# AUTOPHAGY IN ENDOCRINE-METABOLIC DISEASES ASSOCIATED WITH AGING

EDITED BY : Maria Ines Vaccaro, Vincenzo De Tata and Claudio Daniel Gonzalez PUBLISHED IN : Frontiers in Endocrinology

#### Frontiers eBook Copyright Statement

The copyright in the text of individual articles in this eBook is the property of their respective authors or their respective institutions or funders. The copyright in graphics and images within each article may be subject to copyright of other parties. In both cases this is subject to a license granted to Frontiers. The compilation of articles constituting this eBook is the property of Frontiers.

Each article within this eBook, and the eBook itself, are published under the most recent version of the Creative Commons CC-BY licence. The version current at the date of publication of this eBook is CC-BY 4.0. If the CC-BY licence is updated, the licence granted by Frontiers is automatically updated to the new version.

When exercising any right under the CC-BY licence, Frontiers must be attributed as the original publisher of the article or eBook, as applicable.

Authors have the responsibility of ensuring that any graphics or other materials which are the property of others may be included in the CC-BY licence, but this should be checked before relying on the CC-BY licence to reproduce those materials. Any copyright notices relating to those materials must be complied with.

Copyright and source acknowledgement notices may not be removed and must be displayed in any copy, derivative work or partial copy which includes the elements in question.

All copyright, and all rights therein, are protected by national and international copyright laws. The above represents a summary only. For further information please read Frontiers' Conditions for Website Use and Copyright Statement, and the applicable CC-BY licence.

ISSN 1664-8714 ISBN 978-2-88966-079-7 DOI 10.3389/978-2-88966-079-7

#### About Frontiers

Frontiers is more than just an open-access publisher of scholarly articles: it is a pioneering approach to the world of academia, radically improving the way scholarly research is managed. The grand vision of Frontiers is a world where all people have an equal opportunity to seek, share and generate knowledge. Frontiers provides immediate and permanent online open access to all its publications, but this alone is not enough to realize our grand goals.

#### Frontiers Journal Series

The Frontiers Journal Series is a multi-tier and interdisciplinary set of open-access, online journals, promising a paradigm shift from the current review, selection and dissemination processes in academic publishing. All Frontiers journals are driven by researchers for researchers; therefore, they constitute a service to the scholarly community. At the same time, the Frontiers Journal Series operates on a revolutionary invention, the tiered publishing system, initially addressing specific communities of scholars, and gradually climbing up to broader public understanding, thus serving the interests of the lay society, too.

#### Dedication to Quality

Each Frontiers article is a landmark of the highest quality, thanks to genuinely collaborative interactions between authors and review editors, who include some of the world's best academicians. Research must be certified by peers before entering a stream of knowledge that may eventually reach the public - and shape society; therefore, Frontiers only applies the most rigorous and unbiased reviews.

Frontiers revolutionizes research publishing by freely delivering the most outstanding research, evaluated with no bias from both the academic and social point of view. By applying the most advanced information technologies, Frontiers is catapulting scholarly publishing into a new generation.

#### What are Frontiers Research Topics?

Frontiers Research Topics are very popular trademarks of the Frontiers Journals Series: they are collections of at least ten articles, all centered on a particular subject. With their unique mix of varied contributions from Original Research to Review Articles, Frontiers Research Topics unify the most influential researchers, the latest key findings and historical advances in a hot research area! Find out more on how to host your own Frontiers Research Topic or contribute to one as an author by contacting the Frontiers Editorial Office: researchtopics@frontiersin.org

## AUTOPHAGY IN ENDOCRINE-METABOLIC DISEASES ASSOCIATED WITH AGING

Topic Editors: Maria Ines Vaccaro, University of Buenos Aires, Argentina Vincenzo De Tata, University of Pisa, Italy Claudio Daniel Gonzalez, Centro de Educación Médica e Investigaciones Clínicas Norberto Quirno (CEMIC), Argentina

Citation: Vaccaro, M. I., De Tata, V., Gonzalez, C. D., eds. (2020). Autophagy in Endocrine-metabolic Diseases Associated with Aging. Lausanne: Frontiers Media SA. doi: 10.3389/978-2-88966-079-7

# Table of Contents

*05 Editorial: Autophagy in Endocrine-Metabolic Diseases Associated With Aging*

Maria Ines Vaccaro, Vincenzo De Tata and Claudio Daniel Gonzalez

*08 mTORC1 Overactivation as a Key Aging Factor in the Progression to Type 2 Diabetes Mellitus*

Carlos Guillén and Manuel Benito

*15 Pancreatic Cancer Diagnosis and Management: Has the Time Come to Prick the Bubble?*

Pedro Moutinho-Ribeiro, Guilherme Macedo and Sónia A. Melo


Iván E. Alfaro, Amelina Albornoz, Alfredo Molina, José Moreno, Karina Cordero, Alfredo Criollo and Mauricio Budini


M. Bugliani, S. Mossuto, F. Grano, M. Suleiman, L. Marselli, U. Boggi, P. De Simone, D. L. Eizirik, M. Cnop, P. Marchetti and V. De Tata

*73 Palmitic Acid Reduces the Autophagic Flux and Insulin Sensitivity Through the Activation of the Free Fatty Acid Receptor 1 (FFAR1) in the Hypothalamic Neuronal Cell Line N43/5*

María Paz Hernández-Cáceres, Lilian Toledo-Valenzuela, Francisco Díaz-Castro, Yenniffer Ávalos, Paulina Burgos, Carla Narro, Daniel Peña-Oyarzun, Jasson Espinoza-Caicedo, Flavia Cifuentes-Araneda, Fernanda Navarro-Aguad, Cecilia Riquelme, Rodrigo Troncoso, Alfredo Criollo and Eugenia Morselli

*86 Polycystin-2 is Required for Starvation- and Rapamycin-Induced Atrophy in Myotubes*

Catalina Kretschmar, Daniel Peña-Oyarzun, Cecilia Hernando, Nadia Hernández-Moya, Alfredo Molina-Berríos, María Paz Hernández-Cáceres, Sergio Lavandero, Mauricio Budini, Eugenia Morselli, Valentina Parra, Rodrigo Troncoso and Alfredo Criollo

*95 Autophagic and Proteasomal Mediated Removal of Mutant Androgen Receptor in Muscle Models of Spinal and Bulbar Muscular Atrophy* Maria Elena Cicardi, Riccardo Cristofani, Valeria Crippa, Veronica Ferrari, Barbara Tedesco, Elena Casarotto, Marta Chierichetti, Mariarita Galbiati, Margherita Piccolella, Elio Messi, Serena Carra, Maria Pennuto, Paola Rusmini and Angelo Poletti

*109 Secretory Autophagy and its Relevance in Metabolic and Degenerative Disease*

Claudio Daniel Gonzalez, Roxana Resnik and Maria Ines Vaccaro


# Editorial: Autophagy in Endocrine-Metabolic Diseases Associated With Aging

#### Maria Ines Vaccaro<sup>1</sup> \* † , Vincenzo De Tata<sup>2</sup> and Claudio Daniel Gonzalez <sup>3</sup>

*<sup>1</sup> University of Buenos Aires - CONICET, Institute of Biochemistry and Molecular Medicine, Buenos Aires, Argentina, <sup>2</sup> Department of Translational Research and New Technologies in Medicine and Surgery, University of Pisa, Pisa, Italy, <sup>3</sup> CEMIC University Institute, Buenos Aires, Argentina*

Keywords: autophagy, autophagy (macroautophagy), chaperon mediated autophagy (CMA), secretory autophagy, aging, diabetes mellitus, cancer, skeletal muscle atrophy

#### **Editorial on the Research Topic**

#### **Autophagy in Endocrine-Metabolic Diseases Associated With Aging**

Autophagy is a highly regulated self-degradative process of cytoplasmic cellular constituents usually activated under certain conditions such as starvation or other different forms of cell stress that result in breakdown proteins and other cell components to obtain energy. Autophagy is also responsible for removing damaged or aged organelles, eliminating different pathogens and misfolded, aggregated, or altered proteins. Autophagy is an evolutionarily biologically conserved process that sequesters and delivers cytoplasmic components to the lysosome for degradation. It is also involved in the removal of cells that have undergone classical apoptosis. Autophagy is commonly associated with cell survival mechanisms: its dysregulation, however, may be also associated with cell death. In the classical view, according to the pathway that cargo follows to reach the lysosomal compartment, there are three major types of canonical degradative autophagy. These types are: microautophagy/endosomal microautophagy, chaperone mediated autophagy, and macroautophagy, the last one being characterized by the engulfment of cytoplasmic contents by a double membrane vesicle, named autophagosome. However, other non-canonical types of autophagy have been described. One of these unconventional forms is named secretory autophagy, a newly recognized process that is becoming of increasing relevance to explain the non-canonical secretion of a series of cytosolic proteins that have critical biological importance.

Disruption of autophagy is associated with aging and metabolic and degenerative diseases including cancer. This special issue contains a collection of 12 articles covering a broad range of key topics on the interplay of the different types of autophagy alterations with aging, endocrinemetabolic, and degenerative diseases.

Chaperone-mediated autophagy (CMA) represents a major mechanism for degradation of cytosolic proteins. It also plays a significant role in the regulation of lipid and carbohydrate metabolism. Dysregulation of chaperone-mediated autophagy has been found in several models of Parkinson's Disease, Alzheimer's Disease, and Huntington's Disease. Alterations in CMA may also play a role in the pathophysiology of Lateral Amyotrophic Sclerosis and other forms of degenerative disease. Alfaro et al. review in detail the potential involvement of chaperone-mediated autophagy in neurodegeneration as well as in aging, pointing out some critical gaps in knowledge as a rich substrate for further research. CMA consists of the internalization of selected cytosolic substrates into the lysosome by a mechanism that includes recognition of a pentapeptide-like KFERQ in the substrate by the chaperone hsc70, substrate presentation by the chaperone to the receptor LAMP2A, receptor multimerization, and protein internalization for degradation in the lysosome

#### Approved by:

*Antonello Lorenzini, University of Bologna, Italy*

> \*Correspondence: *Maria Ines Vaccaro mvaccaro@ffyb.uba.ar*

#### †ORCID:

*Maria Ines Vaccaro orcid.org/0000-0002-2456-747X*

#### Specialty section:

*This article was submitted to Endocrinology of Aging, a section of the journal Frontiers in Endocrinology*

Received: *29 June 2020* Accepted: *14 July 2020* Published: *20 August 2020*

#### Citation:

*Vaccaro MI, De Tata V and Gonzalez CD (2020) Editorial: Autophagy in Endocrine-Metabolic Diseases Associated With Aging. Front. Endocrinol. 11:572. doi: 10.3389/fendo.2020.00572* assisted by a luminal form of hsc70 (1). LAMP2A is the only known lysosomal receptor for CMA. LAMP2A localization is defective and its function impaired in cystinosis (2, 3). Zhang et al. develop and characterize human cystinotic proximal tubule cells and demonstrate that these cells are characterized by CMA defects that affect vesicular trafficking mechanisms regulating the localization in the plasma membrane of the scavenger receptor megalin.

Secretion of some proteins lacking a "signal peptide" (for instance, some cytokines, insulin-degrading enzyme, alpha-Synuclein, etc.) does not follow the canonical secretion pathway. Those proteins are secreted following different unconventional processes. One of these routes relies on autophagy; it is autophagy dependent. "Secretory autophagy" may then explain the secretion of some relevant peptides involved in several pathophysiological processes (4). Some aggregation-prone proteins, like Amyloid beta or alpha-Synuclein are secreted by secretory autophagy. Some pro-inflammatory mediators such as interleukin-1 beta also follow this non-canonical secretory process. Alteration in secretory autophagy, as extensively described by Gonzalez et al. in this special issue, may play a substantial role in degenerative and metabolic diseases and their treatment.

Skeletal muscle atrophy is a common finding in aging and many degenerative diseases. Different degrees of muscle atrophy can be achieved under diverse physiological conditions, exposure to certain drugs, or starvation. Many of the mechanisms associated with muscle atrophy remain obscure. Kretschmar et al. propose a novel role of polycystin-2, a membrane protein of the transient receptor potential family, as a regulator of skeletal muscle atrophy mediated by the modulation of mTOR in myotubes. Polycystin-2 regulates autophagy in several models, under different stimuli and in diverse tissues. However, Kretschmar et al. suggest that the polycystin-2 role in the regulation of muscle atrophy may be independent of autophagy. This article proposes new mechanisms in the field and opens spaces for further investigation in this area. Spinal and bulbar muscle atrophy is a rare disease associated with a mutation in the androgen receptor. A polyglutamine expansion in the N-terminus of the androgen receptor protein is associated with neurotoxicity. These alterations have a critical role in the pathophysiology of the disease. Cicardi et al. provide a detailed characterization of the autophagy activation and involvement in the disease initiation and progression, reveal some novel mechanisms, and suggest some potential targets for therapeutic exploration.

Hypothalamic arcuate neurons can sense the nutrient status of the organism and accordingly regulate food intake and glucose metabolism. Alterations of this neuronal network can contribute to the pathogenesis of obesity-related diseases such as type 2 diabetes mellitus (5). In their paper, Hernández-Cáceres et al. suggest that the inhibition of autophagy could be a potential mechanism for the high fat diet-induced obesity and diabetes. By using the neuronal cell line N43/5 they demonstrate that palmitic acid exposure causes inhibition of autophagy and decreased insulin sensitivity. Their results indicate that hypothalamic autophagy could represent a promising therapeutic target in diabetic patients. It has been proposed that altered autophagy can promote beta cell dysfunction in diabetic patients (6). To further clarify the relationship between autophagy and beta cell damage, Bugliani et al. explore the effects of autophagy modulation (rapamycin induction or 3-methyladenine inhibition) in human islets under conditions of ER stress. Lipotoxic (palmitate) and chemically induced (brefeldin) ER stress are associated with alteration of beta cell survival and function. Interestingly, the authors reported that, in these conditions, the promotion of autophagy by rapamycin ameliorates the function, survival, and ultrastructure of beta cells. Thus, tuning autophagy could be a useful tool for beta cell protection.

Autophagy is an important homeostatic protective mechanism, but, on the other hand, its alterations can be involved in various pathologic processes (7). The paper by Barbosa et al. reviews recent developments on the role of impaired autophagy in age-related diseases. Aging can be defined as a time–dependent deterioration of cell functioning due to damage accumulation. Autophagic activity has been shown to decrease with age, potentially contributing to the accumulation of damaged molecules and organelles. Thus, to clarify the role of dysfunctional autophagy in establishing the hallmarks of aging could help to define new anti-aging therapeutic strategies and to extend longevity.

The mTORC1 signaling pathway couples energy and nutrient abundance to the execution of cell growth, cell division and metabolism (8). Moreover, the activation of mTORC1 inhibits autophagy. The paper of Guillén and Benito reviews the role of mTORC1 in the progression of diabetes. It has been widely demonstrated that autophagy is crucial to maintain pancreatic beta cell homeostasis (9). The progression of the diabetic disease is associated with a chronic overactivation of mTORC1 and consequently with a sustained inhibition of autophagy in pancreatic beta cells. The failure of such an important protective mechanism could induce the apoptosis of pancreatic beta cells and the impairment of compensatory insulin secretion that characterize the transition to prediabetes to clinically evident type 2 diabetes.

Cancer is emerging as a complex disease where metabolic alterations and inflammatory processes play a critical role. Autophagy and autophagy related proteins are involved in cancer pathophysiology. Arroyo et al. report the cases of two patients with progressing chronic lymphocytic leukemia (CLL). They simultaneously detect autophagy protein LC3B and classical phenotypic markers used for identification of tumoral CLL B cell clones. They found that two patients with progressing CLL showed increased expression of the autophagy protein LC3B, in addition to CD38 and ZAP70 positive expression, suggesting that activation of autophagy may correlate with CLL progression even before ibrutinib treatment. Ropolo et al. present evidence of new regulatory pathways involved in autophagy induction in high resistance pancreatic tumor cells. They demonstrated that gemcitabine requires the expression of a pancreatitis associated and autophagy related protein VMP1 (10) to induce autophagy in pancreatic cancer cells PANC-1 and MIAPaCa-2 that carry activated KRAS. Ropolo et al. identified E2F1, a molecule regulated by the retinoblastoma pathway, as the transcription factor in gemcitabine-induced VMP1-mediated autophagy of highly resistant pancreatic cancer cells. Finally, Moutinho-Ribeiro et al. discuss the potential role of exosomes (11) (small circulating extracellular vesicles of 50– 150 nm in diameter) in several aspects related to pancreatic cancer, from initiation to tumor progression and its applicability in early detection and treatment. Increasing knowledge on cancer exosomes, whose secretory mechanisms are related to autophagy, provides valuable insights on new therapeutic targets and can potentially open new strategies to treat malignant disease.

Scientific knowledge in the field of autophagy grew exponentially along the last decade. A more precise characterization of the different forms of autophagy, their molecular mechanisms and potential implications in the pathophysiology of many diseases reflect the magnitude of such advances. However, several gaps in knowledge and unmet needs remain uncovered and call for further research. Identification of molecular targets for pharmacological modulation of autophagy and its translation into clinical practice remains among the more urgent ones. The role of autophagy modulation in therapy of metabolic and degenerative diseases challenges our current knowledge and opens potentially promising avenues for future investigation.

#### REFERENCES


## AUTHOR CONTRIBUTIONS

All authors listed have made a substantial, direct and intellectual contribution to the work, and approved it for publication.

## FUNDING

This work was supported by grants from: Consejo Nacional de Investigaciones Científicas y Técnicas (CONICET) [PIP GI-11220150100160CO]; Agencia Nacional de Promoción Científica y Tecnológica (ANPCyT) [PICT2016- 1032] and Universidad de Buenos UBACyT 2018-2021; GC-20020170100082BA].

#### ACKNOWLEDGMENTS

The authors would like to thank all the authors who contributed to this Research Topic, as well as the reviewers of the manuscripts for their efforts.


**Conflict of Interest:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2020 Vaccaro, De Tata and Gonzalez. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# mTORC1 Overactivation as a Key Aging Factor in the Progression to Type 2 Diabetes Mellitus

Carlos Guillén1,2 \* and Manuel Benito1,2

<sup>1</sup> Department of Biochemistry and Molecular Biology, Faculty of Pharmacy, Complutense University of Madrid, Madrid, Spain, <sup>2</sup> Spanish Biomedical Research Centre in Diabetes and Associated Metabolic Disorders (CIBERDEM), Instituto de Salud Carlos III, Madrid, Spain

#### Edited by:

Maria Ines Vaccaro, Universidad de Buenos Aires, Argentina

#### Reviewed by:

Claudio Daniel Gonzalez, Centro de Educación Médica e Investigaciones Clínicas Norberto Quirno (CEMIC), Argentina Piero Marchetti, Università degli Studi di Pisa, Italy James Harper, Sam Houston State University, United States

> \*Correspondence: Carlos Guillén cguillen@ucm.es

#### Specialty section:

This article was submitted to Endocrinology of Aging, a section of the journal Frontiers in Endocrinology

Received: 04 July 2018 Accepted: 27 September 2018 Published: 16 October 2018

#### Citation:

Guillén C and Benito M (2018) mTORC1 Overactivation as a Key Aging Factor in the Progression to Type 2 Diabetes Mellitus. Front. Endocrinol. 9:621. doi: 10.3389/fendo.2018.00621 Type 2 Diabetes Mellitus (T2DM), a worldwide epidemics, is a progressive disease initially developing an insulin resistant state, with manifest pancreatic beta islet overwork and hyperinsulinemia. As the disease progresses, pancreatic β cells are overwhelmed and fails in their capacity to compensate insulin resistance. In addition, it is usually associated with other metabolic diseases such as hyperlipidemia, obesity and the metabolic syndrome. During the progression to T2DM there is a chronic activation of mTORC1 signaling pathway, which induces aging and acts as an endogenous inhibitor of autophagy. The complex 1 of mTOR (mTORC1) controls cell proliferation, cell growth as well as metabolism in a variety of cell types through a complex signaling network. Autophagy is involved in the recycling of cellular components for energy generation under nutrient deprivation, and serves as a complementary degradation system to the ubiquitin-proteasome pathway. Autophagy represents a protective mechanism for different cell types, including pancreatic β cells, and potentiates β cell survival across the progression to T2DM. Here, we focus our attention on the chronic overactivation of mTORC1 signaling pathway in β islets from prediabetics patients, making these cells more prone to trigger apoptosis upon several cellular stressors and allowing the progression from prediabetes to type 2 diabetes status.

Keywords: T2DM, autophagy, amylin, mitophagy, mTORC1, TSC2

## mTORC1 SIGNALING NETWORK: UPSTREAM AND DOWNSTREAM EFFECTORS

Tuberous sclerosis complex (TSC) is an essential regulator of mTORC1 signaling pathway. TSC is a heterotrimer protein complex formed by the association of TSC1 (hamartin), TSC2 (tuberin) and Tre2-Bub2-Cdc16-1 domain family member 7 (TDC1D7) (1). This complex is the major inhibitory node of the mechanistic target of rapamycin complex 1 (mTORC1) (2). TSC1-TSC2- TBC1D7 complex exerts its inhibitory effect on mTORC1 by a member of the Ras superfamily, called Rheb. TSC2 presents a GTPase activating domain (GAP) favoring the transformation of GTP-Rheb (ras homology enriched in brain) into GDP-Rheb, avoiding mTORC1 activation (2–5). TSC1 and TSC2 are modulated by multiple mechanisms, including phosphorylation. TSC1/TSC2 integrates signals coming from the energetic status of the cell and nutritional availability, with those from extracellular signaling coming from hormones or growth factors such as insulin or IGF1 respectively (2, 6).

mTOR (mechanistic target of rapamycin) is a serinethreonine kinase, belonging to the phosphatidyl-inositol-3 kinase related kinases (PIKK) (7)**.** Depending on the proteins that are associated with mTOR, it can be found into two different complexes, mTORC1 and mTORC2. mTORC1 is the "rapamycin-sensitive" complex and mTORC2 is the "rapamycin-independent" complex. There are two specific components of mTORC1, the regulatory-associated protein of mTOR (RAPTOR) and the 40-Kda proline-rich Akt substrate (PRAS40) (8–10). However, the rapamycin-independent companion of mTOR, known as RICTOR, the mammalian stress-activated MAP kinase-interacting protein 1 (mSIN1) and the protein observed with RICTOR (PROTOR) are specific members of mTORC2 complex (11). The best characterized substrates of mTORC1 are S6 kinase 1 (S6K1) and eIF4Ebinding protein 1 (4E-BP1), controlling protein synthesis and ribosome biogenesis (12). Activation of mTORC2 leads to phosphorylation and activation Akt at serine 473. mTORC2 it is considered as the PDK2 required for the full activation of Akt (13–15).

In the last years, it has been determined that mTORC1 activation is produced on the lysosomal membrane through a complex mechanism involving different proteins (16). Phosphorylation events coming from growth factors, through the Akt and MAPK kinases, directly phosphorylates TSC2 and dissociates it from the surface of the lysosome, avoiding GAP activity toward Rheb and mTORC1 activation (17). Furthermore, nutrients such as aminoacids are capable to activate mTORC1 signaling in a Rag-dependent manner, another group of small G-proteins, acting as heterodimers for mTORC1 activation (18, 19). The presence of aminoacids affects RagA/B-GTP ratio vs. RagA/B-GDP through GAP activity of GATOR1 (20) and the guanine exchange factor activity (GEF) of the complex called "Ragulator" (21). Then, mTORC1 is regulated by two different systems of small G proteins, named Rag and Rheb, on the surface of the lysosome (16). Our group has demonstrated that TSC2, apart from its phosphorylation state, can be regulated by acetylation in different lysine residues (22). Now, we are analyzing the relevance of acetylation status of TSC2 for its recruitment to the membrane of the lysosome.

Another important effector of mTORC1 signaling is AMPactivated protein kinase (AMPK). AMPK is highly conserved kinase in eukaryotes. AMPK acts as an allosteric sensor of cellular energy status and regulates anabolic and catabolic signaling pathways for the maintenance of energy homeostasis (23, 24). In addition, AMPK inhibits mTORC1 signaling directly by TSC2 phosphorylation, favoring TSC1-TSC2 association (25). Furthermore, AMPK modulates mTORC1, independently from TSC2 by raptor phosphorylation and inactivation of mTORC1 (26)**.** Moreover, AMPK and autophagy are connected directly by phosphorylation of ULK1 that leads to autophagy induction. In fact, several residues of ULK1 protein are directly phosphorylated by AMPK (27, 28). It has recently been proposed that ULK1 acetylation is essential for autophagy activation (29, 30).

## AUTOPHAGY AND ITS RELEVANCE IN PANCREATIC β CELLS

Autophagy is a process that controls cytoplasm quality by the elimination of protein aggregates or altered organelles (31, 32). Autophagy is involved in the shutdown of energyconsuming pathways and the stimulation of catabolic processes through the degradation of cellular components under nutrient deprivation (33). Alternatively, to the ubiquitin-proteasome system (UPS), cells depend on autophagy as an alternative degradation system. Furthermore, autophagy it is involved in the elimination of misfolded proteins or altered organelles. The term autophagy encompasses several conserved mechanisms in eukaryotes that can be classified as: macroautophagy, chaperonemediated autophagy and microautophagy. Chaperone-mediated autophagy consists in the recognition of a specific sequence by a chaperone (hsc70) and then migration of the complex to the lysosome. There, the complex interacts with a receptor in the lysosomal membrane, called LAMP-2A, translocates inside and it is finally degraded (34). In microautophagy, the lysosome directly engulfs the components that are going to be degraded (35). As a general process, macroautophagy it is characterized by the production of a double membrane compartment, termed autophagosome, which surrounds cytoplasmic components and then, fuses with lysosomes where proteolytic activity of the later will degrade the engulfed components (36). Autophagy protects pancreatic β cells, increasing pancreatic β cell survival in the progression to T2DM (37, 38). Using a mouse model Atg-7 deletion specifically in pancreatic β cells, it has been determined the essential implication of autophagy in the survival of pancreatic β cells (37). Furthermore, our group has defined this protection of autophagy in ER-stress using a model of insulin secretion-deficient β cells (39). The protective role of autophagy has been determined for avoiding the toxicity of human amylin (hIAPP) (40, 41). In fact, a mouse model with a chronic overactivation of mTORC1 (β-TSC2-/-) showed an increase in pancreatic β cell death and an impairment in autophagy, as we demonstrated in collaboration with Yoshiaki Kido's lab in Kobe University (42). It has recently been observed in human pancreatic β cells an association between dysregulation in autophagy with an increased in cell death (43, 44). Accordingly, this data point to mTORC1 as a signaling pathway that is overactivated in diabetic patients and, at least partially responsible of pancreatic beta cell failure. All the data exposed in the present paper are derived from animal models and not from pre-diabetic patients. There is a clear effect of mTORC1 hyperactivation and the progression to T2DM, but in animal models. The extrapolation to the possible effect in humans has to be assured.

## mTORC1 AND MITOPHAGY. CONSEQUENCES ON T2DM

mTORC1 it is an endogenous inhibitor of autophagy. Apart from the bulk autophagy, a non-specific mechanism, all the organelles are submitted to a specific degradation. In the case of mitochondria, this process is named mitophagy. Additionally, a mitochondrial unfolded protein response (mtUPR) for the degradation of protein aggregates in different locations inside the mitochondria, such as the matrix and in the inner membrane, occurs (45, 46). The inhibition of mTORC1 slows aging by an increased in autophagy, favoring the elimination of misfolded proteins and impaired organelles such as mitochondria, avoiding its accumulation, and associated with aging and different aging-related diseases such as T2DM, or Parkinson disease, or Alzheimer disease (47). In fact, in the progression to T2DM exists an increasing alteration in mitochondrial morphology and function. Then, an increased in mitophagy could be a potential mechanism for delaying the progression to the disease and maintain a preserved β-cell function (48). Mitophagy is a complex process and there are several possible mechanisms. During the development and maturation of red blood cells, it is necessary a removal of mitochondria in the precursor cell or reticulocyte through a NIP3-like protein X (NIX, also known as BNIP3L) mechanism (49). However, in mammals exist another mechanism with involves the system PTEN-induced putative kinase 1 (PINK1)-Parkin pathway. PINK1 is a serinethreonine kinase recruited to the healthy mitochondria, which is imported into the TIM complex of the inner membrane, where it is cleaved by the mitochondrial processing peptidase (MPP) (50) and presenilin-associated rhomboid-like protein (PARL) (51, 52). PINK1 accumulates specifically in damaged mitochondria, because the processing system is inhibited. Then, the E3-ubiquitin ligase Parkin it is recruited to the outer membrane (OM) of the mitochondria, ubiquitinates different substrates, driving the mitophagic process (53). Our group has determined that in MEF TSC2 KO, which presents a hyperactivation of mTORC1, there is an impairment in the PINK1 recruitment in response to a mitophagic inducer such as carbonyl cyanide m-chlorophenylhydrazone (CCCP). Furthermore, in TSC2–/– cells there is a reduction in Pink1 expression as well as PINK1 protein production. Then, in cells lacking TSC2, an accumulation of damaged and aged mitochondria occurs (54). This situation contributes to the accumulation of reactive species of oxygen (ROS) as occurred in diabetics (48). In addition, amylin accumulates and forms cytotoxic oligomers, which damage the cell membrane and directly affects to the functionality and viability of β cells (55). We have recently published that INS1E overexpressing human amylin (INS1E-hIAPP) presents a hyperactivation of mTORC1 and a diminished level of TSC2. Furthermore, these cells possess an impairment in mitophagic flux with the concomitant accumulation of damaged mitochondria (56). However, the mechanism underlying pancreatic β cell toxicity during the formation of amyloid remains unknown (57). In **Figure 1**, it is depicted the main players in the regulation of pancreatic β cell fate and different drugs acting at different levels.

#### mTORC1, AGING AND ITS EFFECTS ON T2DM

It is well known that insulin signaling is involved in the control of longevity in a wide spectrum of organisms including worms, flies, and mice (58–62). In addition, the use of rapamycin or knocking-down mTOR or S6K1 can promote life extension in several species (63–66). In addition, TSC1 and TSC2 activation, which negatively controls mTORC1, prolongs life span in Drosophila (67). During aging or under a hypercaloric diet exists an mTORC1 hyperactivity, which derives into a disruption in autophagy and, concomitantly an increase in ER stress (16). In this regard, nicotinamide adenine dinucleotide (NAD+) as well as sirtuins levels can be modulated during caloric restriction, activating autophagy, and directly affecting mammalian longevity (68). Mammalian sirtuins are composed of seven members (SIRT1-7) and can be activated by several stimuli including energy deprivation, caloric restriction and resveratrol (69, 70). Sirtuins are a group of NAD+-dependent histone deacetylases with homology to the yeast silent information regulator 2 (Sir2) proteins (71). SIRT1-3 catalyzed deacetylation reaction producing a by-product, nicotinamide and O-acetyl ADP-ribose (OAADPr), along with the deacetylated lysine as it has been reviewed in (72). NAD<sup>+</sup> levels declines in different tissues with aging, altering the functionality of sirtuins and hence plays a key role in multiple diseases, such as metabolic disorders, neurodegenerative diseases and many others (73). Then, NAD<sup>+</sup> plays a central role in aging process and longevity (74).

Sirtuins can be activated by different treatments, such as caloric restriction, rapamycin, resveratrol and many others. In fact, all of these treatments are capable to modulate mTORC1 signaling (75). The overactivation of mTORC1 signaling specifically in pancreatic β cells leads to an augmented in β cell mass, which are related to hyperinsulinemia and hypoglycemia (76–79). However, chronic overactivation of mTORC1 signaling pathway develops a progressive hyperglycemia and a diminished islet mass (76). In this regard, we recently published that overactivation of mTORC1 in beta islets from a β-cell specific deletion of Tsc2 (βTsc2 (–/–) mice) induces an impairment in autophagy with an accumulation of damaged mitochondria. Those deleterious effects were partially reverted by the use of rapamycin (42).

In pancreatic β cells, SIRT1 represses the expression of uncoupling protein 2 (UCP2), leading to an increase in mitochondrial ATP production, thus enhancing insulin secretion (80). Accordingly, a β cell-specific SIRT1-overexpressing transgenic mice (BESTO) has a better glucose tolerance and a boost in insulin secretion (81). Nowadays, SIRT1 it is considered a regulator of different metabolic pathways. There are multiple SIRT1-regulated targets, modulating transcriptional activity of different transcription factors, such as peroxisome proliferatoractivated receptor gamma (PPARγ), PPARα, PPAR gamma coactivator 1 alpha (PGC-1α), and the forkhead box, subgroup O (FOXO) family. Then, SIRT1 is capable to regulate different processes like insulin secretion, gluconeogenesis and fatty acid oxidation (82).

#### TYPE 2 DIABETES MELLITUS AND mTORC1

Type 2 diabetes mellitus (T2DM) is a very complicated disorder and it is considered epidemic in the world

(83). T2DM it is a progressive disease including insulin resistance, β-cell hyperplasia and/or β cell hypertrophy, that mediates a compensatory insulin secretion and subsequently hyperinsulinemia, pancreatic β cell dysfunction and a loss of cell identity or de-differentiation (84). However, the underlying mechanism mediating pancreatic β cell fate in type 2 diabetics is not completely known. T2DM is associated with higher production of glucose and lipids facilitating β cell death. Furthermore, advanced glycation end-products (AGEs) have been proposed as pancreatic β cell death inducers (85). The hyperamylinemia found in obese people and in insulin-resistant patients, contributes to its oligomerization inside pancreatic β cells, being deleterious for pancreatic β cells (86). From the whole amount of autopsy from patients with T2DM, around 80% present amyloid deposits in the pancreas. However, the importance of amylin it is not yet completely understood. Thus, at the insulin resistant prediabetic stage, mTORC1 is a key effector for the growth and survival of pancreatic β cells. However, if mTORC1 remains chronically overactivated, pancreatic beta cell death occurs and the compensatory insulin secretion mechanism it is compromised. Then, mTORC1 is a double-edged sword in the progression to T2DM (87). The location where protein synthesis occurs is in the endoplasmic reticulum (ER). When misfolded proteins are accumulated into the ER, the unfolded protein response (UPR) is activated. UPR is a protective mechanism that alleviates the cell from that overload. However, if UPR activation it is maintained for a long period of time, pancreatic β cell death occurs (88). T2DM alters the capacity of ER to manage the increased demand of protein synthesis of pancreatic β cells (89). Increasing the expression of endogenous chaperones such as Bip (90), the use of chemical chaperones, such as taurine-conjugated ursodeoxycholic acid (TUDCA) or 4-Phenyl butyric acid (4-PBA) (91), or increasing the ER protein folding capacity by the use of azoramide (92), diminish β cell failure and facilitate the proper folding and avoiding protein aggregation and accumulation of damaged organelles.

However, mTORC1 hyperactivation is observed in other tissues, such as in the heart muscle contributing to the dysfunction and complication of T2DM (93). Furthermore, not only in cardiac tissue, in the liver, apart from being involved in the regulation of lipid homeostasis, facilitating lipogenesis and inhibiting lipolysis and lipophagy (94), is upregulated in insulin resistant states, such as T2DM contributing to the dysregulation of glucose as well as lipid homeostasis (95). In adipose tissues (96), mTORC1 is also involved in the generation of insulin resistance as it has been reviewed in (16).

## CONCLUDING REMARKS

Diabetes is a multifactorial and progressive disease with two phases; firstly, a prediabetic stage, with an insulin resistance and hyperinsulinemia, and secondly as manifest diabetes associated with hypoinsulinemia and hyperglycemia. Then, it is crucial to understand the transition from prediabetes to type 2-diabetes status and the underlying molecular mechanisms of disease. At this stage, chronic overactivation of mTORC1 signaling pathway in β islets from prediabetics patients leads to on one hand to the expansion of the pancreatic beta cell mass and, on the other to the inhibition of autophagy as protective mechanism of beta cells against the attack of several stressors, making these cells more prone to trigger apoptosis. Thus, the maintenance of a functional autophagy it is an essential component to protect and prolong pancreatic β cell life span precluding chronic hyperglycemia.

## REFERENCES


## AUTHOR CONTRIBUTIONS

All authors listed have made a substantial, direct and intellectual contribution to the work, and approved it for publication.

## FUNDING

Our research it is supported by grant SAF2017-82133R from M.C.T., Spain.

## ACKNOWLEDGMENTS

We thank all the current and former members of the lab, which directly or indirectly, has contributed to this mini-review. We apologize to all not cite authors, whose work was excluded owing to the space limitation.


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Guillén and Benito. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Pancreatic Cancer Diagnosis and Management: Has the Time Come to Prick the Bubble?

Pedro Moutinho-Ribeiro1,2, Guilherme Macedo1,2 \* and Sónia A. Melo2,3,4 \*

<sup>1</sup> Department of Gastroenterology, Centro Hospitalar São João, Porto, Portugal, <sup>2</sup> Faculty of Medicine of the University of Porto, Porto, Portugal, <sup>3</sup> Institute for Research Innovation in Health (i3S), Porto, Portugal, <sup>4</sup> Institute of Pathology and Molecular Immunology of the University of Porto, Porto, Portugal

Pancreatic cancer (PC) is associated with poor prognosis and very dismal survival rates. The most effective possibility of cure is tumor resection, which is only possible in about 15% of patients diagnosed at early stages of disease progression. Recent whole-genome sequencing studies pointed genetic alterations in 12 core signaling pathways in PC. These observations hint at the possibility that the initial mutation in PC might appear nearly 20 years before any symptoms occur, suggesting that a large window of opportunity may exist for early detection. Biomarkers with the potential to identify pre-neoplastic disease or very early stages of cancer are of great promise to improve patient survival. The concept of liquid biopsy refers to a minimally invasive sampling and analysis of liquid biomarkers that can be isolated from body fluids, primarily blood, urine and saliva. A myriad of circulating molecules may be useful as tumor markers, including cell-free DNA (cfDNA), cell-free RNA (cfRNA), circulating tumor cells (CTC), circulating tumor proteins, and extracellular vesicles, more specifically exosomes. In this review, we discuss with more detail the potential role of exosomes in several aspects related to PC, from initiation to tumor progression and its applicability in early detection and treatment. Exosomes are small circulating extracellular vesicles of 50–150 nm in diameter released from the plasma membrane by almost all cells and exhibit some advantages over other biomarkers. Exosomes are central players of intercellular communication and they have been implicated in a series of biological process, including tumorigenesis, migration and metastasis. Several exosomal microRNAs and proteins have been observed to distinguish PC from benign pancreatic diseases and healthy controls. Besides their possible role in diagnosis, understanding exosomes functions in cancer has clarified the importance of microenvironment in PC progression as well as its influence in proliferation, metastasis and resistance to chemotherapy. Increasing knowledge on cancer exosomes provides valuable insights on new therapeutic targets and can potentially open new strategies to treat this disease. Continuous research is needed to ascertain the reliability of using exosomes and their content as potential biomarkers, so that, hopefully, in the near future, they will provide the opportunity for early diagnosis, treatment intervention and increase survival of PC patients.

Keywords: pancreatic cancer, early diagnosis, liquid biopsy, biomarkers, exosomes

#### Edited by:

Maria Ines Vaccaro, Universidad de Buenos Aires, Argentina

#### Reviewed by:

Alfonso Pompella, University of Pisa, Italy Xinran Ma, East China Normal University, China

#### \*Correspondence:

Guilherme Macedo guilhermemacedo59@gmail.com Sónia A. Melo smelo@ipatimup.pt

#### Specialty section:

This article was submitted to Endocrinology of Aging, a section of the journal Frontiers in Endocrinology

Received: 19 September 2018 Accepted: 11 December 2018 Published: 08 January 2019

#### Citation:

Moutinho-Ribeiro P, Macedo G and Melo SA (2019) Pancreatic Cancer Diagnosis and Management: Has the Time Come to Prick the Bubble? Front. Endocrinol. 9:779. doi: 10.3389/fendo.2018.00779

## INTRODUCTION

Pancreatic ductal adenocarcinoma (PDAC), the most frequent type of pancreatic cancer (PC), is associated with poor prognosis, with 53,670 new cases and 43,090 estimated deaths in 2017 (1). It represents around 3% of the new cancer cases each year, but it is the fourth most common cause of cancer mortality (2). It is expected to become the second cause of death by cancer by 2020 in the USA (2). On the contrary to the death rates for many other cancer, such as lung, colorectal, breast, and prostate, the death rate for PDAC patients has increased (3) as well as its incidence, that has raised in about 30% (2).

Pancreatic cancer is typically asymptomatic in its first stages of development, and as a consequence of the late diagnosis, this disease presents a very low survival rate. Combining all stages, the overall 5-year survival rate is 5% (4).Tumor resection is the only possibility of cure, but recurrence often happens and, therefore, the 5-year survival rate in resected patients is only up to 25% (1). Its unique tumor biology contributes to early recurrence, metastasis, and a subpar response to conventional therapies (5).

Since lifetime risk of PC in general population is low (1.3%), a population-based screening is not recommended (6). Meanwhile, individuals with family history or genetic predisposition have been identified as risk groups. According to the International Cancer of the Pancreas Screening Consortium (CAPS) it has been proposed that these individuals at higher risk for PC should be considered for screening (6, 7). The referred Consortium defined eligible individuals to be those with more than 5% of lifetime risk or with a 5-fold increased relative risk for PC (6, 7). Nonetheless, this group of patients represents only 10% of the total spectrum of PC, the other 90% of cases being sporadic (8).

#### PANCREATIC CANCER PRECURSOR LESIONS: A WINDOW FOR OPPORTUNITY

Recent studies based on whole-genome sequencing indicated 12 important signaling pathways altered in PC. It was also suggested that the initial mutation occurred nearly 20 years before the first symptoms. This evidence has offered a time frame for pancreatic carcinogenesis, suggesting that a large window of opportunity may exist for early detection, which could improve the prognosis of this lethal disease (9, 10).

In recent years, a myriad of biomarkers have been investigated with specificity for PC detection (8). The ideal biomarker should be able to detect the disease in its early stages, when patients are still amenable for a curative treatment, or even in a more favorable scenario, its premalignant precursor lesions. Pancreatic cancer precursor lesions are intraductal papillary mucinous neoplasms (IPMNs), the mucinous cystic neoplasms (MCNs), and the pancreatic intraepithelial neoplasias (PanINs), the first two being macroscopic cystic alterations and the last one representing microscopic non-invasive epithelial proliferations within the pancreatic ducts (5). Intraductal papillary mucinous neoplasms (IPMNs) are papillary proliferations inside the pancreatic ducts that usually secrete thick mucus, leading to its focal dilation. They represent up to 10% of the neoplasms of the pancreas, and the ones harboring high-grade dysplasia carry an important risk of malignant transformation (11, 12). Depending on the extent of ductal involvement, three subtypes can be recognized: the main-duct intraductal papillary mucinous neoplasm (MD-IPMN), the branch duct intraductal papillary mucinous neoplasm (BD-IPMN), and the mixed type. Malignant transformation is more frequent in main duct and mixed types IPMNs, while BD-IPMNs are considered indolent lesions (11, 12).

In relation to cytoarchitectural and immunophenotypic features, four histological subtypes of IPMNs are considered: the gastric type (49–63%), the intestinal type (18–36%), the pancreaticobiliary type (7–18%), and the oncocytic (1– 8%) type (11, 13). Main-duct intraductal papillary mucinous neoplasm (MD-IPMNs) are more frequently of the intestinal type, a combination that carries the highest risk of invasive transformation, usually giving rise to a colloid type carcinoma. On the other hand, the branch duct intraductal papillary mucinous neoplasm (BD-IPMN) are predominantly of the gastric type, which are characterized by an insignificant risk of malignant transformation. The pancreatobiliary type usually harbors high-grade dysplasia and is considered the aggressive evolution of the gastric type. Finally, the rarer oncocytic type presents with cytological atypia, frequently with high-grade dysplasia (14).

MCNs, unlike IPMNs, are neoplasms that develop in the pancreatic parenchyma, without involvement of the ductal system. Characteristically they are seen in women, rarely occurring in men, with a preferential location in the body and tail of the gland (15).

An ovarian-type stroma is an essential feature for the diagnosis of MCNs that clearly separates them from the much more frequent IPMNs. Depending on the grade of cytologic atypia in epithelial lining (low, intermediate or high-grade dysplasia), which has been associated with point mutations in KRAS and p53 genes, mucinous cystic neoplasms (MCNs) may exhibit different risks of malignant potential (15–17). Interestingly, the observation that in invasive MCNs, the inactivation of SMAD4/DPC4 suppressor gene complex occurs only in the epithelial lining but not at the stroma level, suggests that the typical ovarian-type stroma of these lesions is not involved in the process of malignant transformation (18). The reported incidence of invasive carcinoma among MCNs is variable according to different series, ranging from 6 to 36% (15, 19, 20).

With the development and widespread of imaging modalities, the diagnosis of both of these macroscopic cystic PC precursor lesions is increasing. As the majority of them are asymptomatic and discovered incidentally, its proper management is not consensual and is a matter of debate by many International Societies worldwide. In fact, despite the availability of a significant number of guidelines and recommendations (21), there is still a lack of consensus in the decision of which one(s) to follow.

The vast majority of carcinomas originate from microscopic non-invasive epithelial proliferations within the pancreatic ducts, described as PanIN (22). These lesions are considered precursors in the stepwise progression from intraepithelial to invasive neoplasia. This morphological progression is accompanied by accumulation of genetic changes, in which activating KRAS mutations are thought to be the driving force (8, 12). PanIN lesions are characteristically asymptomatic and are composed of columnar to cuboidal cells with varying amounts of mucin and different degrees of cytological and architectural atypia (23). These lesions are classified into 3 grades: PanIN-1A (flat) and PanIN-1B (papillary) are low-grade lesions with minimal cytological and architectural atypia, while high-grade PanINs (PanIN-3), also described as "carcinoma in situ," are characterized by severe cytological and architectural atypia (23). In cadavers of patients over the age of 80 the prevalence of PanIN lesions is about 55%. Also, these lesions are very frequent in patients with concomitant PC and in familial PC kindreds. The major problem concerning these very frequent lesions is their identification and the evaluation of its malignant potential. Currently, there is no imaging technique capable of an accurate diagnosis of a PanIN lesion and many promising biomarkers are being investigated for this purpose.

#### CONVENTIONAL DIAGNOSTIC TOOLS ARE INSUFFICIENT FOR EARLY DETECTION

Early diagnosis of PC is very challenging with the currently available methods (**Figure 1**) (24). Unlike colonoscopies for colorectal cancer and serum prostate-specific antigen (PSA) levels for prostate cancer, there is currently no standardized PC screening strategy, even for high-risk populations. Pancreatic cancer diagnosis and staging depends, substantially, on imaging modalities including ultrasonography (US) and endoscopic ultrasonography (EUS), multidetector computed tomography (MDCT), magnetic resonance imaging (MRI) and magnetic resonance cholangiopancreatography (MRCP), endoscopic retrograde cholangiopancreatography (ERCP), and positron emission tomography (PET) (25–30).

Nowadays, multidetector computed tomography (MDCT) is the imaging technique of choice for pancreatic diseases, especially in the setting of solid tumors, where it has high accuracy to detect and to stage pancreatic malignancies (31). However, its sensitivity may be suboptimal as it misses some target lesions in the PC screening context. Even when considering thin-section, triple-phase helical CT, the sensitivity to detect lesions smaller than 2 cm is only up to 80% (32). Moreover, ionizing radiation exposition is also an important drawback, precluding CT to be an ideal screening and/or surveillance imaging technique.

In high-risk individuals, magnetic resonance imaging (MRI) can be used as a non-invasive screening imaging test considering the possibility to scan the entire abdomen and pelvis, avoiding radiation exposure. Considering the MRCP, this technique offers, in a non-invasive way (in contrast to ERCP), the capacity to characterize the ductal anatomy of the pancreas and diagnose small cystic lesions such as IPMNs. Preliminary data from CAPS3 study, that included high-risk patients submitted to surgical resection, suggests that MRI/MRCP may be superior to CT particularly for detection of IPMNs (71 vs. 14%, p < 0.001) (33). Some MRI features have been recently added to this technique improving its diagnostic capacity. In fact, diffusion-weighted imaging (DWI), a technique based on the Brownian motion of water molecules in tissue (34), has brought functional aspects into conventional anatomic evaluation, allowing higher contrast resolution and the identification of very small PC lesions.

In diagnosing PC, the advantage of EUS over MDCT has been reported for more than 10 years (35, 36). Despite being an imaging technique operator-dependent, with inherent risks of invasiveness and sedation, EUS has progressively being considered as the most accurate tool to investigate pancreatic diseases. It is also a non-radiation technique that can offer highresolution images and can accurately characterize solid and cystic lesions. Moreover, it can evaluate cystic wall features associated with increased risk of malignancy, namely mural nodules, and other focal thickenings. The data analysis from a screening program involving high-risk individuals that were submitted to surgical resection confirmed the superiority of EUS, as it was able to detect almost twice as many neoplastic lesions comparing to MRI/MRCP or CT (33). Also, past studies (37, 38) that used EUS as a screening tool showed its accuracy in detecting asymptomatic precancerous branch duct IPMNs, large PanINs, incidental pancreatic endocrine tumors and ductal adenocarcinomas. The chance to add some recent techniques such as EUS-guided elastography and contrast-enhanced imaging has expanded new and promising fields of investigation (37–40).

Considering EUS-guided fine-needle aspiration (EUS-FNA), when performed in suspected lesions, it shows diagnostic accuracy for malignancy of more than 85–90%. Apart from its role in the study of solid lesions, the investigation of pancreatic cystic lesions by EUS-FNA can also be very useful, allowing a cytological diagnosis of IPMN and MCN in up to 70% of the cases (41). "Cell-block preparation" and "core tissue sampling" are two developments of this technique (42), which might be useful not only in providing more material for histological evaluation, but also for recently developed ancillary diagnostic techniques, namely: microRNA profiling, KRAS mutation detection and chemo sensitivity testing (43–45).

Lately, confocal laser endomicroscopy has surged as a technological improvement to EUS with a particular interest in cystic lesions characterization. In this technique, a dedicated miniprobe is introduced through a 19-gauge needle previously inserted into the cystic lesion, allowing a real time direct visualization of the epithelial lining at a microscopic level, permitting the identification of suspicious architectural changes (46).

Nowadays, no biomarker exists with adequate sensitivity and specificity for routine clinical diagnosis or screening of PC (8). Carbohydrate antigen (CA) 19-9 and carcinoembryonic antigen (CEA) are the most commonly used blood-based tumor biomarkers in clinical practice. The only biomarker approved by the US FDA for monitoring the progression and therapeutic response of PC is CA 19-9, which is also the only one recommended by the NCCN guidelines for clinical management of PC patients (47). Nevertheless, it is not specific and can be found in normal pancreatic and biliary duct cells, as well as in gastric, endometrial, colonic, and salivary epithelia. In

fact, depending on the series, its specificity ranges from 30 to 100% (48–50), so the extremely high rate of false positive results prohibits its routine use for diagnosis. Moreover, CA 19- 9 sensitivity is also imperfect, ranging from 41 to 86% (48–50). Additionally, up to 15% of general population do not express Lewis antigen and, consequently, CA 19-9 levels cannot be measured (50–53). Another clinically relevant issue is the fact that only 65% of patients with resectable tumors have increased levels of CA 19-9. In relation to its ability to differentiate PC from chronic pancreatitis, this biomarker is also inadequate, as up to 40% of patients harboring this last condition exhibit CA 19-9 levels above the normal range (52, 54). Considering all these pitfalls, serum CA 19-9 is used primarily as a prognostic tool in monitoring patients for recurrence or managing those with late stage disease (52, 55).

Taking the aforementioned aspects in account, it is understandable that most patients present with advanced disease, with only up to 25% having resectable tumors (56–58). We should also have in mind that even if these 25% of patients are identified and properly treated under the current standards of care, their survival, at the best, is up to 24 months (4). So, the focus should be, on one side, in aiming to detect the "real" curable lesions, that is the premalignant and the very small malignant ones, and, on the other side, the development of better therapeutic options. For the early detection scenario, the conventional diagnostic tools are far from being competent. In fact, available non-biopsy tests (serum CA19-9) and imaging techniques lack the sensitivity and specificity necessary to detect tumors smaller than 1 cm in the context of a 1.5% lifetime risk disease (59–65).

In consequence, it is crucial to develop new and improved strategies which can overtake all the obstacles described and diagnose primary tumors that can be resectable at very early stage. The development of markers with high sensitivities and specificities for PC and for its precursor conditions should be a priority issue, with extreme importance mainly in the context of high-risk individuals (66).

### LIQUID BIOPSY, A HORIZON FOR EARLY DETECTION AND SURVIVAL IMPROVEMENT

As mentioned before, given the typical late stage of disease at the time of presentation, when treatment is disappointing, PC remains one of the most dismal diseases worldwide, with incidence nearly parallel to mortality. There has been much effort invested in identifying accurate tumor markers, ideally present in the timeframe between cancer onset and invasion, to allow diagnosing PC in early curable stages, to ultimately improve patient's survival (67, 68). In this setting, the ideal biomarker should be easily detected with satisfactory sensitivity and specificity and should distinguish PC from other benign pancreatic lesions. In the context of early detection, the identification of preneoplastic conditions, such as PanINs, IPMNs and MCNs, is of great importance (69).

The concept of liquid biopsy refers to the analysis of biomarkers present in a sample of a body fluid collected through a simple and painless, minimally invasive technique. The body fluids mostly used for biomarker isolation are essentially blood, urine and saliva (70–72). A myriad of circulating molecules may be used as tumor markers, including cell-free DNA (cfDNA), cell-free RNA (cfRNA), circulating tumor cells (CTC), circulating tumor proteins, and extracellular vesicles, more specifically exosomes (66, 73). Blood is easily accessible and relatively stable, making serum an ideal specimen to explore potential biomarkers. However, biomarkers secreted into serum are extremely diluted and probably obscured by other more-abundant serum proteins (74). Technological advances in the last decade have provided more opportunities to discover circulating biomarkers based on "omics" analyses, including methods focused on proteins, nucleic acids, CTCs, and exosomes. Numerous proteins of low abundance can be analyzed by mass spectrometry-based approaches and proteomic technologies (24).

In recent years, based on the expression of transcriptional profiles and structural variations, different molecular subtypes of PC have been described thought genomic analyses (75–78). The early detection of mutant genes that identify PC and its subtypes is essential for an effective strategy for the management of the disease.

In PDAC, there are four major driver genes (one oncogene and three tumor suppressor genes) implicated in tumorigenesis (5). KRAS is the most frequently altered oncogene that encodes a GTPase which mediates downstream signaling from growth factor receptors; somatic mutations, clustered in specific hotspots (most in codon 12), are identified in more than 90% of PC (79). CDKN2A, by turn, is the most frequently altered tumor suppressor gene, with loss of function in more than 90% of tumors; it encodes an important cell-cycle regulator (79). TP53 is another tumor suppressor gene, with essential role in cellular response to stress, also exhibits frequent somatic mutations (79). Lastly, SMAD4, a tumor suppressor gene mediating downstream signaling of the transforming growth factor β (TGFβ) receptor is inactivated in about half of the PC cases (79).

Since Mandel and Metais, in 1948, (80) first described circulating free DNA in body fluids, an exponential interest in non-invasive technology for disease monitoring has been the focus of research in many centers worldwide.

Recently, due to the possibility to pair genomic tests with tests on CTCs, circulating tumor nucleic acids (ctNAs) and tumorderived exosomes, liquid biopsies have gained increased value for clinical application (71, 81, 82).

During PC initiation and progression, many different genetic modifications take place, including genetic diversification, amplifications and homozygous deletions, an increase in duplicate chromosomal number, recapitulation of clonal expansion, clonal selection, driver mutations and losses of heterozygosity (10, 83–90).

Next-generation sequencing techniques provide deeper insight into somatic mutations and epigenetics analysis of the genome and broaden the characterization of circulating tumor DNA (ctDNA) and (cfRNA). With the development of cell tracking techniques and flow cytometry, it is now possible to capture and analyze CTCs and exosomes (24).

In this manner, the capabilities of liquid biopsy are enormous, allowing the characterization of tumor biomarkers in the same way tissue biopsy does, favoring improvement of the knowledge of tumor heterogeneity and, most importantly, contributing to early detection, monitoring of disease progression and response to treatment (**Figure 2**) (71, 82, 91–96).

#### EXOSOMES: THE BUBBLES OF THE FUTURE?

Exosomes are small cup-shaped extracellular vesicles (50–150 nm in size) released from the plasma membrane by almost all cells, including cancer cells (**Figure 1**) (24). They play an important role in intercellular communication, tumorigenesis and cancer metastasis (97–100). Structurally, they are enveloped by a lipid bilayer membrane with tissue-specific content instead of cellular organelles, such as pathogenic mRNA, microRNA, DNA fragments, and proteins (101, 102). After release, exosomes are stable in the extracellular environment or enter the circulation and can be taken up by neighbor or distant cells (103, 104). Exosomes allow the exchange of material and information between cells, altering gene expression or mediating RNA silencing (105).

Considering their extensive distribution and functions, exosomes are ideal candidates to find circulating biomarkers for PC detection and management (67, 104, 106). Moreover, exosomes have some advantages over other biomarkers (106, 107). First, they are widely distributed in nearly all body fluids, including serum, and are relatively stable when stored for long term at −80◦C. Second, cancer cells secrete more exosomes than normal healthy cells. One of the reasons for this can be the acidic conditions of the tumor microenvironment that enhances the release of exosomes (108–110). Third, exosomes contain a combination of proteins, DNA, coding and non-coding RNAs, and lipids that can be used as a natural panel of biomarkers for simultaneous evaluation. Fourth, PC-derived exosomes enter the circulation at an early stage of cancer development and are related to metastasis. In fact, on the contrary of some other cancer markers that are released into the blood only after necrosisrelated cell death occurs, a phenomenon usually associated with advanced stage disease with large volume tumor mass, cell-secreted exosomal nucleic acids (DNA fragments, mRNA, miRNAs, and others) are released in circulation during the initial phases of the tumorigenesis process (66, 91, 107). This aspect is critical for the early detection of PC because PC cells are able to metastasize at an early stage, with great impact on prognosis.

Recent studies have implicated PC-derived exosomes in the early development of PC (103) and showed that they contribute for establishing a premetastatic niche in the liver (98, 111) and that they can promote tumor formation and proliferation (112). In relation to exosomes' content, a main focus of research has been on RNA and microRNA profiling, in part due to the established utility of a variety of some of these molecules in cancer screening and also due to the relative ease feasibility of its amplification (99, 113–118). By turn, in order to determine their cellular origin, exosomal proteomic profiling is being an important focus of research in recent years (114).

Besides their possible role in diagnosis, the study of exosomal function has contributed to the improvement of the comprehension of the microenvironment related to PC and progression of the disease. The ultimate aim of understanding the way exosomes can influence tumor initiation, proliferation and metastasis is improving the knowledge on PC pathophysiology and patient prognosis. Moreover, its role in the development of some paraneoplastic conditions, such as diabetes mellitus and cachexia, and in the resistance to chemotherapy, can provide insights for future therapeutic targets (119–125).

The extraction of exosomes from body fluids has been described using various methods and technologies. Commonly

used methods for isolation are ultracentrifugation, precipitation, size, immunoaffinity, and microfluidics (126).

Strong efforts have focused on developing sensitive diagnostics tools improving early detection of PC via identifying pancreatic cancer-associated exosomal markers (67, 104, 127). The first case–control study on PC exosomes was conducted by Que et al. (128), where four exosomal microRNAs were evaluated as candidates. In this study, a moderate discrimination of cases from controls was seen with miR-21 and miR-17-5p. In 2015, in a study involving patients with PC, chronic pancreatitis and controls, a higher expression of exosomal miR-10b was shown only in cancer patients (129). Although this cohort was a small one, the strength of this study was the use of new technology of label-free nanoplasmonic-based short non-coding RNA sensing. In the same year, a more extended cohort including PC patients, patients with chronic pancreatitis, individuals with benign pancreatic neoplasms and controls was evaluated by Madhavan et al. (130). In this well conducted study, a combination of four miRNAs (miR-1246, miR-4644, miR-3976, and miR-4306) and five proteins (CD44v6, Tspan8, EpCAM, MET and CD104) in circulating exosomes was able to distinguish PC patients from non-cases ones, with a sensitivity of 1.0 (CI: 0.95-1) and a specificity of 0.80 (CI: 0.67-0.90).

In the past years, the number of exosomal miRNAs studied in this context was considerable. Examples of these molecules are miR-21, miR-17-5p, miR-155, miR-34, miR- 196a, miR-181a, miR-181b, miR-138-5p, miR-494, miR- 542-3p, miR-31, and miR-205, that have been implicated for several studies to have the capacity, when upregulated, to promote cellular proliferation and angiogenesis, metastasis and disease progression and even chemo-resistance in PC patients (113, 128, 131–138).

These studies emphasize the importance of exosomal miRNAs not only as diagnostic and prognostic biomarkers, but also open a field of investigation of its use as potential targets for treatment PC.

Recently, Melo et al. showed that glypican-1 (GPC1), a membrane anchored protein, in circulating exosomes may distinguish with 100% specificity patients with PC or precancerous pancreatic lesions from patients with benign pancreatic diseases (139). Melo et al. reported that GPC1 expression patterns in exosomes secreted by PC could be utilized to identify subjects with PC early and offer considerable insights into the disease progress and tumor load. A comparison of exosomes from PC and control cell lines indicated that the exosomes from cancer exhibited enhanced levels of GPC1. In serum specimens from subjects with PC (n = 190), a significantly larger amount of GPC1+ circulating exosomes was present compared with normal controls (n = 100). Interestingly, direct analysis of GPC1 in serum itself revealed lower sensitivity and specificity than measurement of GPC1 in purified serum exosomes. Furthermore, GPC1+ exosomes were also confirmed to contain identical KRAS mutations, which frequently are present in PC and precancerous lesions and have been considered a fundamental mutation (139). Moreover, higher levels of GPC1 positive circulating exosomes were seen in both PC and PC precursor lesions, such as IPMNs, when compared to other benign diseases of the pancreas and healthy controls, with a perfect area under the receiver-operating characteristic curve (AUC) of 1.0. The authors concluded that given the high sensitivity and specificity of exosomal GPC1 in differentiating PC, independently of its stage, from non-PC controls, this biomarker could have a promising role

in PC early detection. Not surprisingly, when compared to the tumor marker CA 19-9 (AUC: ranging between 0.69 and 0.79), exosomal GPC1 was found to be significantly superior (p < 0.001) (139). Additionally, using a genetically engineered mouse that progressively developed into PC, the identification of GPC1+ exosomes exhibited positive results prior to pancreatic lesions being detectable by MRI. Tumor burden was associated positively with levels of GPC1+ circulating exosomes. In most subjects, the exosome levels reduced following the removal of the solid tumor. In this manner, circulating exosomal GPC1 may be seen as a prognostic indicator too, reflecting tumor load and monitoring disease progression and patients' survival. Moreover, evolving research will address if this biomarker can play a role in the treatment of PC as a potential pharmacological target (139). Many others since than have reported findings related to GPC1 detection in exosomes, not only in PC but also breast and colorectal cancer cases (140–148). In any event, in the future, further independent confirmation of exosomal GPC1 performance is needed, assessing its role as a more reliable marker in predicting diagnosis and prognosis of PC when compared to CA19-9, but also, and as it seems to be overexpressed in some precursor lesions such as IPMNs, its potential role in individualizing the management of such conditions (149).

Also interestingly, a paper from Hoshino et al. (98) has described that integrin αvβ5-expressing PC exosomes can determine liver metastasis and that targeting that molecule reduced exosome absorption by resident cells and inhibited liver involvement. The levels of integrin αvβ5 were noticeably enhanced in exosomes isolated from PC individuals with liver metastasis compared with those with no distant metastasis. Together, these findings have raised the possibility that GPC1+ and integrin αvβ5-expressing circulating exosomes may be used as indicators of PC progression and liver metastization.

Taking in account the excellent biodistribution and biocompatibility of exosomes, the idea of its utilization as vehicles for drug, genes or nucleic acids delivery has gained increased acceptance for continuous research in the field of PC treatment (124, 150, 151), especially when considering the fact that they can be particularly targeted to specific cell types by engineered exosomes-producer cells (124, 150, 152). This is important especially when dealing with a tumor with such difficulties in therapy delivery due to the intrinsic resistance of its microenvironment and its dense stroma (153).

#### CONCLUSION AND NEW PATHWAYS OF INVESTIGATION

Given the timeframe from pancreatic tumor initiation to invasion and metastatic capacity, there is a large window of opportunity for early management of this lethal disease. An ideal diagnostic method for PC should definitively distinguish malignant lesions from benign ones and detect early-stage disease and preneoplastic conditions, such as PanINs and cystic mucinous lesions with risk of progression.

There are many challenges in the early detection of PC, including its asymptomatic nature, the lack of imaging exams able to detect minimal lesions and the absence so far of sensitive and specific molecules in body fluids. Recently described circulating biomarkers associated with PC initiation and progression, easily detectable in blood, followed by confirmative diagnosis based on imaging and pathologic results might be the future ideal strategy for screening and diagnosing PC.

A number of circulating biomarkers have been widely studied, but validation for routine clinical use is still needed. The lack of sufficient samples from non-invasive precursor lesions and early-stage PC must be addressed, and animal models are important tools for research. In fact, extensive understanding of the fundamentals of PC development and the nature of precursor lesions is crucial prerequisite toward discovering and applying novel biomarkers.

Exosomes have been proving to be reliable candidates as PC biomarkers, as its contents, namely DNA, RNASs, proteins, lipids, and metabolites are largely derived from tumor cells. In this way, tracking these molecules will conduct to the knowledge of cell-type of origin, and also importantly the specificity for pre-metastatic niches formation and distant colonization.

Its potential utilization as biomarkers, besides the ultimate goal of early detection of PC, can also play an important role in monitoring disease progression.

Moreover, considering its excellent biodistribution, biocompatibility and cell-specific nature, exosomes can be used, in the future, as drug delivery vehicles.

The majority of studies addressing exosomes as biomarkers have been based on patients with established diagnosis of PC. However, and having in mind that when facing this dismal disease, the earlier the diagnosis the better the chance for cure, it would be of great interest to access the performance of these biomarkers in the context of high-risk individuals screening.

Continuous investigation is needed to ascertain the potential of these biomarkers, so that, hopefully, in the near future, they will provide the opportunity for early diagnosis, treatment intervention and increasing survival of PC patients.

## AUTHOR CONTRIBUTIONS

PM-R did the bibliographic search and wrote the initial manuscript. GM and SM critically revised the manuscript and actively contributed to it.

## ACKNOWLEDGMENTS

The Melo team is supported by the project NORTE-01-0145- FEDER-000029, Norte Portugal Regional Programme (NORTE 2020), under the PORTUGAL 2020 Partnership Agreement, through the European Regional Development Fund (ERDF) and national funds through FCT—Foundation for Science and Technology IF/00543/2013/CP1184/CT0004, PTDC/BIM-ONC/2754/2014, POCI-01-0145-FEDER-32189 and Astrazeneca Foundation, FAZ Ciencia Award.

## REFERENCES


from malignant (or premalignant) neoplasms? Ann Surg. (2000) 231:205–12. doi: 10.1097/00000658-200002000-00009


neoplasm of the pancreas is highly predictive of pancreatic neoplasia. Diagn Cytopathol. (2006) 34:457–62. doi: 10.1002/dc.20446


**Conflict of Interest Statement:** SM has ownership interests (patents).

The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Moutinho-Ribeiro, Macedo and Melo. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Hallmarks of Aging: An Autophagic Perspective

#### María Carolina Barbosa<sup>1</sup> , Rubén Adrián Grosso<sup>1</sup> and Claudio Marcelo Fader 1,2 \*

<sup>1</sup> Laboratorio de Biología Celular y Molecular, Instituto de Histología y Embriología (IHEM), Universidad Nacional de Cuyo, CONICET, Mendoza, Argentina, <sup>2</sup> Facultad de Odontología, Universidad Nacional de Cuyo, Mendoza, Argentina

Autophagy is a major protein turnover pathway by which cellular components are delivered into the lysosomes for degradation and recycling. This intracellular process is able to maintain cellular homeostasis under stress conditions, and its dysregulation could lead to the development of physiological alterations. The autophagic activity has been found to decrease with age, likely contributing to the accumulation of damaged macromolecules and organelles during aging. Interestingly, failure of the autophagic process has been reported to worsen aging-associated diseases, such as neurodegeneration or cancer, among others. Likewise, it has been proposed in different organisms that maintenance of a proper autophagic activity contributes to extending longevity. In this review, we discuss recent papers showing the impact of autophagy on cell activity and age-associated diseases, highlighting the relevance of this process to the hallmarks of aging. Thus, understanding how autophagy plays an important role in aging opens new avenues for the discovery of biochemical and pharmacological targets and the development of novel anti-aging therapeutic approaches.

#### Edited by:

Maria Ines Vaccaro, Universidad de Buenos Aires, Argentina

#### Reviewed by:

Rong Yuan, Southern Illinois University Carbondale, United States Claudio Acuña-Castillo, Universidad de Santiago de Chile, Chile

#### \*Correspondence:

Claudio Marcelo Fader cfader@fcm.uncu.edu.ar

#### Specialty section:

This article was submitted to Endocrinology of Aging, a section of the journal Frontiers in Endocrinology

Received: 30 July 2018 Accepted: 17 December 2018 Published: 09 January 2019

#### Citation:

Barbosa MC, Grosso RA and Fader CM (2019) Hallmarks of Aging: An Autophagic Perspective. Front. Endocrinol. 9:790. doi: 10.3389/fendo.2018.00790 Keywords: autophagy, ROS, aging, hallmarks of aging, mitophagy

#### THE AUTOPHAGIC PROCESS

Autophagy, literally meaning "self-eating," is an evolutionarily conserved catabolic process in eukaryotic cells by means of which intracellular components and extracellular incorporated material are delivered into lysosomes, where their degradation occurs (1). Since its discovery, autophagy has been associated with the maintenance of cellular homeostasis, as well as the cytoplasmic quality control process (1, 2). Its dysregulation has being related to a diversity of pathological or physiological processes such as neurodegenerative, infectious, and metabolic disorders, as well as cancer and aging, among others (3–5). Several studies have demonstrated that autophagy can be very selective in targeting its cargo for degradation. Three major types of autophagy have been identified: macroautophagy, microautophagy, and chaperone-mediated autophagy (CMA). Macroautophagy (hereafter referred to as autophagy) begins with the extension of a specialized membrane, known as the phagophore, derived from the endoplasmic reticulum (ER), the mitochondria, and the Golgi cisternae (6, 7). The phagophore engulfs the molecules and organelles to be eliminated, forming a double membrane vesicle called autophagosome (7, 8). Finally, autophagosomes are targeted to lysosomes and fusion occurs, the sequestered material is degraded and released back into the cytosol (8). In microautophagy, the lysosome picks up cytosolic components directly via invagination of the lysosomal membrane (9). On the other hand, CMA is a process involving the direct transport of cytosolic components across the lysosomal membrane via chaperone proteins. Several studies have demonstrated that CMA is a highly regulated and degradative event, involving HSC70 (heat shock protein 70 complex) and multimerization of

**26**

the LAMP2A receptor (lysosome-associated membrane protein type 2A). Interestingly, not all proteins are able to be CMA substrates. To undergo CMA degradation, proteins must contain a KFERQ motif in their amino acid sequences, which is necessary to bind the chaperone HSC70 (10, 11). Substrate and the HSC70 complex can bind a 12-amino-acid cytosolic tail of LAMP2A for lysosomal docking. In addition, LAMP2A multimerization is necessary for substrate translocation into the lysosomal lumen. Cytosolic HSC70 is released from the multimeric complex, and then a chaperone HSP90 (located at the lumen of the lysosomal membrane) interacts with LAMP2A, stabilizing it during the substrate translocation. Finally, a luminal chaperone HSC70 is required to end the translocation process, and once inside the targeted protein is degraded by the lysosomal enzymes (12) (**Figure 1**).

Autophagy can be induced by a variety of stressors, and nutrient restriction is one of the major stimuli, capable of rapidly activating the autophagic process with the concomitant inhibition of protein synthesis (1). Studies in both yeasts and mammals have characterized at least 40 autophagy-related genes (Atg), which encode proteins that participate in autophagy (13). In addition, the canonical autophagy pathway includes the inactivation of mammalian target of rapamycin complex 1 (mTORC1), allowing the phosphorylation and activation of the Unc-51-like kinase complex (Ulk1/2), with the subsequent cascade activation of the other ULK complex members such as FIP200 and ATG13 (14, 15). Another important complex that is activated is the BECLIN1, in which VPS34, one of its members, is translocated into the ER membranes and it produces high levels of phosphatidylinositol-3-phosphate, which is necessary for the recruitment of other effectors such as WIPI2b (16). Next, this effector interacts and recruits ATG16L, which binds ATG5-ATG12 conjugated to generate the ATG12-ATG5- ATG16L complex. This complex is required for the lipidation of LC3 (Microtubule-associated protein light chain 3), because it determines the site where LC3 will be conjugated and activated to LC3-II (17). Moreover, ATG3 (an E2-like protein) is associated with LC3-I and it binds to the complex through ATG12, allowing the conjugation of LC3-I with phosphatidylethanolamine to generate LC3-II. LC3-II, which is present in both inner and outer membranes of autophagosomal structures and is necessary for phagophore extension, cargo engulfment, and vesicle closure to form the autophagosome. Additionally, the targeted cargo is able to bind receptor/adaptor molecules like p62, NDP52, and NIX. These proteins contain a LC3 interacting region (LIR), which allows the recognition of elements to be engulfed by the phagophore and eliminated in an autophagic manner (18, 19).

In addition to degradation, autophagy, or part of its machinery, can mediate a regulated cell death, named autophagydependent cell death (ADCD). Moreover, autophagy can participate in other cell death types [reviewed in reference (20)]. Interestingly, despite the fact that regulated cell death of malignant cells is a pro-survival mechanism at the whole organism level, it can also lead to tissue degeneration and function loss, and this can reduce the fitness of the aged individual (21).

## AGING

Aging, the natural event occurring in all living organisms, can be defined as a deterioration of the cell functioning due to damage accumulation over time (22–25). This is an important biological, demographic and socio-economic issue all over the world. Dr. Barja points out that all living organisms have different longevity, indicating that evolution has played an important role in regulation and flexibilization of aging between species, in a relatively fast process (26). The understanding of the molecular basis of aging and longevity could let us manipulate it somehow in the future. In this regard, in the last 50 years numerous investigations related to aging have emerged, trying to explain this unstoppable process.

Despite the general accepted concept that aging is a multifactorial process, several theories have emerged in an attempt to explain it as a single predominant age-related change. A popular aging theory is the "Stochastic Theory," which suggests that aging results from random damage accumulation. This can be due to external and internal sources over time, in addition to a failure of the repairing capacity. On the other hand, other theories support the idea that aging is a regulated process, mainly by the genetic code, such as the telomere length, the number of divisions that a somatic cell can go through (the "Hayflick limit") and spatio-temporal regulation of gene expression (27, 28). Nevertheless, one of the most popular theories is the Free Radicals (or Oxidative Stress) Theory of Aging, which hypothesizes that an accumulation of Reactive Oxygen Species (hereafter ROS) falls into an oxidative damage of biomolecules, with the consequent cell functioning decline (27–29). A considerable body of evidence supports this theory, because it points to an increase in ROS cellular levels as we age, due to a higher production of them as well as a failure in the anti-oxidant systems (30, 31).

## AUTOPHAGY IN AGING

Several animal models have contributed to our understanding of how the impairment of autophagy and redox homeostasis can result in age-related diseases. In the same way, numerous studies involving genetic ablation or induction of autophagic genes have revealed the importance of this process in aging of yeast, nematodes, flies, and mammals (32). The most important work that links an overexpression of a single Atg gene with an increment in mammals' lifespan was conducted by Pyo and collaborators. The authors overexpressed Atg5 in mice and found an enhancement of the autophagy process and antiaging features, compared with the wild type mice. The mean lifespan was also incremented, suggesting the importance of autophagy in the longevity of mice (33). Another approach that demonstrates the importance of autophagy in aging has been done in Ana María Cuervo's laboratory. In aged mice, they generated a double transgenic mouse model, in which it was possible to modulate the expression of the lysosomal receptor for CMA. The results revealed that the enhancement of this receptor can prevent features of aging at cellular and organ levels (34). In addition, mice overexpressing Atg5 showed a better resistance to

age-related obesity and enhanced insulin sensitivity, exhibiting an improved metabolism in aged individuals (33). Despite the mentioned studies, several others failed to demonstrate that upregulation of a single autophagic component can extend lifespan (32). Moreover, several KO mouse models have been shown to have extended lifespan, although the molecular mechanisms behind it and the connection with aging are not yet clear (35).

Microautophagy: invagination of the lysosomal membrane engulfs cytosolic cargo in small vesicles for its degradation inside.

Notably, another relationship between autophagy augmentation and extended lifespan has been reported in exceptionally healthy centenarian humans, who have increased levels of BECLIN1, compared to young people (36). We hope that in the next years these preliminary studies in humans will be more advanced, providing insights into our species longevity mechanisms from clinical case studies.

The knockout for essential Atg genes is lethal in mice, and tissue-specific ablation has a less-dramatic phenotype, manifesting premature signs of aging (37). Specific-Atg5 or Atg7 KO leads to neurodegeneration or tissue abnormalities in most of the cases available in the literature [for a more detailed summary see reference (37)].

Finally, as we age, the incidence of cancer rises, probably because of the decline of homeostatic processes and the increase in the accumulation of potentially harmful molecules such as ROS and protein aggregates. Autophagy has been proposed to have a dual role in tumorigenesis, being important both in suppression as well as in tumor progression and surveillance (38, 39).

## ROS GENERATION AND AGING

The ROS are considered metabolites of molecular oxygen during cellular respiration, being very reactive due to an unpaired electron (40). Mitochondria are the major ROS producers and perhaps the organelle most affected by them. In order to avoid detrimental effects of ROS, two important processes arise: Mitophagy and antioxidant system. Mitochondrial ROS can activate the autophagic pathway upon starvation by the activation of ATG4 (41), and this in turn leads to autophagic degradation of mitochondria (mitophagy) in order to reduce the ROS levels by limiting the number of mitochondria per cell (42). In addition, hypoxia and exercise can also trigger redox-dependent autophagy, suggesting that ROS might regulate the autophagic response to several stresses (43). Regarding the second process, the antioxidant system consists of several enzymes and molecules that react with ROS and neutralize them somehow, but the connection of antioxidants with lifespan is controversial (44). Notably, overexpression of a mitochondrial-targeted catalase in mice extends lifespan and reduces overall ROS, reinforcing the Free Radicals Theory of Aging in such model (45). Moreover, these mice showed a reduction in age-related pathologies (46). Additionally, Mn-superoxide dismutase (SOD2) heterozygous mice showed a life-long reduction, but surprisingly they did not have an accelerated aging phenotype. Nevertheless, this Sod2+/<sup>−</sup> mice showed a higher oxidative damage to DNA and had higher cancer incidence compared with wild type individuals (47). By contrast, knockout of 17 genes involved in the antioxidant system exhibited no effect in lifespan: Only the knockout for Cu/Znsuperoxide dismutase (Sod1) resulted in a decrease in longevity and premature aging as well (48). This mouse model showed an increase in senescent markers, suggesting that the oxidative stress that Sod1−/<sup>−</sup> mice suffer leads to DNA damage, promoting an irreversible state of quiescence (49). In addition, these Sod1−/<sup>−</sup> mice showed an accelerated sarcopenia, manifesting muscular mass loss and altered neuromuscular junctions (50). Despite these controversial and unexpected results in mice, the relationship between antioxidants and their role in healthy or pathologic aging needs to be deeply studied in the future.

Finally, it is important to highlight that ROS have been proposed to be implicated in proliferation and survival signaling in certain conditions (42). A new concept has emerged recently in the aging field, termed "hormesis," according to which low doses of a stressor can improve the cell response for a more detrimental condition in the future (32, 51). This could increase lifespan and cellular fitness (52). In this context, low levels of ROS can be beneficial due to the trigger of homeostatic responses, but its disproportional augmentation can lead to damage or aging (42, 53). From an autophagic perspective, an augmentation in ROS levels and a decline in mitophagy occur simultaneously, leading to aging (43, 52, 54).

## HALLMARKS OF AGING: AN AUTOPHAGIC VIEW

In the last years, aging has begun to be seen as an active and highly regulated process (55). Age-related changes at cellular level include an increase in ROS, loss of proteostasis, genome instability, and telomere exhaustion, among others (23, 56, 57). These characteristic features of aging were termed "hallmarks of aging" by López-Otín et al. (52). In the following sections, we discuss how autophagy plays an important role in some of these hallmarks of aging, in both health and disease.

### Loss of Proteostasis

Proteostasis is one of the major functions of autophagy in normal tissues. Imbalance of proteostasis due to aging leads to protein aggregation, accumulation of misfolded proteins and in the end to cellular dysfunction, among others (23, 56, 57). Notably, carbonylation due to oxidative stress is one of the changes that leads to loss of proteostasis (44). To avoid cell death or dysfunction, numerous homeostatic mechanisms turn on, mainly autophagy (58) and the Ubiquitin-Proteasome-System (UPS). Because autophagy is considered one of the most important intracellular homeostatic processes, an alteration or deterioration of this pathway could modify the normal cell functioning, including a variety of diseases and normal cell physiology declination. Autophagosomes and lysosomes decline in an agedependent manner in muscles (59), heart (43), and several other tissues. Moreover, CMA has also been implicated in removing oxidized and potentially dangerous proteins by direct lysosomal degradation (60).

The UPS is mostly implicated in the degradation of misfolded proteins, as well as short- and long-lived proteins by their ubiquitination. This process is achieved thanks to three major proteins that sequentially activate the ubiquitin tag (E1), transfer it to a second enzyme (E2), and finally ligate the ubiquitin tag to the target molecule (by E3 ligase), which eventually reaches the proteasome for degradation (61, 62). It is important to note that almost all regulatory proteins are substrates for this system (61, 63), and UPS declines with age [reviewed in (64)]. Interestingly, mTORC1 was found to regulate not only lysosomal protein degradation, but also proteasomal proteolysis of longlived proteins, independently of protein synthesis, suggesting a common regulation of both proteolytic systems by nutrientsensing (63). In addition, overexpression of a sole subunit of the proteasome enhanced its activity and the survival against several oxidants in two cell lines as well as primary culture of human fibroblasts (65). Moreover, proteasome activity decreased in an age-dependent manner (66). Overexpression of proteasome subunits in aged dermal human fibroblasts ameliorated the aged phenotype and restored the oxidized and ubiquitinated proteins to young levels (66). In the same way, transgenic mice with reduced proteasomal activity accumulated oxidized and ubiquitinated proteins, accelerating the aging phenotype and the age-related metabolic diseases (67). Besides, inhibition of proteasome activity impaired cell proliferation and shortened lifespan (68), reinforcing the importance of a correct proteostasis in healthy aging and longevity.

Several studies have been done on neurodegenerative diseases related to aging and autophagy, including those most relevant for their high impact on human population. Most of them share the accumulation of ROS, misfolded proteins, and damaged organelles, aging being the main risk factor (69, 70). This accumulation interferes with proper axonal traffic, enhancing neurotoxicity. Both autophagy and CMA impairment hamper the correct protein-aggregates degradation and the remodeling of dendrites and axons, thus diminishing the nervous plasticity (71, 72). In Parkinson's disease (PD), the cytoplasmic aggregates are formed by α-synuclein and ubiquitin (or Lewy bodies) in dopaminergic neurons of substantia nigra, leading to their death (32). Alterations in UPS (71) and also in CMA can develop the disease too. It is worth noting that overexpression of Lamp2A improved CMA performance and decreased αsynuclein cytoplasmic levels (73). In the same way, Alzheimer's disease (AD) is characterized by intracellular accumulation of tau protein as well as β-amyloids (Aβ), derived from the amyloid precursor protein (APP). This aggregate formation impairs normal cell function, finally leading to cell death (32). Also, extracellular Aβ-plaques secreted by autophagosomes can interrupt intercellular communication (72), another hallmark of aging (52). Besides, Apolipoprotein E4 (ApoE4) is the main genetic risk for sporadic AD and was found to promote the disruption of the lysosomal membrane together with Aβ, leading to neuronal degeneration (72, 74). Mutations in Presenilin1 or 2 (PS1 and PS2, two transmembrane subunits of gamma-secretase), as well as in tau protein or in APP are common causes of the familial AD (72). Other neurodegenerative diseases implicate alterations in the autophagic process, i.e., SENDA, Huntington, Amyotrophic Lateral Sclerosis, and Frontotemporal Dementia disorders [reviewed in (70, 75, 76)]. In all these cases, the lack of proper degradation by autophagy promotes the aggregation of several proteins and the consequent malfunctioning of axonal transport.

Regarding metabolic diseases and autophagy, it is well known that islet amyloid deposition leads to type 2 diabetes in humans due to the amyloidogenic property of human islet amyloid polypeptide (hIAPP). It is important to note that mice do not develop such aggregation. In order to bypass this model difference, Kim et al. developed transgenic mice expressing hIAPP specifically in β-cells and bred them with Atg71β−cell mice. Male mice had premature diabetes, while females had hyperglycemia but never developed the disease, suggesting a synergism between autophagy deficiency and human amyloid overexpression. Moreover, primary culture of monkey islet cells overexpressing precursors of hIAPP showed that autophagic inhibition by 3-methyladenine (3-MA) increased pro-hIAPP dimer or trimer accumulation, blocking the autophagic activity in these pancreatic cells (77). It is also important to highlight that diabetes or glucose handling deficiencies are risk factors for the AD, as the amyloids properties of proteins implicated in metabolic diseases and AD are similar and probably interconnected (78). More studies regarding the connection between metabolic and neurodegenerative diseases are required for a better understanding of the molecular basis of such relationships at systemic level.

Finally, sarcopenia is characterized by a progressive loss of muscle mass and strength thanks to an imbalance between production and degradation of proteins (79). Aged-related declination of autophagy (both mitophagy as well as CMA) promotes sarcopenia by protein accumulation interference with normal myofibers functioning, but an exacerbation of autophagy can also result in cellular stress and finally death (79). Thus, an age-related imbalance of proteostasis could drive a variety of diseases involving both protein accumulation and degradation.

#### Mitochondrial Dysfunction

Mitophagy is a basal process involved in the autophagic degradation of mitochondria (76, 80, 81). It is necessary in normal differentiation of certain cell types such as red blood cells (82), in embryogenesis, immune response, cell programming, and cell death (80). Mitophagy is required not only to remove damaged mitochondria, but also to promote the biosynthesis of new ones, supporting the mitochondrial quality control (76, 80). Given that mitochondria are implicated in bioenergetics and ROS production, the mitophagy plays an important role in cell homeostasis. Additionally, a decrease in mitophagy is observed in aged animals and this contributes to aging phenotype (81).

Canonically, mitophagy is triggered by the cytosolic exposition of mitochondrial outer membrane (MOM) proteins, which have a LIR domain. The mitophagy is tightly regulated by several molecules, NIX and BNIP being two of the most widely characterized mitochondrial adaptors for autophagic machinery (83). NIX activation is associated with an increment in mitochondrial degradation in HeLa cells, protecting them against cellular stress (84). Interestingly, NIX has a LIR domain which binds LC3 once it is activated by phosphorylation (84). Additionally, PINK1 and PARKIN have been involved in the regulation of mitophagy when the mitochondrion loses its membrane potential (80). These proteins have been considered as key components in controlling the activation of mitophagy (85) and also as participants of mitophagy-associated cancer resistance. PINK1 and PARKIN are activated in response to an increment of intracellular ROS levels, which stimulate the MAPK and ERK1/2 signaling cascades, triggering parapoptosis in non-malignant cells, which bypass the caspases activation and, thus, the apoptosis (86) (**Figure 2**).

Additionally, Mitofusin 2 (MFN2) is a mitochondrial membrane fusing protein involved in several processes, including mitochondria fusion and mitophagy. Its expression declines with age, and its deficiency provokes precocious sarcopenia, accumulation of damaged mitochondria, and metabolic disorders in young mice (87). In addition, Humanin, an antiapoptotic mitochondrial protein, is capable of activating the CMA machinery, thus protecting several cell types from oxidative stress (88). Interestingly, both CMA and Humanin decline with age (89, 90), contributing to the age-related deterioration of proteostasis and mitochondrial functionality. As can be seen, several proteins regulate the mitophagy and contribute to the mitochondria homeostasis. As almost all of them decline with age, the modulation of mitophagy regulatory proteins could be a novel anti-aging therapeutic approach in the future. Despite this, more studies are needed in order to understand the complex regulation of mitophagy and the relationships between the players.

Different compounds, intracellular changes or stimuli could drive an activation of mitochondrial dysfunction. Normally, ROS oxidative stress, loss of membrane potential, MOM permeability, and aging are able to cause mitochondrial dysregulation. This

imbalance triggers signaling pathways involving activation of pro-apoptotic proteins of the BCL-2 family such as BAK and BAX, resulting in an MOM damage and the consequent release of cytochrome c and SMAC (second mitochondrial-derived activator of caspase) to the cytoplasm (91), activating the intrinsic apoptotic pathway through caspase 9 (92, 93) (**Figure 2**). Furthermore, hypoxia inducible factor 1α (HIF1α) is able to trigger mitophagy by stabilization and activation of NIX protein, and is also responsible for autophagic activation through VMP1 (vacuole membrane protein 1) promoter, causing colon cancer resistance to photodynamic therapy (94).

Disruption in mitophagy, and thus in redox homeostasis, can produce different cardiovascular pathologies (43). PINK1-KO mice developed a decline in cardiac function due to dysfunctional mitochondria and an increase in oxidative stress (95). Notably, the same features were observed in human end-stage heart failure samples, where diminished PINK1 levels were also found (95). Despite the ubiquitous KO model, the authors were capable of demonstrating the importance of PINK1 in heart functioning. Additionally, altered mitophagy due to elevated ROS production has been linked to Alzheimer Disease (AD), but there is controversy as to whether this disturbance in the autophagic pathway is a cause or a consequence of AD (72). Remarkably, PARKIN and PINK1 have been found mutated in Parkinson Disease (PD) patients, suggesting the importance of this pathway in dysfunctional mitochondria clearance by autophagy (76, 96). In 2015, Sun et al. published a new approach for measuring mitophagy in vivo, using a transgenic mouse model consisting in the mitochondria-targeted overexpression of a fluorescence reporter named Keima (81). We think that this tool could be very interesting for the in vivo study of mitophagy and its regulation under a wide variety of conditions. Deeper studies are then required to fully understand this process and its role in healthy and pathologic aging.

#### Deregulated Nutrient Sensing

Cellular response to nutrient privation implies some kind of intracellular sensor, capable of triggering the corresponding survival mechanisms. It has been proved that nutrient sensing is a highly conserved process across eukaryotes (97). Several nutrient-related signaling pathways converge on mTOR [mammalian Target Of Rapamycin; (97)], which triggers the response to growth factors, energy, glucose or amino acid changes (38, 97–100). Interestingly, nutrient sensors can also be activated under oxidative stress conditions, suggesting a common regulatory mechanism linking redox homeostasis and nutrients availability [reviewed in (101)].

The kinase mTOR is capable of linking environmental conditions with reproduction and somatic maintenance, thus influencing the individual lifespan (102). In addition, mTOR and mslt8 (a positive regulator of mTORC1) haploinsufficient female mice showed an increment in lifespan, compared to wild type mice (103). Mice carrying hypomorphic alleles of mTOR also had an increase in lifespan, a reduction in aging biomarkers, and a normal metabolism (104), showing the importance of nutrient sensing in aging. Moreover, Ribosomal S6 kinase 1 (S6K1) knockout mice showed an extended lifespan, compared with wild type counterparts, demonstrating again the importance of the mTOR pathway inhibition in longevity (105).

Besides, mTOR forms two distinct complexes, the Complex 1 (mTORC1) being capable of integrating different responses depending on nutrients availability (99, 106, 107). Upstream mTORC1, there are multiple regulating complexes, responding to each amino acid, glucose or growth factor input (99, 106, 107). Notably, p62 is one of the mTOR interactors upon amino acid stimulation, and it has been proposed to specifically stabilize the activated mTORC1 at the lysosome surface (108). It is well known that p62 is an autophagic adaptor protein whose role in nutrient sensing pathway might be another connection between mTOR and the regulation of autophagy.

Because autophagy is a catabolic mechanism, it can be assumed to be implicated in cellular and systemic metabolism. Metabolic stress responses could be compromised due to a decline in autophagic activity (109). As an important process regulating the general cellular status, autophagy can also crosslink metabolic pathways to maintain the homeostasis under a variety of conditions (43). In this sense, it has been demonstrated that, after nutrient or growth factor deprivation, ULK1 and ULK2 are activated, and these kinases phosphorylate and activate several glycolytic enzymes as well as autophagic proteins. This makes it possible to obtain metabolites thanks to glucose uptake, gluconeogenic pathway blockage, and autophagic degradation of cytosolic components (110). Supporting this, mTOR hyperactivation was found in several diseases such as obesity, metabolic syndrome, and type 2 diabetes (100), which highlights the importance of a tight regulation of autophagy as well as the nutrient sensing pathway.

Given that mTOR is capable of sensing the nutritional state of the cell, it was proposed to play an important role in Caloric Restriction (CR) therapy. Indeed, mTOR signaling network was shown to mediate lifespan extension by CR. Notably, the sole amino acid-restriction is enough to promote CR-response (111). Sirtuins 1 and 3 (NAD+ deacetylases) are activated in response to CR, as well as SOD1, in order to change the metabolism accordingly and prevent oxidative damage, respectively. Such ROS regulation rules out the Free Radicals Theory of Aging, at least partially (101, 112, 113). It is important to highlight the fact that inhibition of mTOR (specifically mTORC1) in embryogenesis is lethal, while the ablation of these pathways in adulthood can extend life [reviewed in (111)]. In fact, weight loss is suggested as prevention as well as therapy for a variety of age-related diseases (98).

#### Genomic Instability

In the last decade, several studies have demonstrated that autophagy or autophagic-related molecules act as a "safeguard" of genome stability both directly (DNA repair modulation) and indirectly (by acting as a homeostatic response) (114). Several mouse models have provided substantial information regarding genomic instability and its connection with healthy and pathological aging (55).

Regarding oxidative stress and DNA damage, ROS increase is thought to be mainly harmful for the mitochondrial DNA (mtDNA), generating the mutagenic 8-hydroxy-20 deoxyguanosine (8-OHdG), as well as mutations and deletions in mtDNA that result in a dysfunctional mitochondrion (69). Moreover, mitochondrial dysfunction promotes telomere attrition, telomere loss, and chromosome alterations, culminating in apoptosis in mouse embryos (115). Besides, Donati and collaborators demonstrated that, upon autophagic stimulation with an anti-lipolytic agent in 16-month-old rats, 8- OHdG accumulation in liver was successfully supressed, reaching the values obtained from young animals in only 6 h. When they measured the cytochrome c oxidase activity, they found that this decrease was not associated with lower mitochondrial enzyme activity, demonstrating the selective mitophagy of a small population of 8-OHdG-positive mitochondria and the importance of this proteostatic process in anti-aging mechanisms (54, 116). In the same way, dietary restriction reduced 8-OHdG levels in mitochondrial DNA (mtDNA) of aged rats and mice compared with those fed ad libitum (117), supporting the importance of dietary restriction in prevention of mtDNA damage by ROS in aged animals. Notably, Sod2−/<sup>−</sup> mice accumulated high levels of 8-OHdG both in nuclear and mitochondrial DNA, compared with wild type mice. Nevertheless, they showed no changes in lifespan or age biomarkers (47). On the other hand, Atg7−/<sup>−</sup> mouse keratinocytes presented premature aging after oxidative stress induction, supporting the importance of autophagy in healthy aging (118). In addition, Bender et al. found high levels of mtDNA deletions in dopaminergic neurons of PD patients, compared to controls (119). As we have already mentioned, PARKIN and PINK1 are mutated in PD, thus altered mitophagy can explain, in part, the accumulation of mtDNA damage in PD patients.

Autophagy has emerged as an important process in genome maintenance. After treatment with several cell cycle blockers, human osteosarcoma cells (U2OS) increased the micronuclei frequency as well as autophagosomes. Importantly, the authors observed a small but significant colocalization between them. Knockdown of Atg5 or Atg7 abolished this colocalization. P62/SQSTM1 also colocalized with micronuclei, indicating that micronuclei can be degraded by autophagy and this may contribute to genome stability (120). Moreover, NDP52 and p62-dependent autophagy can degrade retrotransposon RNA, preventing new insertions into the genome of long and short interspersed elements (121). Additionally, autophagy deficiency leads to an accumulation of RHOA with p62. This phenotype drives cytokinesis failure, aneuploidy, and multinucleation due to inappropriate formation of contractile ring (122). Furthermore, allelic loss of Beclin1 promotes tumorigenesis and activation of DNA-damage response in neoplasic cells. In this context, autophagy deficiency leads to genome instability under metabolic stress in these mouse mammary epithelial cells (123). Artificially aneuploid mouse cells showed increased autophagy to protect cells from genome instability (124).

Autophagic adaptor p62 has been found to be implicated in genome instability in several studies. Accumulation of p62 led to the activation of DNA-damage response (125). By contrast, overexpression of p62 (or autophagy deficiency) suppressed DNA-damage response by its direct inhibitory interaction with RNF168, an important E3 ligase for histone H2A ubiquitination and DNA-damage response (126). In this regard, p62 downregulates the protein levels of several molecules involved in homologous recombination (HR) of damaged DNA, inducing at the same time non-homologous end-joining (NHEJ), stressing the importance of p62 nuclear accumulation upon several stresses (127). More studies are needed to completely elucidate the role of p62 and other autophagic components in genome maintenance throughout life.

#### Epigenetic Alterations

Epigenetic changes due to external or internal factors drive several processes, including development and aging (128). In muscular and hematopoietic stem cells, different histone modifications help to establish the quiescence state with age (129). Besides, epigenetic alterations lead to different responses in aging and longevity in several mouse models [reviewed in (55)]. Notably, macrophages of old mice had hypermethylated LC3 and Atg5 promoter regions, compared with those of young mice, downregulating the protein levels and promoting aging decline of autophagy (130).

The natural polyphenol resveratrol, an autophagic inducer, has been proved to exert its action by inositol 1,4,5-triphosphate receptor (IP3R) and by protein acetylation decrease (131, 132). In this sense, a balance between different degrees of protein acetylation is presumed to be important for autophagy (32). High levels of acetyl-CoA, which serves as an acetyl group donor, were found to promote histone hyperacetylation, downregulating the expression of Atg genes, thus inhibiting autophagy and shortening lifespan in aged yeast (133). In addition, these authors found that the knockout of a certain acetyl-CoA in Drosophila melanogaster brain enhances autophagy and prolongs lifespan. Morselli et al. used specific siRNA knockdown of Sirtuin1, a NAD<sup>+</sup> deacetylase in HCT116 cell line, which resulted in a suppression of the autophagic response to resveratrol. Importantly, resveratrol induced autophagy by AMP-dependent kinase/mTOR-independent pathway and changed the acetylation of 375 proteins, half of them involved in autophagy. On the other hand, deacetylations were often observed in metabolism-related proteins, thus activating autophagy (131). Another autophagic inducer, spermidine, is a well-known pro-longevity polyamine (56), largely studied as an anti-aging agent. Like resveratrol, spermidine can induce both acetylation and deacetylation changes that promote the autophagic pathways via AMPdependent kinase signaling. In this regard, low doses of this polyamine together with low doses of resveratrol synergistically caused the same autophagic response as higher doses of each inducer separately (131).

Taken together, organismal models as well as in vitro studies highlight the importance of epigenetics throughout life. The relationship between epigenetic changes and autophagy needs to be deeply studied in order to understand the regulatory loop that seems to be involved in development and aging.

#### Telomere Attrition

Telomeres are specific repetitive sequences at the end of chromosomes, the telomerase is the special polymerase involved in the elongation of this protective zone. It is important to note that somatic cells of mammalian species lack the expression of telomerase. Thus, telomerase is highly regulated in a spatiotemporal manner (134). In each cell cycle, the telomere becomes shorter and shorter, leading to a vulnerability of the chromosomal end. This process is called telomere attrition and is considered another hallmark of aging (52).

Telomere shortening can cause genomic instability, and thus contributes to cancer or age-related diseases (115). This progressive diminishment in telomere length can influence gene expression and provoke several pathologies without damaging DNA (134). Moreover, ROS can trigger telomere attrition or loss and genomic instability, which can be prevented by using an antioxidant (115). In addition, telomerase activity can support cell cycle progression by preventing the arrest due to short telomeres, leading to a putative malignancy. Remarkably, overexpression of Beclin1 in HeLa cells revealed that telomerase activity is reduced after autophagy induction (39). This approach argues in favor of the hypothesis that autophagy plays an important tumor suppressor role by the modulation of telomerase activity in somatic cells. Similarly, Guanosine-rich zones, like telomeres, can suffer the formation of G-quadruplexes, interfering with gene expression and cell growth (135). In this regard, melanoma cells treated with an anthracene-based ligand (Ant 1,5, capable of stabilizing G-quadruplexes) showed an increase in genome instability and telomere dysfunction. Furthermore, Ant 1,5 induces a p21-dependent augmentation of autophagy levels. This autophagic response arises in order to avoid genome instability and telomeric dysfunction, thus promoting cell survival (135).

#### Other Hallmarks of Aging

Cellular senescence is considered another hallmark of aging (52). Senescence can be induced by DNA damage, telomere attrition or other stress signals, with the consequent cell cycle arrest. Several mouse models have contributed to the current knowledge of senescence and its characteristics (55). For example, Apolipoprotein E (ApoE) knockout mice present atherosclerotic plaques, which can be prevented by the addition of rapamycin. Furthermore, autophagy regulates the senescence of vascular smooth muscle cells (VSMCs) of ApoE−/<sup>−</sup> mice, involving the mTORC1/ULK1/ATG13 pathway in atherosclerosis progression (101). In addition, senescent MEF cells accumulate copper due to higher import and lower export, enhancing antioxidant defense mechanisms. In addition, rapamycin treatment can prevent and reverse copper accumulation, suggesting that autophagy mediates the copper homeostasis (136). Along the same lines, primary cultures from human fibroblasts depleted for Atg7, Atg12, or Lamp2 showed cell cycle arrest and high levels of SA-β-Gal staining, a characteristic feature of replicative senescence cells (137). Intriguingly, autophagy can mediate the transition to a senescent phenotype in IMR90 human diploid oncogene-induced senescence fibroblasts, making possible the protein remodeling needed to establish the senescent phenotype under oncogene activation (138). An interesting review conducted by Kwon et al. put forward a "toolkit" of differential diagnosis to resolve the apparent contradiction of autophagy in the cellular senescence. They proposed that type of autophagy, the exact moment when it acts, and the place where it occurs can define the pro or anti-senescence role of autophagy (139).

Another hallmark of aging, according to López-Otín et al., is stem cell (SC) exhaustion, with the consequent decline in tissue regenerative potential (52). Self-renewal is important to maintain the population of tissue-specific stem cells throughout life. Importantly, as we age, stem-cell activity decreases (140). In addition, SCs function is highly regulated in response to external stimuli (129). Ho et al. have shown that autophagy is necessary for preservation and quiescence of hematopoietic stem cells (HSCs). The authors demonstrated, in an Atg12- KO mice model, that autophagy-defective HSCs resemble old HSCs, in terms of accumulation of mitochondria, augmentation of myeloid-to-lymphoid ratio, diminishment of the regenerative potential, and decrease in self-renewal (141). This study supports the importance of autophagy in blood system homeostasis (142). Autophagy is also important to maintain stemness in bone marrow-derived mesenchymal stem cells (BMMSCs), and induction with rapamycin restores the biological properties of BMMSCs (143). In addition, Atg7 loss in aged muscle stem cells (satellite cells) of transgenic mice caused altered mitophagy and an accumulation of ROS, all features of senescence that diminish the regenerative potential of aged satellite cells (144). It is important to note that autophagic modulation could be an interesting therapeutic approach to prevent stem cell senescence and decline. Due to SC complexity, more studies are required to fully elucidate the role of autophagy in maintenance of stem cells.

## CURRENT AND FUTURE AUTOPHAGIC TREATMENTS IN AGING AND AGE-RELATED DISEASES

It can be presumed that induction or restoration of autophagy and antioxidant cellular systems could alleviate aging symptoms. Three major anti-aging therapies were evaluated over the last 30 years: Autophagic inducers, antioxidant (polyamine-rich) consumption, and caloric restriction.

Rapamycin, an immunosuppressive macrolide, is a wellknown autophagic activator via mTORC1 inhibition of the mTOR complex. This pharmacological treatment has proved to increase lifespan in flies, nematodes, yeast, and mice [reviewed in (111)]. Moderate doses of this drug can alleviate atherosclerosis, achieving the same effects as higher concentrations (101). Furthermore, rapamycin added during reperfusion after heart infarction in a C57 myocardial ischemia mouse model improved the survival and cardiac functioning, reducing the infarcted zone size as well as apoptosis post-ischemia/reperfusion. In addition, AKT phosphorylation increased after treatment, suggesting that AKT-ERK pathways were selectively activated by rapamycin (145). Finally, rapamycin improves whole metabolism in several ways, strengthening the importance of regulating autophagy activity by external compounds in order to ameliorate metabolic diseases at cellular and whole organism levels (100, 109).

In addition, exogenously administered spermidine extends lifespan in mice treated throughout life or at pre-aged adulthood, and in hypertensive rats as well. In addition, spermidine dietary intake was inversely correlated with cardiovascular pathologies in a human population-based cohort (146). Spermidine, thus, has been proposed to be a neuro and cardioprotector in aging models and humans, highlighting the importance of polyamine-rich diets (56).

Caloric restriction (CR) is the reduction of total calorie intake by 30-40% without malnutrition. There is strong evidence supporting this therapy as one of the most effective in reducing oxidative stress in rats and mice, prolonging lifespan (31, 111). CR has been found to provoke a decrease in ROS generation (69) and a diminishment of DNA damage (31). Regarding CR and autophagy, it was demonstrated that 8%-CR in combination with exercise or 8%-CR alone were capable of upregulating Atg7, LC3, and LAMP2 in type II skeletal muscle in rats (147). The authors also found that aging augmented BECLIN1 protein expression and oxidative stress, but CR alone or with exercise diminished this modification as well as the apoptotic index, both correlating negatively with LAMP2 gene expression (147). Sod−/<sup>−</sup> mice

#### REFERENCES


treated with CR attenuated the age-related-like phenotype of this knockout in terms of DNA damage, cellular senescence, and inflammation (49). CR could be beneficial to human health, according to epidemiological studies (37). As CR regulates several pathways, more integrative studies are required in order to fully understand its anti-aging effect.

Finally, several reviews and research studies highlight the importance of autophagic modulation as an anti-aging therapy for the future (32, 37, 43). Rapamycin, resveratrol, polyamines, and CR are possible candidates to be tested more carefully in order to improve the putative treatments for human age-related diseases.

## CONCLUDING REMARKS

Aging involves several features that can promote the development of a variety of disorders in aged individuals, such as neurodegenerative, heart, and metabolic diseases, as well as cancer. These age-related characteristics involve a gradual increase in ROS production and genome instability, and a progressive decrease in antioxidant, DNA repairing, and proteostatic systems, among others (**Figure 3**). Autophagy, as a homeostatic process, plays an important role in the maintenance of cell physiology and avoidance of any internal or external damage that could eventually appear. Several attempts were made to improve age-related features, such as caloric restriction as well as antioxidants and autophagy inducers. Rapamycin, resveratrol, and polyamines are autophagic inductors clinically available that could improve aging and some age-related disorders. It is important to highlight the fact that chemotherapies combined with autophagic inhibitors (i.e., chloroquine and derivatives) could be more effective in cancer treatment. Further studies are required to make autophagy modulation a more promising anti-aging and anti-tumoral therapy in the next decades.

#### AUTHOR CONTRIBUTIONS

MB analyzed the data, wrote the manuscript, and critically read the manuscript. RG wrote the manuscript. CF designed the research, analyzed the data, wrote the manuscript, produced the figures and critically read the manuscript.

#### FUNDING

This work was partly supported by grants from Agencia Nacional de Promoción Científica y Tecnológica PICT 2013-2335, PICTO 2016-0055, SeCTyP K013 (Universidad Nacional de Cuyo) to CF.


therapy in colon cancer cells. Photochem Photobiol Sci. (2017) 16:1631–42. doi: 10.1039/C7PP00161D


a safe and easy way to meet goals in the SENS agenda. Rejuvenation Res. (2006) 9:408–12. doi: 10.1089/rej.2006.9.408


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Barbosa, Grosso and Fader. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Chaperone Mediated Autophagy in the Crosstalk of Neurodegenerative Diseases and Metabolic Disorders

Iván E. Alfaro<sup>1</sup> \*, Amelina Albornoz <sup>1</sup> , Alfredo Molina<sup>2</sup> , José Moreno<sup>2</sup> , Karina Cordero<sup>2</sup> , Alfredo Criollo2,3,4 and Mauricio Budini 2,3 \*

<sup>1</sup> Fundación Ciencia & Vida, Santiago, Chile, <sup>2</sup> Dentistry Faculty, Institute in Dentistry Sciences, University of Chile, Santiago, Chile, <sup>3</sup> Autophagy Research Center (ARC), Santiago, Chile, <sup>4</sup> Advanced Center for Chronic Diseases (ACCDiS), University of Chile, Santiago, Chile

Chaperone Mediated Autophagy (CMA) is a lysosomal-dependent protein degradation pathway. At least 30% of cytosolic proteins can be degraded by this process. The two major protein players of CMA are LAMP-2A and HSC70. While LAMP-2A works as a receptor for protein substrates at the lysosomal membrane, HSC70 specifically binds protein targets and takes them for CMA degradation. Because of the broad spectrum of proteins able to be degraded by CMA, this pathway has been involved in physiological and pathological processes such as lipid and carbohydrate metabolism, and neurodegenerative diseases, respectively. Both, CMA, and the mentioned processes, are affected by aging and by inadequate nutritional habits such as a high fat diet or a high carbohydrate diet. Little is known regarding about CMA, which is considered a common regulation factor that links metabolism with neurodegenerative disorders. This review summarizes what is known about CMA, focusing on its molecular mechanism, its role in protein, lipid and carbohydrate metabolism. In addition, the review will discuss how CMA could be linked to protein, lipids and carbohydrate metabolism within neurodegenerative diseases. Furthermore, it will be discussed how aging and inadequate nutritional habits can have an impact on both CMA activity and neurodegenerative disorders.

Keywords: CMA, neurodegeneration, lipids, carbohydrates, metabolism

## INTRODUCTION

#### Autophagy

Autophagy is a cellular process in which proteins and organelles become degraded through lysosomes in order to maintain an adequate cellular homeostasis. In mammals, three different types of autophagy pathways have been described, (i) macroautophagy (mostly known as "autophagy"), (ii) microautophagy, and (iii) chaperone-mediated autophagy (CMA).

Macroautophagy and CMA, together with proteasome, are involved in the recycling of cellular proteins, including those that can compromise the normal physiology of the cell by adopting aberrant folded forms which are able to generate aggregates or inclusions (1). Macroautophagy is a multistep pathway that starts with the formation of an autophagosome, a double membrane vesicular structure that includes the cargo to be degraded. Autophagosome formation involves the participation of different factors like the Atg protein family, LC3-II, p62, Beclin, among others.After its formation, the autophagosome fuses to the lysosome where it releases its proteolytic material to form an autolysosome structure that finally degrades the cargo (1).

#### Edited by:

Maria Ines Vaccaro, Universidad de Buenos Aires, Argentina

#### Reviewed by:

Angelo Poletti, University of Milan, Italy Carsten Merkwirth, Ferring Research Institute, Inc., United States

#### \*Correspondence:

Iván E. Alfaro ialfaro@cienciavida.org Mauricio Budini mbudini@u.uchile.cl

#### Specialty section:

This article was submitted to Endocrinology of Aging, a section of the journal Frontiers in Endocrinology

Received: 19 September 2018 Accepted: 11 December 2018 Published: 31 January 2019

#### Citation:

Alfaro IE, Albornoz A, Molina A, Moreno J, Cordero K, Criollo A and Budini M (2019) Chaperone Mediated Autophagy in the Crosstalk of Neurodegenerative Diseases and Metabolic Disorders. Front. Endocrinol. 9:778. doi: 10.3389/fendo.2018.00778

In addition to proteins, macroautophagy targets can also include damaged and/or oxidized organelles such as mitochondria (mitophagy), endoplasmic reticulum (ERphagy), peroxisomes (pexophagy), lipid droplets (lipophagy), ferritin (ferritinophagy) and zymogen granules (zymophagy) (2–5).

CMA, which will be further described below, involves the participation of HSC70 (heat shock-cognate chaperone 70 KDa) and LAMP-2A (lysosomal associated-membrane protein 2A) proteins. With respect to macro- and microautophagy, CMA has been reported only to degrade cytosolic proteins previously bound to HSC70-specific sequence (see below) (6, 7). In the case of microautophagy, it carries the degradation of proteins and organelles out, by their direct engulfment of the lysosome. In microautophagy, whereas some of the proteins can be incorporated into the lysosome through an "unspecific way", others, like those in CMA, also need the interaction with the Hsc70 chaperone. The similarities and differences between CMA and microautophagy have recently been discussed in the review of Tekiradg and Cuervo (8).

#### Chaperone Mediated Autophagy (CMA) CMA Overview

Chaperone-mediated autophagy (CMA) is a specific lysosomaldependent protein degradation pathway (9). It differs from macroautophagy because, (i) it does not involve the formation of autophagosomes and autolysosomes, (ii) its targets are cellular proteins and not organelles, (iii) the protein cargo is directly delivered into the lysosomal lumen through the interaction with HSC70 and LAMP-2A (10). HSC70 belongs to the Hsp70 protein family and it has a constitutive expression, participating principally in the CMA pathway but also in microautophagy (8). LAMP-2A protein is the A isoform of Lamp-2 (lysosome associated-membrane protein type 2) and restricts the CMA degradation process. Importantly, in CMA, selectivity resides in the fact that all CMA substrate proteins contain at least one amino acidic motif biochemically related to the penta-peptide KFERQ (11, 12). In the cell cytoplasm, HSC70 recognizes the KFERQ sequence in target proteins to form the complex HSC70 substrate. Although the temporality of this process is not well known, the complex HSC70-substrate interacts with the cytosolic tail of LAMP-2A that, in turns, drives the translocation of the target protein into the lysosome lumen (13). All cells have a basal CMA activity which helps to maintain the homeostasis of many cellular proteins. However, under certain stimulus, like nutrient starvation, serum deprivation or cellular stress (e.g., protein aggregation), CMA activity increases. This up-regulation condition can be visualized by mRNA LAMP-2A overexpression, HSC70 and LAMP-2A co-localization and LAMP-2A positive lysosomes with increased perinuclear distribution (14–16).

#### CMA Target Recognition

In 1985 Dice et al. found a pentapeptide in the Ribonuclease A (RNase A) which was necessary for the degradation of this protein by lysosomes in fibroblasts (11, 17–20). When different proteins were analyzed, it was found that the best aminoacidic sequence, to be present in a protein that is going to be degraded through CMA, corresponds to the pentapeptide KFERQ (11). The general rule, is that this sequence is composed of an amino acid Glutamine (Q), that must contain in one side (right of left) a tetrapeptide, including basic-, acidic- and hydrophobic- amino acids (11). Target proteins for CMA should contain at least one KFERQ-like domain, however some of them include more than one (21). Proteins containing a KFERQ-like domain do not enter the lysosomes by themselves, they first need to be recognized and bound by the chaperone HSC70 (22). This protein, also known as HspA8, is a 73 kDa protein belonging to the Hsp70 protein family (23). Thus, HSC70 specifically recognize the KFERQ-like motive in cytosolic proteins to target them for lysosomal degradation through CMA (22). In some cases, other co-factors, or covalent modifications have been found to be necessary for the binding between the complex HSC70-protein substrate and LAMP-2A. For example, STUB1/CHIP (a ubiquitin E3 ligase) assists the degradation of HIF1A through CMA by the ubiquitination of Lysin 63 (K63) (24) and also by supporting the interaction of LAMP-2A through its chaperon binding domain (24). In addition, covalent modifications, like acetylation and phosphorylation have been described to create KFERQ-like domains capable of targeting proteins for CMA degradation (25–29).

#### CMA Substrate Translocation and Degradation

In the lysosome membrane, the HSC70-substrate complex interacts with the lysosome receptor LAMP-2A. The protein LAMP-2A was first described by Dr. Ana María Cuervo and Dr. J. Fred Dice (30). It was identified as the LGP96 (lysosome membrane glycoprotein 96 kDa) and it was found to be a necessary receptor for the specific degradation of GAPDH (Glyceraldehyde- 3-phosphate dehydrogenase) and RNase I (Ribonuclease I) through lysosomes (30). The authors showed that both GAPDH and RNase I, were able to bind the LGP96, which were isolated from rat liver (30). The Lysosome-associated membrane glycoprotein 2 (LAMP-2) is able to generate three variants through its pre-mRNA alternative splicing, (LAMP-2A, Lamp-2B, and Lamp-2C) but, to date, it is well established that only the variant A is responsible for the CMA activity (31). Compared to variants B and C, LAMP-2A has four positively charged residues in its C-terminal domain that specifically allow its interaction with target proteins (31). Characterization of LAMP-2A showed that multimerization of this variant (and not the others) in the lysosomal membrane is directly correlated with CMA activity (13, 31). The substrate up-take by the lysosome is a step that depends on LAMP-2A (32). Monomeric LAMP-2A is able to interact with substrates, however, substrate translocation is dependent on the formation of LAMP-2A oligomers (32). Further evidence showed that LAMP-2A arranges in a stable homotrimer, with helical transmembrane domains bound by a coiled-coil conformation, and with the cytosolic tails interacting with the complex HSC70-substrate protein (13). In addition, LAMP-2A oligomerization was shown to be regulated by different proteins. Glial fibrillary acidic protein (GFAP) helps to stabilize the translocation complex in an EF1α dependent manner. The stabilizing effect of GFAP is disrupted by the association of EF1α to GTP, which in turn is released from the translocation complex and allows the self-association between

GFAP molecules. The self-interaction between GFAP molecules have a negative impact on the stabilization of the translocation complex. Thus, GTP acts as an inhibitor of CMA activity (33). Phosphorylating and dephosphorylating signals also regulate CMA activity at the level of the translocation complex (34). For example, mTORC2 (mammalian target of rapamycin complex 2) has been shown to activate, through phosphorylation, Akt kinase, which in turn inhibits the CMA activity. The mechanism by which active Akt decreases the CMA activity is not yet well established. However, it has been postulated that active Akt phosphorylates GFAP, destabilizing its binding to the LAMP-2A translocation complex. In this scenario, both, mTORC2 and Akt act as negative regulators of CMA activity. On the contrary, Pleckstrin homology (PH) domain and leucine-rich repeat protein phosphatase 1 (PHLPP1), has been shown to be recruited to the lysosome membrane in a Rac1 dependent manner. On the lysosomal membrane, PHLPP1 induces CMA activation by dephosphorylating Akt at the same residues previously phosphorylated by mTORC2. Thus, PHLPP1 is a positive regulator of CMA activity (34). In addition, the binding and up-take of the CMA substrates through the translocation complex is assisted by HSC70 and Hsp90 (heat shock protein 90) chaperones. In the case of Hsp90, it was shown that this protein stabilizes the binding of LAMP-2A at the lysosomal membrane (32). HSC70, apart from its important role in binding to the KFERQ domain, participates in two additional steps of the CMA process. One step is related to the recycling of LAMP-2A in the absence of substrates, where HSC70 has been proposed to support the destabilization of LAMP-2A from the translocation complex (32). The other step occurs at the lumen of the lysosome, where a lysosomal HSC70 (lys-HSC70) is also involved in the up-take process of the substrate. The blockage of lys-HSC70, by a specific antibody, has been directly correlated with an inhibition of the CMA activity (35). **Figure 1** summarizes the main aspects of the CMA mechanism.

#### CMA Activation

Koga et al. (36) showed that basal CMA activity is present in a variety of cell types. However, an up-regulation in this pathway can be observed under different stimulus or conditions. The most common stimulus for CMA activation is nutrient deprivation (or starvation). Starvation activates CMA both in vitro and in vivo (11, 20, 36) and, although the exact mechanism has not been described yet, at least in vivo, it has been proposed that it depends on the circulating ketone bodies (37). CMA over-activation has also been observed under DNA damage. Under this condition, CMA activity is up-regulated with the purpose to degrade the checkpoint protein kinase 1 (Chk1). The accelerated degradation of Chk1 by CMA reduces its nuclear entrance and consequently decreases the phosphorylation and destabilization of the MRN (Mre11–Rad50–Nbs1), a complex that participates in the early steps of particular DNA repair pathways (38). In line with this result, DNA irradiation of a hepatocellular carcinoma cell line also over-activates CMA. In this cell line, CMA up-regulation was responsible for the degradation of HMGB1 (high-mobility group box 1 protein) degradation, which in turn, provoked the down-regulation of the p53 protein and the consequent decrease in the apoptotic rate. In this case, CMA activation could be considered a mechanism by which this type of carcinoma is able to resist the irradiation treatment (39). Oxidative stress also activates CMA constitutively. Both, mRNA LAMP-2A levels and the recruitment of LAMP-2A protein to the lysosomal membrane (40, 41) were augmented under oxidative stress. The reasons by which CMA is activated under oxidative stress are unknown.

FIGURE 1 | Chaperone mediated autophagy mechanism. The scheme shows the principal events in CMA mediated translocation of proteins through the lysosomal membrane and known regulatory mechanisms. (A) LAMP-2A exist as an inactive oligomer at the lysosomal membrane. CMA protein substrates containing KFERQ motifs and HSC70 can bind to the cytosolic tail of LAMP-2A with similar affinity. The short C-terminal tail of LAMP-2A exposed to the cytoplasm and conformed by 12 amino acids probably binds to the chaperone and the substrate in a competitive manner. However, on the substrate, binding sites for HSC70 and LAMP-2A do not overlap allowing the HSC70-substrate complexes to interact with LAMP-2A oligomer in a substrate dependent manner. (B) In a dynamic fashion, GFAP favors the formation of high molecular weight aggregates of LAMP2A. Somehow, coupled and simultaneous binding of the substrate to HSC70 and LAMP-2A is sensed by the transmembrane domains of the LAMP-2A for rearrangements in the supramolecular complexes. Next, substrate unfolding and translocation to the lysosomal lumen occurs. A lysosomal-HSC70 (Lys-hsc70) collaborates in this process. (C) A signaling pathway involving phosphorylating and dephosphorylating signals, where AKT, MTORC2, PHLPP1, RAC1, and GFAP, regulate the stabilization of the LAMP-2A complex at lysosomal membranes.

However, a possibility could be that CMA acts as a degradative backup system for an inhibited macroautophagy pathway (42). With respect to the molecular mechanism, recent work showed that under hydrogen peroxide treatment, the transcription factor NFE2L2/NRF2 (nuclear factor, erythroid derived 2, like 2) can drive the expression of the lamp2 gene. Pharmacological activation of NFE2L2/NRF2 further supported the role of this transcription factor on lamp-2A gene overexpression (43). However, under oxidative stress, CMA up-regulation could also have a toxic effect. It was shown that CMA over-activity strongly reduces the levels of MEF2A (myocyte enhancer factor 2a), a protein necessary for the adaptive response of the cells to oxidative stress (41). Another stressor that up-regulates CMA activity is hypoxia. Over-activation of CMA was directly correlated with the survival of neuronal cells exposed to different hypoxic conditions and, on the contrary, the down-regulation of this pathway sensitized the cells to the stress (44). On the other hand, it was reported that HIF-1α, a protein that activates HIF-1 (Hypoxia-inducible factor-1) and mediates an adaptive response to hypoxia, is degraded by CMA in cells exposed to hypoxic conditions (45). Therefore, the exact mechanisms by which CMA protects cells against hypoxia, needs to be further elucidated. A constitutive activation of CMA has also been observed in different models of neurodegenerative diseases. For example, in cellular and mouse models of Huntington Disease, CMA was activated as a compensatory mechanism for the inhibition of macroautophagy. In this case, both lys-HSC70 and LAMP-2A protein levels were markedly augmented. In particular, there was an observed stability in LAMP-2A proteins at the lysosomal membrane and a transcriptional upregulation of this splice variant (46). Up-regulation of CMA was also observed in molecular mechanisms related with Parkinson's disease. For example, recent studies performed with a dopaminergic neuronal cell line, using an endoplasmic reticulum (ER) stress mouse model, showed that ER stressors are able to induce CMA activation. The mechanism for such activation involved the recruitment of MKK4 protein kinase to the lysosomal membrane and the subsequent phosphorylation of LAMP-2A by p38 MAPK protein kinase (47). This regulatory mechanism for CMA upregulation is relevant as neurotoxins associated with PD are directly correlated with an activation of the ER-p38 MAPK–CMA response pathway (48). Additionally, in PD models, LAMP-2A and HSC70 were observed to be up-regulated when α-syn was over-expressed in vivo, indicating that a CMA over-activation is triggered in response to a α-syn pathological condition (49)

#### CMA and Metabolism Participation of CMA in the Regulation of General Protein Metabolism

Protein homeostasis is regulated by a series of interconnected signaling pathways that sense amino acid availability, among them, the integrated signaling pathway (ISR) through the general control nonderepressible-2 protein (GCN2) and the V-ATPase/Ragulator/mTOR pathway located in lysosomes (50, 51). GCN2 is a protein kinase that is activated in the presence of accumulated deacetylated tRNAs, sensing amino acid depletion. The main target of GCN2 is the eukaryotic initiation factor 2 (eIF2), whose phosphorylation induces the inhibition of general translation and favors the synthesis of a specific set of mRNAs that regulates cell survival and promotion of macroautophagy (52). On the other hand, mTORC1 kinase is activated in response to amino acid availability, promoting translation and inhibiting macroautophagy. The mechanism of amino acid sensing, coupled with mTOR activation, involves the lysosomal amino acid transporter SLC38A9, and mTORbinding proteins, Sestrin2 and Castor1 (50). These factors that regulate mTOR activation or inhibition, also modulate mTOR activity and thus mTOR-downstream targets involved in promotion of protein translation, cell growth and anabolism, and inhibition of macroautophagy initiation. In response to amino acid deprivation, mTOR activity is attenuated, protein translation is inhibited and macroautophagy is maximally activated to restore amino acid levels.

These non-selective mechanisms of protein homeostasis, added to a variety of mechanisms of stabilization and destabilization of protein pools, define the average life time of every protein in the cell, in nutrient rich conditions and in response to nutritional stress. Selective mechanisms of protein degradation are in contrast, generally linked to the modulation of the function of specific proteins or to the degradation of misfolded proteins that have been targeted for efficient elimination. Selective and non-selective mechanisms of proteasomal and macroautophagy protein degradation pathways have been extensively reviewed in several articles (53–56). Whereas proteins degraded by CMA need to be recognized specifically by HSC70 protein through its KFERQ domain, CMA can be considered a selective protein degradation process. An initial analysis indicated that at least 30% of cytosolic proteins contain CMA KFERQ-like motifs (7). These proteins mainly correspond to a fraction of long-lived proteins that were more rapidly degraded by lysosomes in response to serum starvation or amino acids withdrawal. Indeed, at least 90% of enhanced proteolysis observed during serum withdrawal in fibroblast cells is thought to be part of the protein pools targeted for degradation by CMA (20). In comparison to macroautophagy, the increase in degradation rates of the pool of proteins targeted to CMA is observed in response to prolonged periods of serum or nutrient deprivation (7, 12, 57). Indeed, macroautophagy activity decreases in activity after 6–12 h of nutrient removal, but CMA progressively increases in response to prolonged periods of serum or nutrient deprivation (57). Thus, CMA is a protein degradation pathway specifically activated during long term starvation. The pool of proteins degraded by CMA is very specific and different to the protein pools degraded by other protein catabolic neutral or acid proteolytic pathways. Additionally, CMA-targeted proteins accumulate in lysosomal membranes under starvation and, probably, because of this reason, this pool is excluded for degradation by proteasome or macroautophagy.

The pool of proteins targeted by CMA depends, most likely, on the periods of deprivation, type of deprivation (glucose, amino acids, serum or specific growth factors) and the cell type in the study (12). In addition, the selective pool of proteins degraded by CMA would also be associated with certain functions capable of overcoming metabolic changes in response to starvation. This has been observed in the regulation of glycolytic flux, through the degradation of glycolytic enzymes by CMA, in order to protect cells from apoptosis (7, 21). On the other hand, proteins that do not contain KFERQ-like regions would most likely be protected against degradation, to sustain the critical functions of cells under nutritional stress. It has also been suggested that activation of non-selective macroautophagy in the first stages of starvation, help to provide metabolite building blocks (as amino acids) for the continuous synthesis of macromolecules. However, at prolonged times of starvation, a relief of macroautophagy by selective CMA-dependent protein degradation, would be necessary to avoid the elimination of essential proteins for cell survival (57). It is unknown whether nutrient deprivation directly controls the CMA activity, as with macroautophagy. The main hypothesis is that an indirect regulation of CMA by nutrient deprivation would be carried out by translational and transcriptional up-regulation of lysosomal proteins and genes involved in lysosome biogenesis (58–60). However, a recent report indicates that upregulation of CMA was responsible for mTOR activation by a mechanism that sensed the augmentation of the free amino acid content dependent on CMA activation (61). In addition, increased CMA activity, in response to LAMP2-A overexpression, was also shown to down-regulate macroautophagy through mTOR activation (62). Overall, this latter evidence reveals that CMA activation functions as a negative feedback loop for mTOR and, thus, would regulate the general amino acid and protein homeostasis signaling.

As mentioned before, basal CMA activity can be detected in different tissues, suggesting that this pathway has a constant role in the regulation of certain proteins, even under conditions of nutrient or growth factor availability (57). However, whether the degradation of these target proteins increases under CMA activation (e.g., starvation), or another specific cell response, remains to be clarified. Most likely, the latter will depend on the different affinities that these proteins would have for HSC70 and LAMP-2A, and/or on the availability of HSC70 and LAMP-2A (63).

As a consequence of its important role in controlling protein homeostasis, CMA is able to regulate metabolic pathways that are crucial to maintain balance in cellular and systemic physiology.

#### CMA and Regulation of Lipid Metabolism

Neutral cytosolic lipolysis and lysosome-associated acid lipolysis were classically classified in two different pathways. However, both are now recognized as synergistic, cooperative and interconnected mechanisms that contribute to lipid catabolism (64). Lysosomal lipolysis involves the degradation of lipids derived from endocytosis, as well as from cytosolic lipid stores (lipid droplets) through an autophagic process known as Lipophagy (65). Lipid droplets (LDs) are specialized organelles that store neutral lipids during fatty acid availability, mainly triglycerides (TG) and cholesterol esters, for their posterior use as energy or precursors for lipids and steroids (66). LDs consist of a core of neutral lipids separated from the aqueous cytoplasm by a phospholipid monolayer decorated by a set of proteins. These proteins are enzymes involved in lipid synthesis and lipid hydrolysis, membrane-trafficking proteins and a set of specialized LDs-associated structural proteins called perilipins (PLINs). PLINs cover a family of 5 major proteins and some additional splice variants expressed in lower amounts. Perilipin 1 (PLIN1) and 2 (PLIN2) are exclusively associated with LDs whereas perilipins 3–5 are found in the cytoplasm, associated with LDs, as well as microdroplets and lipoprotein particles (67). Perilipins 1, 2 and 5 control, at the surface of the lipid droplets, the access of cytosolic neutral lipases. For example, phosphorylation of PLIN1 and PLIN5 by protein kinase A (PKA), in response to hormones or growth factors, trigger the recruitment of neutral lipases like hormone sensitive lipase (HSL) and triacyl-glycerol lipase (ATGL) (68, 69). In contrast to PLN1 and PLIN5, PLN2 is not phosphorylated by PKA and does not recruit lipases at LDs (67). Similar to PLIN2, PLIN3 and PLIN4 protect LDs from degradation and they seem to have a tissuespecific function in the regulation of LDs size, maturation and mitochondrial association (70–72).

The involvement of autophagy in lipid turnover was first demonstrated for macroautophagy (65). Electronmicroscopy studies indicated that autophagosomes engulf LDs or small portions of large LDs for degradation, forming autolipophagosomes in a process now known as macrolipophagy. In hepatocytes, macrolipophagy and the presence of lipidcontaining autophagolysosomes, is stimulated by nutrient starvation and low fatty acid treatments (65). Members of the Rab GTPase family, RAB7 and RAB10 have been involved in the early steps of autophagosome recruitment to LDs (73). Moreover, the Rab effectors, EHBP1 and EH2, have been suggested to help in the elongation of autophagosomal membranes around LDs (74, 75). Neutral lipases ATGL and HSL contain LIR motifs (LC3-interaction region) and interact with the cytosolic face of autophagosomes through LC3-II, being recruited to LDs in response to macrolipophagy activation (64, 76).

In addition to macroautophagy, CMA also plays a central role in lipid homeostasis. In a mouse model, where LAMP-2A was specifically down-regulated in liver by deleting the exon 9 of lamp2 gene, it was shown that the constitutive blockade of CMA activity induces pronounced steatosis without a substantial increase in TG synthesis (77). In addition, LAMP-2A knock-out cells were insensitive to lipolysis induced by starvation and displayed increased density, size and occupancy area of LDs (78). Interestingly, in this model, non-selective macroautophagy was intact, indicating a specific and new role of CMA in lipophagy (77). In line with this evidence, further studies found the presence of KFERQ-like motifs in PLIN2 and PLIN3 proteins, and identified these two proteins as substrates for CMA lysosomal degradation. In addition, lysosomes actives in CMA isolated from rat liver, were enriched in PLIN2 and PLIN3, supporting the fact that these proteins are CMA substrates. This evidence was further complemented by experiments indicating that PLIN2 and PLIN3 degradation was reduced in LAMP-2A knockout cells (78). With regards to the mechanism, the degradation of PLIN2 through CMA in response to nutrient starvation and lipid challenges, was correlated with an increment in the association between HSC70 and perilipins (78). Reduced degradation of PLIN2 in cells that are deficient for the CMA pathway, leads to reduced lipolysis, likely due to an impairment in the association of ATGL with LDs as well as due to the defective recruitment of autophagosomal proteins and Rab7 to LDs (25, 78). Moreover, PLIN2 phosphorylation by AMPK was found to be a necessary modification for PLIN2 association with HSC70 in a form dependent on CMA activation (25, 78). Therefore, the role of CMA in lipophagy not only relays in the regulation of the recruitment of neutral lipases to LDs trough perlipin degradation, but also in the integration of a regulatory step that involves a post-transcriptional modification guided by AMPK (25). Thus, CMA itself could be considered a nutrient sensing pathway supporting the fact that lysosomes can play a role as initial sensors and integrators of lipid homeostasis (79). Additional research, however, is needed to determine how cellular pathways involved in cellular lipid homeostasis communicate with CMA and lipophagy, and how CMA-dependent lipolysis affects different tissues.

#### CMA and Regulation of Carbohydrate Metabolism

As mentioned before, glycolytic enzyme glyceraldehyde 3 phosphate dehydrogenase (GAPDH) was one of the first characterized as a CMA substrate (80). Additionally, at least 8 of 10 glycolytic enzymes contain CMA-targeting motifs (77). Substrates of CMA that participate in glycolysis, validated by LAMP2-A loss of function or lysosome-transport assays, also include hexokinase-2 (HK-2) (28), the M2 isoform of pyruvate kinase (81), aldolase-A, enolase-1, 3-phosphoglyceric phosphokinase, phosphoglycerate mutase, glucose-6-phosphatedehydrogenase and the mitochondrial proteins glutamate dehydrogenase, ornithine transcarbamoylase and malate dehydrogenase (77). Other related substrates that have been suggested to be regulated by CMA in their function include phosphofructokinase-2 (82) and aldolase B (83). Glycolytic flux regulation by CMA has been confirmed in vivo in a mouse model with specific down-regulation of LAMP-2A in hepatocytes (77). These mice displayed higher protein levels of glycolytic enzymes and enzymes from the tricarboxylic acid cycle (TCA), a reduction in hepatic gluconeogenesis, lower glycogen synthesis and an increase in lactate production and TCA intermediates (77). This metabolic profile suggests a switch in hepatic metabolism to carbohydrate consumption as a source of energy vs. glucose biosynthesis in response to low CMA activity (77). On the other hand, classical inhibition of hepatic glycolysis caused by serum starvation (84) was not observed in mice with liver-specific CMA down-regulation (77). These results suggest that CMA activity would be necessary for a metabolic adaptive mechanism that triggers glucose production in liver to support peripheral organs under nutritional stress conditions.

The mechanism regulating CMA in response to changes in glucose availability are not fully understood. Pointing to a central role of the lysosome in sensing glucose homeostasis, new evidence indicates that glucose starvation induces changes in lysosomal acidification in an AMPK activity dependentmanner (84, 85). The mechanisms implicated in this regulation may involve a glucose-dependent regulation of the lysosome biogenesis through the transcription factor EB (TFEB) (86). Additional research is needed to elucidate how these lysosomal changes, induced by carbohydrate availability, regulate the CMA activity and, in turn, how this affects the cellular glycolytic flux.

### CMA and Neurodegenerative Diseases

There is increasing evidence supporting the idea that dysregulation in the CMA pathway plays a crucial role in neurodegeneration.

#### Parkinson's Disease (PD)

Evidence indicates that a dysregulation in CMA could impact on the onset or progression of Parkinson's Disease (PD). As mentioned above, the main protein associated with this neurogenerative disorder, alpha-synuclein protein (α-syn), has been identified as a CMA substrate (87). More specifically, reduced α-syn degradation was observed when its KFERQ motif was mutated and the expression of LAMP-2A was knockeddown. The involvement of CMA in α-syn degradation was confirmed in different neuronal cell lines (PC12 and SH-SY5Y) and primary cultures of cortical and midbrain neurons (87). One of the hallmarks of PD is the neurotoxicity caused by the abnormal aggregation of α-syn. In this context, mutations in the protein impair its degradation through a CMA pathway, causing the accumulation α-syn oligomers that are unable to be degraded by the lysosome. This event blocks the entire CMA pathway, enhancing the oligomers formation and compromising the degradation of other CMA substrates (88, 89). As mentioned above, LAMP-2A and HSC70 were observed to be up-regulated when α-syn was over-expressed in vivo (49). In line with these results, it was shown that the down-regulation of LAMP-2A in adult rat substantia nigra, via an adeno-associated virus, induced intracellular accumulation of α-syn puncta. In addition, LAMP-2A down-regulation was also correlated with a progressive loss of dopaminergic neurons, severe reduction in striatal dopamine levels/terminals, increased astro- and microgliosis and relevant motor deficits (90). Furthermore, studies using the Drosophila melanogaster model, showed that the overexpression of human LAMP-2A protein protected the flies from progressive locomotor and oxidative defects induced by neuronal expression of a human pathological form of α-syn (91). Also, other PD related proteins seem to be regulated by CMA activity. This is the case of PARK7/DJ-1 and MEF2D. PARK7/DJ-1 is an autosomal recessive familial PD gene that plays a critical role in the antioxidative response and its dysfunction leads to mitochondrial defects. CMA was shown to degrade, preferentially, an oxidized and altered form of PARK7, providing protection from mitochondrial damage and impairing cell viability (92). For MEF2D (myocytespecific enhancer factor 2D protein), it was shown that an inhibition of CMA provoked the accumulation of this protein in the cell cytoplasm (93). These results are in line with the observation that MEF2D levels are increased in brains of αsyn transgenic mice and in samples from patients with PD (93). Thus, an impairment in CMA activity, caused by the αsyn oligomers, could be the reason for MEF2D accumulation in PD. Beyond the studies performed in cell cultures and in vivo models, strong evidence indicates that CMA could also be dysregulated in individuals affected by PD. In this regard, LAMP-2A and HSC70 were observed to be down-regulated in PD patients, suggesting that a CMA dysregulation can occur before the appearance of α-syn aggregation and other PD associated disorders (94, 95). Although the mechanisms that lead to a reduction in LAMP-2A and HSC70 levels in PD remain unknown, some studies suggest that genetic variations in the promoter of the lamp-2A gene and the upregulation of different microRNAs that target both LAMP-2A and HSC70, could be implicated in their down-regulation (96, 97).

#### Alzheimer Disease (AD)

The association between CMA and Alzheimer's Disease (AD) can be described by the degradation of the RCAN1 protein (Regulator of calcineurin 1) through this pathway. Two KFERQ-like motifs identified in RCAN1 were responsible for its degradation by the CMA pathway (98). In addition, the inhibition of CMA pathway increased RCAN1 protein levels and consequently reduced the NFAT transcription factor activity (98). Interestingly, NFAT has been involved in the transcriptional regulation of the lamp2 gene (99). Thus, inhibition of CMA activity could be further enhanced by increased levels of RCAN1, that will subsequently impair NFAT-dependent transcription of lamp2. On the other hand, the Tau protein, one of the principal factors associated with AD, has also shown to be a CMA substrate (100, 101). The Tau protein is degraded by the autophagy– lysosomal system producing different fragments that, in turn, bind to HSC70 and become CMA substrates. Although these fragments are able to reach the lysosomes, they remain bound to the lysosomal membrane, causing the formation of pathological Tau aggregates that cause lysosomal damage and block the degradation of other CMA targets (100). Additionally, the amyloid precursor protein (APP), can be processed to produce another pathogenic molecule associated with AD, the β-amyloid peptide (Aβ), which contains a KFERQ related motif (102). However, deletion of KFERQ did not affect its binding to HSC70, but did somehow keep the APP and its C-terminal fragments (CTFs) away from lysosomes. Thus, the KFERQ-like domain of APP is relevant to preclude the accumulation of APP and its CTFs but, in this case, in an CMA degradation independent manner (102).

#### Huntington Disease (HD)

As previously mentioned, a constitutive activation of CMA has been observed in cell lines and mouse models of HD (46, 103). In a mouse model for this disease, a strong co-localization between LAMP-2A and HSC70 was correlated with augmented lysosomal degradation of the native and aberrant huntingtin protein (Htt) (46). Furthermore, additional studies performed in vivo also showed that Htt aggregates could be forced to be degraded through CMA, by targeting them with a polyQ binding protein (QBP1) including a KFERQ domain in its amino acidic sequence (104). The data indicated that modulation of the CMA pathway can be a plausible strategy for HD treatment.

#### Amyotrophic Lateral Sclerosis (ALS) and Frontotemporal Lobar Degeneration (FTLD)

Amyotrophic Lateral Sclerosis (ALS) and Frontotemporal Lobar Degeneration (FTLD) are two important neurodegenerative diseases where the implication of CMA has been poorly studied. So far, only two studies have connected these disorders with CMA and both focus on the ribonuclear protein TDP-43. Huang et al. described an KFERQ-like domain in TDP-43 that was responsible for the interaction of this protein with HSC70 under ubiquitination condition. Mutation of the KFERQ-like domain disrupted the ubiquitin-dependent binding of TDP-43 with HSC70. In addition, the down-regulation LAMP-2A by siRNA treatment, seems to increase the level of the pathologicallyrelated 25-KDa and 35-KDa TDP-43 C-terminal fragments, but not the full-length protein (105). On the other hand, in a recent study, Tamaki, Y., et al designed an antibody able to recognize abnormal (unfolded or mislocated) TDP-43. The addition of the KFERQ sequence to this antibody, was able to promote the degradation of abnormal TDP-43 through the CMA pathway. Although in this case the connection of TDP-43 with CMA was not in a physiological context, this work indicates that a forced CMA target degradation of TDP-43 can be used as a strategy to ameliorate neurodegenerative diseases associated with this protein (106).

#### CMA in the Crosstalk With Metabolic Disorders and Neurodegeneration Reduced CMA Activity With Aging

Aging can be considered as a physiological process that is associated with metabolic and neurodegenerative disorders via mechanisms that in many cases are completely unknown. With regards to autophagy, it has been observed that this lysosomal degradative mechanism also declines with age (107). In the case of CMA, this was determined by isolating lysosomes from the liver of young and old rats (108). Lysosomes from old rats had reduced LAMP-2A protein levels at both, total and lysosomal levels. The decrease of the LAMP-2A protein in lysosomes is associated with changes in lipid composition of these organelles that accelerate the degradation rate of the protein (108). With respect to the HSC70 protein, although the total protein levels remained unchanged, lysosomal HSC70 levels were augmented, probably to compensate the decline observed in the CMA activity (108). Further experiments demonstrated that, although the levels of Lamp2A mRNA were unchanged, there was a problem in the localization of LAMP-2A at the lysosomal membrane in old rats, consequently reducing the CMA activity (109). Furthermore, by using an animal model with specific downregulation of CMA in the liver, it was possible to demonstrate that reduced CMA activity in old animals had an impact on the ability to overcome proteostasis induced by different stressors like oxidative stress, lipid challenging and aging (110). The mentioned evidences clearly demonstrate that CMA activity is reduced in older individuals and that this protein degradative pathway helps to stabilize the different disorders associated with aging.

Thus, considering that CMA activity is reduced with age, it is possible to argue that a decline in CMA activity during aging can be a risk factor for the development of neurodegenerative disorders associated with adult and senior people. In fact, beyond of what it can be concluded from in vitro and in vivo models, more evidence is coming out revealing that a dysregulation in CMA activity can be present in human neurodegeneration. Most of the evidence is a result of studies on PD. For example, a study performed with peripheral blood mononuclear cells from PD patients, showed that affected people had reduced HSC70 protein levels (111). Another study, analyzing CSF (cerebrospinal fluid) of PD patients with mutations in the repeat kinase 2 (LRRK2) gene, confirmed that LAMP-2A protein levels were reduced in affected females, compared to healthy people (112).

FIGURE 2 | Consequences of downregulated CMA activity. The scheme shows the principal mechanisms of CMA downregulation. Reduced expression of the lamp2 gene has been related with genetic polymorphisms (96), aging (108) and metabolic disorders (132). In relation to the latter situation, HFD has been shown to modify the lipid composition of lysosomal membranes, including increased levels of cholesterol and ceramide (132). These modifications induces the formation of organized lipid microdomains, which prevent LAMP-2A multimerization and stimulate subsequent degradation through proteolysis by Cathepsin D (CatD) (133). Independently of the mechanism, a reduction in CMA activity can induce an increase in the levels of some proteins related with the development of neurodegenerative diseases. In the case of tau, α-synuclein or other less characterized CMA protein substrates that are able to aggregate, increases the total mass that these proteins would contribute to protein oligomerization, an event associated with a blockade of CMA transport. In addition, oxidative damage, would enhance the aggregation process. Additionally, changes in signaling pathways that control LAMP2A oligomerization and membrane stabilization can alter LAMP-2A membrane dynamics and CMA activity. Finally, CMA down-regulation can alter a series of other specific proteins related with neuronal survival or neuroinflammation, that participate in the development of neurodegenerative diseases.

However, studies focused on analyzing the levels of LAMP-2A and HSC70 in healthy people during normal aging, failed to find any positive association (113). Thus, future studies need to focus on whether changes in CMA activity (or CMA protein players) are altered during normal aging in healthy people, or if they are only associated with patients affected by neurodegeneration. The latter will help to elucidate whether observed changes in CMA protein players are a cause or a consequence of these disorders.

#### CMA Activity, Metabolic Disorders and Neurodegenerative Diseases

In vivo models strongly support the hypothesis that CMA activity decreases with aging. Considering this evidence, which of these could be the consequence of CMA activity in people with a high fat diet (HFD)? Rodriguez-Navarro et al. showed that chronic exposure of rats to an HFD or cholesterolenriched diet, provoked a decrease in CMA activity (114). Further characterization of these animals showed that isolated lysosomes from liver have reduced LAMP-2A protein at lysosomal membranes. The latter was triggered by an accelerated degradation of lysosomal LAMP-2A, something also observed in older animals. Interestingly, authors also found that lysosomal membranes from animals with a HFD presented changes in lipid composition, in a similar way than that was observed with age (114). Thus, a dysregulation of CMA activity produced by aging, could impact on the LDs composition, affecting the aggregation of some proteins involved in neurodegenerative diseases.

In addition, a systemic decrease in CMA activity, and the onset or progression of neurodegenerative disorders, could be enhanced by an HFD. Different works demonstrate that a HFD can be a risk factor for the development of neurodegenerative diseases. For example, in a mouse model of PD, it was observed that HFD provoked a reduction in Parkin protein levels (115). In addition, similar results were obtained with a α-syn transgenic mouse model of PD. Compared with wild type mice, an HFD causes obesity and glucose intolerance in transgenic mice. Furthermore, transgenic mice also had an accelerated onset of PD disease and premature movement phenotype and death (116). The same situation was observed in transgenic mice models for AD. The model with an HFD, presented exacerbated neuropathology, defects in synaptic stability/plasticity, apoptotic neuronal cell death (117) and increased levels of insoluble amyloid-β (AB) and Tau (118). On the other hand, LDs have been found to play a role in neurodegenerative diseases. For example, as reviewed by Pennetta et al. LDs seem to be important for the progression of motor neuron diseases (MND) (119). The mechanisms by which LDs are affecting neurodegeneration processes are unknown, however it has been observed that they can affect protein aggregation. For example, in cells treated with an HFD, α-syn accumulated in triglyceride-riche LDs (120). In a similar way, amyloidal fibrils from amyloidotic polyneuropathy (FAP), were observed to be colocalizing with a high density lipoprotein (HDL) (121). Altogether, these data support the idea that LDs could be involved with the aggregation process of neurodegenerative disease associated proteins.

On the other hand, there is increasing evidence which now supports the idea that a high carbohydrate diet (HCD) is a risk factor for certain neurodegenerative diseases (122). Probably, because of a dysregulation in protein glycation, which can control the balance between protein solubility and aggregation (123, 124). Moreover, many reports indicate that there is a strong association between diabetes and people affected by AD (125, 126), suggesting that a dysregulation in glucose metabolism could impact on the onset of this and other neurodegenerative diseases (127). Although the mechanisms are not clear, evidence has shown that there is an increase in the activity of glycolytic enzymes in AD patients (128). In addition, α-syn was observed to bind GAPDH, increasing the activity of this enzyme (129). Also, analyses performed with postmortem human brain tissue, reported a reduction of glucose-6-phosphate dehydrogenase and 6-phosphogluconate dehydrogenase in AD and PD (127).

Carbohydrate metabolism dysregulation has also been found in different models of HD. Gouarné et al. found that a rat model of HD had a deficit in glycolysis in striatal neurons (130). In a mouse model of HD, it was observed that glycolysis inhibition decreases the levels of glutamate transport and provoked neurotoxicity (131). On the contrary, in two different models of HD, it was found that the activity of some glycolytic enzymes was higher. The latter results were not correlated with the enhanced conversion of glucose to lactate and increased ATP in the brain and tissue, respectively. Altogether, this evidence indicates that in neurodegenerative diseases, an imbalance in a glycolytic pathway could play an important role in the onset or progression of the disease. Considering that CMA regulates the homeostasis of glycolytic enzymes, changes in the activity of this pathway along with aging could strongly impact on the risk to develop a neurodegenerative disease. In addition, the risk can also be augmented by aging and by a continuous intake of a high carbohydrate diet. **Figure 2** shows the main mechanisms that can influence CMA down-regulation and its impact in neurodegenerative diseases.

## CONCLUDING REMARKS

As part of the autophagic pathways, CMA has been involved in regulating the metabolism of different proteins. Many of these proteins are strongly associated with neurodegenerative diseases affecting humans, while others are involved in the regulation of metabolic pathways such as lipids and carbohydrate metabolism. In addition, neurodegenerative diseases and metabolic disorders are common features of aged humans and many reports associate the dysregulation in lipids or carbohydrate metabolism, with the risk of developing some type of neurodegenerative disease. It has also been reported that a high fat, or high carbohydrate diet (or glucose reach diet), can also increase neurodegeneration prevalence or progression. On the other hand, CMA activity, has been observed to decline with age and with a non-balanced dietary intake such as a high fat diet. Overall, CMA can be considered a common factor in the regulation of metabolic and neurodegenerative pathways and a dysregulation in CMA, provoked by normal aging, or by metabolic disorders induced by deficient nutritional habits, which could tilt the balance toward a pathological situation. **Figure 3** shows an overview of how CMA could be connected with metabolic dysfunctions and neurodegenerative disorders.

#### AUTHOR CONTRIBUTIONS

AA and MB wrote the article. IEA wrote the article and created the figures. AM, JM, KC, and AC made important contributions, participated in discussions and provided corrections to the manuscript.

#### REFERENCES


#### FUNDING

This work was supported by the Programa de Apoyo a Centros con Financiamiento Basal AFB 170004 to Fundación Ciencia & Vida. By CONICYT Fondo Nacional de Desarrollo Científico y Tecnológico de Chile, FONDECYT N◦ 11161056 to IEA, FONDECYT N◦ 1161123 to MB, FONDECYT N◦ 1171075 to AC. By FONDAP-ACCDis N◦ 15130011 to AC. By the PEW Latin American Fellows Program in the Biomedical Science to AC. By the International Centre for Genetic Engineering and Biotechnology, ICGEB, N◦ CRP/CH113-04RG to MB. This work was also funded by the CONICYT PIA ACT172066 to MB and AC.


complexes at the lysosomal membrane. Mol Cell Biol. (2008) 28:5747–63. doi: 10.1128/MCB.02070-07


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Alfaro, Albornoz, Molina, Moreno, Cordero, Criollo and Budini. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Chaperone-Mediated Autophagy Upregulation Rescues Megalin Expression and Localization in Cystinotic Proximal Tubule Cells

Jinzhong Zhang<sup>1</sup> , Jing He<sup>1</sup> , Jennifer L. Johnson<sup>1</sup> , Farhana Rahman<sup>1</sup> , Evripidis Gavathiotis <sup>2</sup> , Ana Maria Cuervo<sup>3</sup> and Sergio D. Catz <sup>1</sup> \*

<sup>1</sup> Department of Molecular Medicine, The Scripps Research Institute, La Jolla, CA, United States, <sup>2</sup> Department of Biochemistry, Albert Einstein College of Medicine, Bronx, NY, United States, <sup>3</sup> Department of Developmental and Molecular Biology, Albert Einstein College of Medicine, Bronx, NY, United States

#### Edited by:

Maria Ines Vaccaro, Universidad de Buenos Aires, Argentina

#### Reviewed by:

James Harper, Sam Houston State University, United States Mina Konigsberg, Universidad Autónoma Metropolitana, Mexico

> \*Correspondence: Sergio D. Catz scatz@scripps.edu

#### Specialty section:

This article was submitted to Endocrinology of Aging, a section of the journal Frontiers in Endocrinology

Received: 10 October 2018 Accepted: 14 January 2019 Published: 01 February 2019

#### Citation:

Zhang J, He J, Johnson JL, Rahman F, Gavathiotis E, Cuervo AM and Catz SD (2019) Chaperone-Mediated Autophagy Upregulation Rescues Megalin Expression and Localization in Cystinotic Proximal Tubule Cells. Front. Endocrinol. 10:21. doi: 10.3389/fendo.2019.00021 Cystinosis is a lysosomal storage disorder caused by defects in CTNS, the gene that encodes the lysosomal cystine transporter cystinosin. Patients with nephropathic cystinosis are characterized by endocrine defects, defective proximal tubule cell (PTC) function, the development of Fanconi syndrome and, eventually, end-stage renal disease. Kidney disease is developed despite the use of cysteamine, a drug that decreases lysosomal cystine overload but fails to correct overload-independent defects. Chaperone-mediated autophagy (CMA), a selective form of autophagy, is defective in cystinotic mouse fibroblasts, and treatment with cysteamine is unable to correct CMA defects in vivo, but whether the vesicular trafficking mechanisms that lead to defective CMA in cystinosis are manifested in human PTCs is not currently known and whether PTC-specific mechanisms are corrected upon CMA upregulation remains to be elucidated. Here, using CRISPR-Cas9 technology, we develop a new human PTC line with defective cystinosin expression (CTNS-KO PTCs). We show that the expression and localization of the CMA receptor, LAMP2A, is defective in CTNS-KO PTCs. The expression of the lipidated form of LC3B, a marker for another form of autophagy (macroautophagy), is decreased in CTNS-KO PTCs indicating decreased autophagosome numbers under basal conditions. However, the autophagic flux is functional, as measured by induction by starvation or by blockage using the v-ATPase inhibitor bafilomycin A, and by degradation of the macroautophagy substrate SQSTM1 under starvation and proteasome-inhibited conditions. Previous studies showed that LAMP2A accumulates in Rab11-positive vesicles in cystinotic cells. Here, we show defective Rab11 expression, localization and trafficking in CTNS-KO PTCs as determined by confocal microscopy, immunoblotting and TIRFM. We also show that both Rab11 expression and trafficking in cystinotic PTCs are rescued by the upregulation of CMA using small-molecule CMA activators. Cystinotic PTCs are characterized by PTC de-differentiation accompanied by loss of the endocytic receptor megalin, and megalin recycling is regulated by Rab11. Here we show that megalin plasma membrane localization is defective in CTNS-KO PTCs and its expression is rescued by treatment with CMA activators. Altogether, our data support that CMA upregulation has the potential to improve PTC function in cystinosis.

Keywords: lysosomal storage disorder (LSD), fanconi syndrome, megalin, chaperone-mediated autophagy (CMA), cystinosis, vesicular trafficking, Rab 11 GTPase, LAMP2A

#### INTRODUCTION

Cystinosis is a lysosomal storage disorder (LSD) caused by genetic defects in CTNS (Ctns in mouse), the gene that codes for the cystine transporter cystinosin. Increased levels of intralysosomal cystine (1) lead to cell malfunction in cystinosis. Tissue deterioration manifests in kidneys and eyes but also affects other organs including the liver, brain, and muscle (2). Kidney proximal tubule cells (PTCs) are the first cell type to be affected in nephropathic cystinosis, causing, in the long term, end-stage kidney disease. Patients with severe cystinosis require kidney transplants. Endocrine disorders are also common in cystinosis such as hypothyroidism, growth retardation, and hypogonadism (3). Hypothyroidism is the most frequently reported endocrine manifestation of the disease (4). Altered thyroglobulin biosynthesis associated with endoplasmic reticulum stress is the cause of this manifestation. Cystinotic patients also suffer from insulin-dependent diabetes (5), which contributes other complications including muscle (6) and bone (7) alterations that are pathognomonic of the disease.

The current treatment for patients with cystinosis is cysteamine which reduces intra-lysosomal cystine, conjugates, and transports cysteine out of the lysosome through the exporter PQLC2 (8). Despite the efficiency of cysteamine in retarding the rate of renal deterioration and improving linear growth in children with cystinosis (9), cell malfunction, tissue failure, progressive renal disease, endocrine complications, and muscle abnormalities still occur (10), suggesting that cystine accumulation is not the only cause for all the defects observed in cystinosis (10, 11). Thus, to improve treatment of this LSD, it is crucial to understand the defective molecular mechanisms that lead to the various tissues dysfunction and injury. In order to understand these mechanisms, it is essential to develop and characterize models of the disease. To this end, the establishment of new cellular models of cystinotic proximal tubule cells, with defined genotypic and phenotypic characteristics, is essential to study disease-relevant mechanisms, to develop knowledge and to implement novel strategies for treating renal disease progression in this devastating disease.

Chaperone mediated autophagy (CMA) is a selective form of autophagy that contributes to proteostasis in several physiological and pathological conditions (12). CMA consists of the internalization of selected cytosolic substrates into the lysosome by a mechanism that includes: Recognition of a pentapeptide-like KFERQ in the substrate by the chaperone hsc70; substrate presentation by the chaperone to the receptor LAMP2A; receptor multimerization and protein internalization for degradation in the lysosome, assisted by a lumenal form of hsc70 (13). LAMP2A the only known lysosomal receptor for CMA, shows defective localization and impaired function in cystinosis (14, 15). Defects in CMA in cystinosis lead to the cytosolic accumulation of CMA substrates and are proposed to contribute to the pathological processes of the disease that are cysteamine treatment-insensitive (14). However, the specific CMA mechanism(s) that are defective in cystinotic proximal tubule cells are currently unknown and the impact of CMA upregulation on PTC function requires further analysis. Under oxidative stress CMA is activated. This activation correlates with increased expression levels of the lysosomal lumenal chaperone protein hsc70 (required for substrate uptake), and also correlates with a selective increase of the expression of the CMA receptor LAMP2A at the lysosomal membrane, leading to higher rates of CMA (16). However, despite the observations that cystinosis is associated with increased oxidative stress and that cystinotic patients have high serum levels of oxidative stress markers (11), cystinotic cells are actually susceptible to oxidative stress, most likely caused by downregulation of CMA. Remarkably, CMA induction by pharmacological enhancers protects cystinotic cells from the increased susceptibility to oxidative stress and reconstitutes the resistant levels observed in wild type cells, an effect dependent on LAMP2A expression and its lysosomal membrane localization (15). It then becomes clear that the correct lysosomal localization of LAMP2A is necessary to maintain cellular homeostasis in cystinosis. However, the mechanisms that mediate lysosomal localization of LAMP2A are not well-understood and the possible consequences of downregulated CMA in cystinotic PTCs is unknown.

In cystinosis, cystine accumulation induces apical PTC dedifferentiation (17). PTCs, which play a central role in maintaining homeostasis by mediating reabsorption of electrolytes and nutrients in the renal tube, rely on specialized apical receptors that control the internalization of specific substrates. In particular, megalin (gp330, LRP-2), a member of the low-density lipoprotein receptor family, is expressed in proximal tubule epithelial cells, and together with cubilin, mediates the endocytosis of an extensive number of diverse ligands including lipoproteins, vitamin-binding proteins, hormones, and enzymes (18). Based on their function as nonspecific protein reabsorption molecules, megalin and cubilin are considered to operate as scavenger receptors (19). Megalin's apical localization in PTCs is mediated by fast-recycling mechanisms from apical recycling endosomes, a mechanism regulated by the small GTPase Rab11 (18). Fast megalin recycling allows for its reutilization and delays its degradation (18). A possible role for CMA in the regulation of megalin expression and localization is currently unknown. However, the role of CMA in the control of trafficking mechanisms that are affected in cystinotic fibroblasts, as previously described by our group (14), suggests that autophagy and, in particular CMA, may be an important determinant in the control of megalin functions through the regulation of vesicular transport.

In this work, we developed and characterized human cystinotic proximal tubule cells (CTNS-KO PTCs) and demonstrated that these cells are characterized by CMA defects that affect vesicular trafficking mechanisms regulating megalin localization.

## MATERIALS AND METHODS

## Generation of Cystinosin Knockout Human Proximal Tubule Cell Line Using CRISPR/Cas9

To generate CTNS knockout human proximal tubule cells, Cas9 nickase was used along with a pair of gRNAs targeting exon 4 of the human CTNS gene. gRNAs were designed using the online tool (http://crispr.mit.edu/). Each gRNA was sub-cloned into pSpCas9n(BB)-2A-GFP(PX461) (Addgene), two pSpCas9n(BB)- 2A-GFP(PX461)-CTNS gRNA constructs were then transfected into a model of human proximal tubule cell (HK-2 cells) at a 1:1 ratio using Lipofectamine LTX (Thermo Fisher Scientific). After 48 h, GFP-positive cells were sorted by FACS, seeded into plates and grown into single colonies. Single colonies were then expanded, the junction sequence was amplified by PCR, subcloned into pBluescript and sequenced to identify the variety of insertions and deletions (indels) at targeting sites. As a result, we obtained different indels in two alleles. One of the modifications has a 37 bp insertion, which abolishes the splicing site of the exon 4 at its 3′ end. The other allele has a 17 bp deletion, causing an amino acid shift, which led to a 52-amino acid product.

## Constructs and Transfections

The GFP-Rab11 construct was obtained from Addgene. Cells were transfected with this construct using Lipofectamine <sup>R</sup> 2000 (Thermo Fisher Scientific) following the manufacturer's instructions.

## Gel Electrophoresis, Immunoblotting, and Antibodies

Cells were lysed in RIPA lysis buffer in the presence of protease-inhibitors (Roche). Following electrophoresis using NuPAGE 4–12% gels (Thermo Fisher Scientific), proteins were transferred onto 0.45µm nitrocellulose membranes and the membranes were incubated overnight at 4◦C with the indicated primary antibodies, followed by incubation with HRPconjugated secondary antibodies. The following antibodies were used in this study: mouse anti-LAMP1 (Santa Cruz, sc-20011), rabbit anti-LAMP2A (Abcam, ab18528), rabbit anti-actin (Sigma, A2066), mouse anti-Rab11 (Thermo Fisher Scientific, MA1- 24919), rabbit anti-SQSTM1/p62 (Cell Signaling, 5114), rabbit anti-LC3B (Cell Signaling, 2775), and goat anti-Megalin (Santa Cruz, sc-16476).

## Immunofluorescence and Confocal Microscopy Analysis

Cells were seeded on a 4-chamber 35 mm-glass bottom dish (in vitro Scientific). Where indicated, cells were treated with the indicated treatments, then fixed with 3.7% paraformaldehyde for 15 min and blocked with 1% BSA in PBS for 1 h. Samples were labeled with the indicated primary antibodies overnight at 4◦C in the presence of 0.01% saponin and 1% BSA. Samples were washed 3 times and subsequently incubated with the appropriate combinations of Alexa Fluor (488 or 594)-conjugated anti-rabbit, anti-rat, or anti-mouse secondary antibodies (Thermo Fisher scientific). Samples were analyzed with a Zeiss LSM 710 laser scanning confocal microscope (LSCM) attached to a Zeiss Observer Z1 microscope, using a 63 × oil Plan Apo, 1.4 NA objective. Images were collected using ZEN-LSM software keeping the laser power and gain constant during all acquisitions for comparative analysis of wild type and knockout cells or treated with vehicle or various treatments. Images were then processed using ImageJ (National Institutes of Health, Bethesda, MD) and Photoshop CS4 (Adobe). The fluorescence intensities were quantified using the ImageJ software.

## Total Internal Reflection Fluorescence (TIRF) Microscopy

Pseudo-TIRF microscopy analyses were performed using a 100X 1.45 numerical aperture TIRF objective (Nikon) on a Nikon TE2000U microscope custom modified with a TIRF illumination module as described (20). Images were acquired on a 16-bit, cooled charge-coupled device camera (Hamamatsu) controlled through NIS-Elements software. For live experiments, the images were recorded using 300–500 ms exposures depending on the fluorescence intensity of the sample.

## Cystine Accumulation

Cystine levels were measured by mass spectrometry at the Biochemical Genetics Laboratory, University of California, San Diego as described previously (20).

## Starvation Protocols

For studies of macroautophagy, cells were briefly washed in serum-free DMEM (containing 1x amino acids), media was aspirated, and fresh serum-free DMEM was added followed by 5 h incubation at 37◦C, in the presence or absence of lysosomal inhibitor Bafilomycin A (LC laboratories, 100 nM) or proteasome inhibitor Clasto-Lactacystin β-lactone (Cayman Chemical, 1 uM), as described previously (14).

## PCR and RT-PCR Analysis

For RT-PCR analysis, RNA was isolated from wild-type or Ctns−/<sup>−</sup> mouse fibroblasts using the RNeasy mini-kit for RNA purification (Qiagen). A total of 100 ng of RNA for each cell line was reverse-transcribed (RT) using iScript cDNA synthesis kit (Bio-Rad). PCR analysis was then performed using Taq DNA Polymerase(Thermo Fisher Scientific, 18038042), with the following primer mixes: human CTNS forward, TCCTCCTGTCGTAAAGCTGGA, and human CTNS reverse, GCCGGTCTGATTGGAGTGAT.

#### Statistical Analysis

Data are presented as mean, and error bars correspond to standard errors of the means (SEMs) unless otherwise indicated. Statistical significance was determined using the pair or unpaired Student's t-test or one-way ANOVA, Tukey's multiple comparisons test for multiple groups using either Excel software or GraphPad Prism (version 4) software.

## RESULTS

## Generation and Characterization of Cystinotic Human Proximal Tubule Cells (PTCs)

Proximal tubule cells (PTCs) are the most affected cells in cystinosis and are characterized by the accumulation of cystine crystals, as well as by PTC de-differentiation accompanied by loss of the endocytic receptors megalin and cubilin (17). In cystinotic fibroblasts, the cystinosin deficiency leads to defects in autophagic pathways caused by defective intracellular trafficking of the CMA receptor LAMP2A (14). In cystinotic kidneys, the accumulation of substrates suggests defective CMA but whether LAMP2A function is defective in CTNS-KO PTCs and whether CMA upregulation improves PTC function are both currently unknown. To study the mechanisms of vesicular trafficking and CMA regulation in CTNS-KO PTCs, we developed human PTCs deficient in cystinosin using CRISPR-Cas9 technology. To this end, we used the human PTC cell line HK-2, which has been largely demonstrated to be an authentic PTC model (21). The strategy to design and generate CTNS-KO PTCs is described in **Figure 1**. Two gRNAs targeting Human CTNS exon 4 were designed and are shown in **Figure 1A**. The gRNA1 targeted one of the alleles, generated a 37 bp insertion, destroyed the splicing site, and was predicted to generate a 149aa product. The gRNA2, targeted the second allele to generate a 17 bp deletion, causing a shift leading to a theoretical 52aa product (**Figure 1A**). Next, the CTNS-KO PTCs were analyzed for the expression of CTNS mRNA by PCR (**Figure 1B**). Using a 5′ primer that sits just upstream of the mutation site and a downstream reverse primer, we show that wild type CTNS mRNA is not expressed in CTNStargeted (CTNS-KO) cells. Although the mutations deplete expression of CTNS, aberrant CTNS mRNA could still be detected, at least partially (not shown). Possible off-target effects of CRISPR/Cas9 were also analyzed. Thus, three potential offtarget loci were amplified by PCR followed with Surveyor assay and Sanger sequencing. Our data show that these three

FIGURE 2 | Mislocalization and decreased expression of LAMP2A in CTNS-KO PTCs. (A,B) Confocal microscopy analysis and quantification of the distribution of endogenous LAMP1 and LAMP2A in WT and CTNS-KO PTCs. Scale bar, 10µm. mean ± SEM. \*\*\*p < 0.001, Student's t-test. (C) Western blot analysis of LAMP1, LAMP2A, and LC3B expression levels in WT and CTNS-KO PTCs. Left panel, Representative immunoblots. Ponceaus S staining is shown for equal loading. Middle and right panels, Quantification of 3 independent experiments. The individual symbols correspond to independent biological replicates from 3 independent experiments. The bars represent the mean ± SEM. \*p < 0.05; \*\*p < 0.01, Student's t-test. (D) SQSTM1/p62 and LC3B-II levels in WT and CTNS-KO PTCs were analyzed by Western blot under fed and serum starvation conditions, in the absence or presence of 100 nM bafilomycin A (BafA) or 1µM Clasto-Lactacystin β-lactone (proteasome inhibitor) for 5 h. Left panel, Representative immunoblots. Middle and right panel, Quantification of the expression levels of SQSTM1 (p62) and LC3B-II, respectively. Each value has been normalized to the actin expression level in the same sample. Results are expressed relatively to the wild type fed condition. The individual symbols correspond to independent biological replicates from 3 independent experiments. The bars represent the mean ± SEM. \*p < 0.05.

loci are intact (**Figure 1C**). For confirmation of cystinosin activity deficiency, we measured cystine accumulation using mass spectrometry analysis of cystine content. In **Figure 1D**, we show that cystine levels are barely detectable in wild type cells but cystine is significantly increased in CTNS-KO cells. Thus, CTNS-KO PTCs accumulate cystine at levels that are similar to those observed in cystinotic cells from the Ctns−/<sup>−</sup> mouse (14).

## Decreased Expression and Mislocalization of LAMP2A in CTNS-KO PTCs

To study possible CMA defective mechanisms in CTNS-KO PTCs, and because previous studies from our laboratory suggested that the CMA receptor LAMP2A is mislocalized in cystinosis, we first analyzed the subcellular distribution of LAMP2A in wild type and CTNS-KO PTCs. In **Figures 2A,B**, we show that while LAMP2A shows high colocalization with LAMP1 in wild type cells, CTNS-KO PTCs have a subpopulation of LAMP2A-positive vesicles that does not overlap with the distribution of LAMP1. This is in agreement with previous studies from our laboratory showing that intermediate vesicles containing LAMP2A are mislocalized and have defective trafficking in cystinotic fibroblasts (15). Further, to better understand whether the mislocalization of LAMP2A affects protein stability, we determined the levels of expression of LAMP2A in CTNS-KO PTCs. Interestingly, LAMP2A expression was decreased in CTNS-KO PTCs, suggesting either downregulation at the transcriptional level or increased degradation (**Figure 2C**). LAMP2A-decreased expression was specific since LAMP1 expression is elevated in CTNS-KO PTCs (**Figure 2C**). These data recapitulate the defective phenotypes observed in Ctns−/<sup>−</sup> fibroblasts (14, 20) and thus confirmed that CTNS-KO PTCs constitute a valid cystinotic model.

Next, to determine whether changes in CMA are accompanied by defects in macroautophagy, we analyzed the expression levels of the lipidated form of LC3B (LC3B-II) by Western blot. In **Figure 2C**, we show that LC3B-II expression is decreased in CTNS-KO PTCs.

The autophagic flux in CTNS-KO PTCs was analyzed using well-established biochemical methods. Wild type and CTNS-KO PTCs were exposed to fed, serum starvation and autophagyblocking conditions, using bafilomycin to inhibit the fusion of autophagosomes with lysosomes. In **Figure 2D**, we show that despite having decreased number of autophagosomes, CTNS-KO PTCs respond to starvation by degrading LC3B, indicating that macroautophagy is inducible by starvation in these cells. This was also confirmed by the decrease in protein levels of the autophagy substrate SQSTM1/p62 in these cells after starvation. Treatment with the proteasome inhibitor β-lactone did not affect protein degradation (**Figure 2D**), indicating that SQSTM1 is degraded mainly through macroautophagy in these cells. Increased LC3B levels after treatment with bafilomycin confirmed that autophagosome formation is functional in CTNS-KO PTCs and that the reduced LC3B levels together with reduced SQSTM1 levels are indicative of accelerated clearance rather than reduced autophagosome

biogenesis. This is in agreement with previous studies showing increased autophagic flux in cystinotic fibroblasts, albeit with accumulation of autophagosomes (14), and goes hand-in-hand with previous studies showing upregulation of macroautophagy as a compensatory mechanism in response to reduced CMA. This suggests that CTNS-KO PTCs have a dysregulated macroautophagy phenotype, similar to other cystinotic cellular systems.

We have previously shown that LAMP2A localizes at Rab11 positive vesicles in cystinotic cells (14), suggesting that Rab11 positive carrier vesicles may be responsible for trafficking of LAMP2A to the lysosome. Here, we next analyzed the distribution of endogenous Rab11 in CTNS-KO PTCs and found that Rab11 is mislocalized in these cells. In particular, we observed that the distribution of Rab11-positive vesicles is defective in CTNS-KO cells, showing a homogeneous distribution throughout the cystinotic cells as opposed to the more perinuclear distribution of Rab11 puncta observed in wild type cells (**Figure 3A**). This suggests defective trafficking of Rab11 vesicles in CTNS-KO cells. Next, we analyzed the expression levels of endogenous Rab11 and found that Rab11 is significantly downregulated in CTNS-KO PTCs (**Figure 3B**). Because LAMP2A mislocalization suggests defective CMA and as CMA upregulation may have a direct effect on the

expression levels of trafficking associated proteins (15), we next studied the effect of CMA enhancers on Rab11 expression in CTNS-KO cells. To this end, we treated wild type and CTNS-KO PTCs with the CMA enhancer QX77, a small molecule known to dysregulate the repressive function of retinoic acid receptor-alpha over the CMA machinery (15, 22). We found that treatment with this CMA enhancer significantly increased Rab11 expression in CTNS-KO PTCs, bringing the expression of this small GTPase to wild type levels (**Figure 3B**).

Next, to determine whether the mislocalization of Rab11 was caused by defective trafficking, we transfected wild type and CTNS-KO cells with the recycling endosome trafficking reporter, GFP-Rab11 (**Figure 4A**), and measured vesicular transport using Total Internal Reflection Fluorescence microscopic analysis. In **Figure 4B**, we show that cystinotic PTCs have defective Rab11 trafficking characterized by significant decrease in Rab11-positive vesicle movement. Thus, CTNS-KO cells showed increased numbers of vesicles moving at very low speed or not moving at all (**Figure 4B**), with a concomitant decrease of faster-moving Rab11-positive vesicles (**Figure 4C**). Interestingly, treatment with the CMA enhancer QX77 rescued the defective trafficking phenotype in CTNS-KO PTCs thus favoring a net and significant increment in Rab11 transport (**Figures 4B,C** and **Movies 1**–**3**).

## Upregulation of Chaperone-Mediated Autophagy Increases Megalin Localization at the Plasma Membrane in Cystinotic PTCs

Megalin, an endocytosis receptor located apically in PTCs and other epithelial cells including endocrine glands, the lung and the brain, plays a fundamental role in PTC function by facilitating the uptake of plasma solutes for ultrafiltration. Cystinotic PTCs have progressive loss of megalin expression and other apical receptors (17), and megalin-KO mice are characterized by urinary loss of ultra-filtrated plasma proteins (23). The apical distribution of megalin in PTCs is proposed to be mediated by the trafficking of apical recycling endosomes, regulated by the small GTPase Rab11 (18). Because Rab11 trafficking is defective in CTNS-KO PTCs, we next investigated the expression and localization of endogenous megalin in CTNS-KO PTCs. In **Figure 5A**, we show that CTNS-KO PTCs are characterized by decreased plasma membrane localization of megalin. Further, because treatment with CMA enhancers has a direct impact on Rab11 trafficking in cystinotic cells, we studied whether the upregulation of CMA has a beneficial effect on megalin localization. To this end, we treated CTNS-KO PTCs with the CMA activator QX77 (20µM) for 72 h and determined

the plasma membrane distribution of megalin in these cells. In **Figure 5A**, we show that CMA upregulation significantly increases megalin localization at the plasma membrane in CTNS-KO PTCs, indicating that upregulation of CMA has a direct positive implication on megalin distribution, likely affecting PTC function. Finally, because recycling through the plasma membrane is known to stabilize megalin and prevent its degradation (18), we next analyzed the effect of CMA enhancers on megalin expression. In **Figures 5B,C**, we show that megalin expression is decreased in CTNS-KO PTCs. We also show that treatment with the CMA enhancer, QX77, induces the significant upregulation of megalin expression in cystinotic PTCs.

## DISCUSSION

Proximal tubule cell dedifferentiation is a hallmark of nephropathic cystinosis (17). One of the most important changes underwent by PTCs in cystinosis is the loss of megalin apical distribution which is accompanied by a net increase of protein and electrolyte loss (17). This progressive loss of the expression of megalin and other endocytic receptors, including the glucose and sodium transporter SGC2 and phosphate transporters, is proposed to account for apical dedifferentiation leading to the development of Fanconi syndrome, which is not corrected by cysteamine treatment in cystinotic patients. Megalin is a fast recycling receptor and has been proposed to define the apical recycling pathway of epithelial cells. Its localization and recycling at the apical membrane depends on the function of Rab11 (18). In this work, using newly generated cystinotic PTCs, we show that Rab11 expression and trafficking is defective in CTNS-KO cells. We also demonstrate that both the impaired Rab11 trafficking and the defective localization of the CMA receptor LAMP2A at lysosomes are corrected by treatment with CMA enhancers, thus linking CMA defects to PTC disfunction and de-differentiation in cystinosis.

Rab GTPases and their effectors are determinants of membrane identity and master regulators of vesicular function. Decreased expression or defective functions of these small GTPases are associated with specific defects in vesicular transport. Cystinotic cells are characterized by defective endolysosomal transport, while the induction of vesicular trafficking mechanisms through the upregulation of Rab27a or Rab7 improves vesicular trafficking in cystinosis (20). Furthermore, a Rab11-specific defect in cystinotic cells was recently described (15). In particular, the CMA receptor LAMP2A shows an anomalous distribution in mouse cystinotic fibroblasts, where it localizes to Rab11-positive carrier vesicles instead of reaching the lysosomes (14). LAMP2A mislocalization was directly associated with defects in CMA, observed both in isolated cystinotic fibroblasts and, in vivo, in cystinotic mice (14, 15). Constitutively active Rab11 partially rescues the localization of LAMP2A at the lysosomal membrane (15). It is current knowledge that not all defects in cystinosis are corrected by cysteamine, that Fanconi syndrome is developed before the apparition of cystine crystals and independently of early cysteamine treatment, suggesting that CMA defects occur independently of lysosomal overload. Because the defects in CMA function are not repaired by treatment with cysteamine, the only lysosomal depletion therapy currently available for cystinotic patients, CMA function acquires important clinical significance. In this context, the observations that CMA upregulation, through treatment of proximal tubule cells with novel CMA enhancers, upregulates Rab11 expression, corrects Rab11 trafficking and corrects LAMP2A localization in PTCs is highly significant. Of note, although LAMP2A is cleaved by lysosomal cathepsin A (24) and lysosomal protease inhibitors increase the amount of lysosomal membrane-localized LAMP2A in cystinosis (14), it is not yet clear whether this mechanism of LAMP2A upregulation may have direct implications in proximal tubule cell function.

How defective CMA affects PTC function in cystinosis is currently unknown but since dysfunction of CMA has been implicated in PTC hypertrophy (25), it is expected that CMA upregulation would be beneficial in cystinotic patients. Kidney proximal epithelial cells are characterized by low levels of macroautophagy, but high basal levels of CMA activity. In PTCs, CMA has been shown to regulate the protein levels of important transcription factors that control PTC differentiation, including Pax2 (25). In mouse cystinotic PTCs in vivo, the accumulation of CMA substrates (including GAPDH) is caused by mislocalization of LAMP2A (15), suggesting that CMA may directly affect PTC function in this lysosomal disorder and that LAMP2A trafficking may be affected in these cells. Despite this knowledge, a direct impact of CMA deficiency on PTC function in cystinosis has not been demonstrated thus far and the mechanisms underlying PTC defects caused by defective CMA in cystinosis remain unknown. Here, the observation that CMA enhancers upregulate Rab11 trafficking in cystinotic PTCs is highly significant, not only because Rab11 is able to re-localize LAMP2A at the lysosomal membrane (15) but also because Rab11 upregulation has direct consequences on plasma membrane expression of the endocytic receptor megalin, which as mentioned above, is essential for

#### REFERENCES

1. Town M, Jean G, Cherqui S, Attard M, Forestier L, Whitmore SA, et al. A novel gene encoding an integral membrane protein is substrate resorption from the ultrafiltrate in the proximal tubule. Thus, it is expected that CMA enhancers would have a direct impact on cystinotic PTC function and potentially on the reduction of some of the symptoms associated with Fanconi syndrome. Whether CMA enhancers increase upregulation of other apical receptors in cystinosis is currently unknown, but because of their effect on Rab11 trafficking and the function of Rab11 as a recycling endosome regulator, it is likely that the effects of CMA enhancers go beyond the benefits mediated by megalin upregulation.

In conclusion, CMA upregulation in cystinotic PTCs has a direct positive impact on Rab11 trafficking and megalin expression (**Figure 6**), whose reduced expression in cystinosis is associated with Fanconi syndrome. Thus, CMA constitutes a potential new therapeutic target, independent of lysosomal overload-reducing therapies such as cysteamine.

## AUTHOR CONTRIBUTIONS

JZ and JH: performed experiments and organized data; JJ: supervised experiments, analyzed data, and edited manuscript; FR: performed experiments; EG and AC: contributed important reagents and edited manuscript; SC: conceived the idea, analyzed data, and wrote the manuscript.

## ACKNOWLEDGMENTS

This work was supported by U.S. Public Health Service grants R01 HL088256, R01 AR070837, R21 EY028642, and R01 DK110162 to SC. JZ and FR are fellows of the Cystinosis Research Foundation.

### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fendo. 2019.00021/full#supplementary-material

Movie 1 | Pseudo-TIRF microscopy analysis of GFP-Rab11 in WT HK-2 cells. Cells were transfected with GFP-Rab11 plasmid. After 48 h, cells were analyzed by TIRFM. The video was recorded for 60 s, 61 frames and 200 ms exposure. Playback is 7 fps.

Movie 2 | Pseudo-TIRF microscopy analysis of GFP-Rab11 in CTNS-KO HK-2 cells. Cells were transfected with GFP-Rab11 plasmid. After 48 h, cells were analyzed by TIRFM. The video was recorded for 60 s, 61 frames and 200 ms exposure. Playback is 7 fps.

Movie 3 | Pseudo-TIRF microscopy analysis of GFP-Rab11 in CTNS-KO HK-2 cells treated with QX77. Cells were transfected with GFP-Rab11 plasmid and incubated with QX77. After 48 h, cells were analyzed by TIRFM. The video was recorded for 60 s, 61 frames and 200 ms exposure. Playback is 7 fps.

mutated in nephropathic cystinosis. Nat Genet. (1998) 18:319–324. doi: 10.1038/ng0498-319

2. Gahl WA, Thoene JG, Schneider JA. Cystinosis. N Eng J Med. (2002) 347:111– 21. doi: 10.1056/NEJMra020552


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Zhang, He, Johnson, Rahman, Gavathiotis, Cuervo and Catz. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Modulation of Autophagy Influences the Function and Survival of Human Pancreatic Beta Cells Under Endoplasmic Reticulum Stress Conditions and in Type 2 Diabetes

M. Bugliani <sup>1</sup> , S. Mossuto<sup>1</sup> , F. Grano<sup>1</sup> , M. Suleiman<sup>1</sup> , L. Marselli <sup>1</sup> , U. Boggi <sup>2</sup> , P. De Simone<sup>3</sup> , D. L. Eizirik <sup>4</sup> , M. Cnop<sup>4</sup> , P. Marchetti 1† and V. De Tata<sup>3</sup> \* †

Université Libre de Bruxelles, Brussels, Belgium

<sup>1</sup> Department of Clinical and Experimental Medicine, University of Pisa, Pisa, Italy, <sup>2</sup> Department of Surgical Pathology, Medicine, Molecular and Critical Area, University of Pisa, Pisa, Italy, <sup>3</sup> Department of Translational Research and New Technologies in Medicine and Surgery, University of Pisa, Pisa, Italy, <sup>4</sup> Medical Faculty, ULB Center for Diabetes Research,

#### Edited by:

Antonello Lorenzini, University of Bologna, Italy

#### Reviewed by:

James Harper, Sam Houston State University, United States Chad Ashyr Lerner, Buck Institute for Research on Aging, United States

\*Correspondence:

V. De Tata vincenzo.detata@med.unipi.it

†These authors have contributed equally to this work

#### Specialty section:

This article was submitted to Endocrinology of Aging, a section of the journal Frontiers in Endocrinology

Received: 06 July 2018 Accepted: 21 January 2019 Published: 26 February 2019

#### Citation:

Bugliani M, Mossuto S, Grano F, Suleiman M, Marselli L, Boggi U, De Simone P, Eizirik DL, Cnop M, Marchetti P and De Tata V (2019) Modulation of Autophagy Influences the Function and Survival of Human Pancreatic Beta Cells Under Endoplasmic Reticulum Stress Conditions and in Type 2 Diabetes. Front. Endocrinol. 10:52. doi: 10.3389/fendo.2019.00052 Autophagy is the major mechanism involved in degradation and recycling of intracellular components, and its alterations have been proposed to cause beta cell dysfunction. In this study, we explored the effects of autophagy modulation in human islets under conditions associated to endoplasmic reticulum (ER) stress. Human pancreatic islets were isolated by enzymatic digestion and density gradient purification from pancreatic samples of non-diabetic (ND; n = 17; age 65 ± 21 years; gender: 5 M/12 F; BMI 23.4 ± 3.3 kg/m<sup>2</sup> ) and T2D (n = 9; age 76 ± 6 years; 4 M/5 F; gender: BMI 25.4 ± 3.7 kg/m<sup>2</sup> ) organ donors. Nine ND organ donors were treated for hypertension and 1 for both hypertension and hypercholesterolemia. T2D organ donors were treated with metformin (1), oral hypoglycemic agents (2), diet + oral hypoglycemic agents (3), insulin (3) or insulin plus metformin (3) as for antidiabetic therapy and, of these, 3 were treated also for hypertension and 6 for both hypertension and hypercholesterolemia. Two days after isolation, they were cultured for 1–5 days with 10 ng/ml rapamycin (autophagy inducer), 5 mM 3-methyladenine or 1.0 nM concanamycin-A (autophagy blockers), either in the presence or not of metabolic (0.5 mM palmitate) or chemical (0.1 ng/ml brefeldin A) ER stressors. In ND islets palmitate exposure induced a 4 to 5-fold increase of beta cell apoptosis, which was significantly prevented by rapamycin and exacerbated by 3- MA. Similar results were observed with brefeldin treatment. Glucose-stimulated insulin secretion from ND islets was reduced by palmitate (−40 to 50%) and brefeldin (−60 to 70%), and rapamycin counteracted palmitate, but not brefeldin, cytotoxic actions. Both palmitate and brefeldin induced PERK, CHOP and BiP gene expression, which was partially, but significantly prevented by rapamycin. With T2D islets, rapamycin alone reduced the amount of p62, an autophagy receptor that accumulates in cells when macroautophagy is inhibited. Compared to untreated T2D cells, rapamycin-exposed diabetic islets showed improved insulin secretion, reduced proportion of beta cells showing signs of apoptosis and better preserved insulin granules, mitochondria and

**63**

ER ultrastructure; this was associated with significant reduction of PERK, CHOP and BiP gene expression. This study emphasizes the importance of autophagy modulation in human beta cell function and survival, particularly in situations of ER stress. Tuning autophagy could be a tool for beta cell protection.

Keywords: autophagy, human islets, type 2 diabetes, ER stress, insulin secretion, beta cell dysfunction

#### INTRODUCTION

Autophagy represents a highly conserved intracellular recycling pathway by which cellular components are degraded through the lysosomal machinery (3). Classically considered as a mechanism to promote cell survival during starvation (2), autophagy can also be induced by several physiological and pathological conditions, such as growth factors deprivation, hypoxia, oxidative stress, and physical exercise (4). Furthermore, autophagy seems to be constitutively activated at low levels to remove misfolded proteins and damaged and/or senescent organelles (1). Hence, autophagy can be viewed as a mechanism to protect cells against several stressors as well as a cellular response to wear-and-tear processes (5). On the other hand, it has been demonstrated that dysregulated activation of autophagy can also induce different types of cell death (6). Thus, it seems that autophagy can either protect or promote cell death in relation to the cellular and environmental context (7). Accordingly, altered autophagy could play a key pathogenic role in several disease processes, especially where the accumulation of damaged molecules and organelles might elicit a condition of increased cellular stress (8).

Pancreatic beta cells are specialized to secrete insulin in response to variations in blood glucose concentration. In order to maintain glucose homeostasis, beta cells are able to raise several fold their insulin synthesis and secretion in response to increased plasma glucose levels. Therefore, they must continually deal with a high protein burden, as proinsulin biosynthesis has been calculated to reach 10<sup>6</sup> molecules/min (9). This represents a major challenge for their ER, where protein translation and quality control takes place, and therefore beta cells are particularly susceptible to ER stress (10–12). When faced with ER stress, beta cells respond to activating the unfolded protein response (UPR) (13, 14), whose signaling is mediated via three main transmembrane sensors: IRE1α (endoribonuclease inositol requiring protein), PERK (protein kinase RNA-like endoplasmic reticulum kinase), and ATF6 (activating transcription factor 6) (10–15). In basal conditions, the chaperone immunoglobulin heavy chain binding protein (BiP), a key member of the Hsp70 family, is constitutively bound to the luminal domain of these three sensors and prevents their activation. When misfolded protein accumulates in the ER, BiP dissociates from the UPR sensors leading to their consequent activation. While a moderate ER stress-induced UPR represents a compensatory mechanism, a chronic or overwhelming ER stress impairs cellular functions and can induce apoptosis to remove irreversibly damaged cells (11–15).

Autophagy has been shown to have a protective role against ER stress (16) and facilitate mitochondrial turnover (17). Indeed, it has been demonstrated that beta-cell Atg7−/<sup>−</sup> mice are characterized by islet degeneration, impaired insulin secretion, and glucose intolerance (18, 19). In addition, we have previously shown that a proportion of beta cells of type 2 diabetic (T2D) subjects presents a major increase of autophagic vacuoles and autophagosomes, associated with cell damage, which further suggests that altered autophagy might contribute to the loss of beta cell functional mass (20).

To shed further light on these issues, we presently explored the effects of autophagy modulation in isolated human islets under conditions of metabolically (palmitic acid) or chemically (brefeldin A) induced ER stress (21, 22). In addition, activators and inhibitors of autophagy were tested with pancreatic islets from T2D organ donors.

#### MATERIALS AND METHODS

#### Human Islet Isolation and Culture

Human islet collection and handling were approved by the local Ethics Committee. Human pancreatic islets were isolated by enzymatic digestion and density gradient purification from pancreatic samples of non-diabetic (ND; n = 17; age 65 ± 21 years; gender: 5 M/12 F; BMI 23.4 ± 3.3 kg/m<sup>2</sup> ) and T2D (n = 9; age 76 ± 6 years; 4 M/5 F; gender: BMI 25.4 ± 3.7 kg/m<sup>2</sup> ) organ donors as detailed elsewhere (23, 24). For the experiments with palmitate, ND islets were cultured for 1–5 days in M199 medium containing 1% BSA with 10 ng/ml rapamycin (autophagy inducer) (25), 5.0 mmol/l 3-methyladenine or 1.0 nmol/l concanamycin-A (autophagy blockers) (25), either in the presence or absence of 0.5 mmol/l palmitate, prepared as previously reported (24, 26). For the experiments with brefeldin A, ND islets were exposed to the autophagy modulators either in the presence or absence of 0.1 ng/ml of this chemical ER stress inducer. The islets prepared from T2D donors were studied after 24 h incubation with M199 medium containing or not 10 ng/ml rapamycin.

#### Electron Microscopy Evaluation

Electron microscopy studies were performed on isolated islets as previously described (27). Islets were fixed with 2.5% glutaraldehyde in 0.1 mol/l cacodylate buffer, pH 7.4 for 1 h at 4 ◦C. After rinsing in cacodylate buffer, the sample was postfixed in 1% cacodylate buffered osmium tetroxide for 2 h at room temperature, then dehydrated in a graded series of ethanol, briefly transferred to propylene oxide and embedded in Epon-Araldite. Ultrathin sections (60–80 nm thick) were cut with a diamond knife, placed on formvar-coated copper grids (200 mesh), and stained with uranyl acetate and lead citrate. The presence of marked chromatin condensation and/or blebs was considered as signs of apoptosis (28). Morphometric analyses were performed by stereological techniques (19, 24, 27). In particular, volume density of insulin granules, mitochondria and rough endoplasmic reticulum (RER) was estimated. Micrographs, obtained at 10,000× were analyzed by overlay with a graticule (11 × 11 cm) composed of 169 points. Volume density was calculated according to the formula: VD = Pi/Pt, where Pi is the number of points within the subcellular component and Pt is the total number of points, and expressed in ml/100 ml of tissue (ml%) (19, 24, 27).

## Insulin Secretion

Insulin secretion studies were performed by the batch incubation technique as previously described (29–31). Groups of approximately 15 islets of comparable size were incubated at 37◦C for 45 min in Krebs-Ringer bicarbonate solution (KRB), 0.5% albumin, pH 7.4, containing 3.3 mmol/l glucose. Then, the medium was removed and replaced with KRB containing 16.7 mmol/l glucose. After additional 45 min incubation, medium was collected. Insulin levels were measured by a commercially available immunoradiometric assay (Pantec Forniture Biomediche, Turin, Italy). Insulin secretion was expressed as stimulation index (SI), i.e., the ratio of stimulated (16.7 mmol/l glucose) over basal (3.3 mmol/l glucose) insulin secretion (29–31).

## Quantitative RT-PCR Experiments in Isolated Islets

Gene expression studies were performed as previously detailed (30). Total RNA was extracted using the PureLinkTM RNA Mini kit (Life technologies, Carlsbad, CA, USA) according to manufacturer recommendations and quantified by absorbance at A260/A280 nm (ratio >1.95) in a Nanodrop 2000c spectrophotometer (Thermo Scientific, Waltham, MA, USA). After it was reverse-transcribed with SuperScript VILO Master Mix (Life technologies), the levels of the genes of interest were normalized for the housekeeping gene beta actin and quantified by the 2−11Ct method in a VIIA7 instrument (Life technologies). The primers/ probe for the analyzed genes were purchased from Taqman <sup>R</sup> Assay on-demand library (Life technologies).

## p62 Evaluation in Isolated Human Islets

The levels of p62 were assessed in T2D islets by the p62 Elisa kit (Enzo Life Sciences, Lausen, Switzerland) following the manufacturer protocol. In brief, after 24 h exposure to rapamycin or 3-MA, the islets were collected, protein were extracted and aliquoted on a plate pre-coated with a p62 specific antibody. After having incubated the samples in presence of a second anti-p62 antibody (rabbit polyclonal), the amount of p62 was revealed by adding a secondary donkey anti-rabbit IgG antibody conjugated to horseradish peroxidase and a mix composed by TMB and hydrogen peroxide. The plate was read in a FLUOstar Omega plate reader (BMG Labtech, Ortenberg, Germany) and the amount of p62 normalized for the total amount of proteins.

## Statistical Analysis

Data are presented as mean ± SD. Differences between groups were assessed by the two-tailed Student's t-test or one-way ANOVA followed by the Bonferroni correction, as appropriate. A P < 0.05 was considered statistically significant.

## RESULTS

## Beta Cell Apoptosis and Insulin Secretion

Exposure of ND human islets to 0.5 mmol/l palmitate for 5 days or 0.1 ng/ml brefeldin-A for 24 h induced a significant increase of beta cell apoptosis compared to control islets (**Figures 1A,B**), confirming previously reported results (32, 33). In agreement with previous reports (20, 24) apoptosis was identified on the basis of internationally acknowledged criteria based on the appearance of marked chromatin condensation and blebs (28). Electron microscopy can be considered one of the best methods to identify apoptotic cells, because it enables not only the detection of apoptosis but also enables to identify which type of cell is undergoing apoptosis (see also **Supplementary Figure 1** ESM). The deleterious effects of both ER stressors were prevented by rapamycin (autophagy activator), whereas 3-MA, but not concanamycin (both are autophagy inhibitors), enhanced the rate of beta cell apoptosis in presence of palmitate (**Figures 1A,B**). Insulin secretion was significantly decreased by islet exposure to either palmitate or brefeldin A, and the presence of rapamycin protected beta cells from palmitate-induced insulin secretion alterations (**Figure 1C**).

## Electron Microscopy Studies

We then investigated the ultrastructural changes induced by the exposure of human islets to palmitate or brefeldin by quantitative morphometry. In particular, we assayed the volume density of insulin granules, mitochondria, and ER in beta cells exposed to 0.5 mmol/l palmitate for 5 days and 0.1 ng/ml brefeldin for 24 h. In addition, we evaluated the effects of the stimulation of autophagy by the concomitant exposure to 10 ng/ml rapamycin. Five days of palmitate exposure significantly decreased the volume density of insulin granules, whereas volume density of both mitochondria and ER was significantly increased (**Figures 2A–C**). In all cases, co-incubation with rapamycin was able to prevent the changes caused by palmitate exposure (**Figures 2A–C**). Brefeldin A also induced a significant reduction of insulin granule volume density, although quantitatively less compared to palmitate, and both mitochondria and ER volume density was significantly increased (**Figures 2A–C**). Rapamycin had no effects on the changes in insulin granules and mitochondria induced by brefeldin, whereas it was able to partially prevent brefeldin A-induced increase of ER volume density (**Figures 2A–C**).

#### Gene Expression in Isolated Human Islets

The expression of selected ER markers was then studied in isolated ND human islets exposed for 24 h to 0.5 mmol/l palmitate or to 0.1 ng/ml brefeldin; the modulating effect of the concomitant exposure to 10 ng/ml rapamycin was also evaluated. Palmitate exposure significantly increased the

expression of PERK, CHOP, and BiP with respect to control human islets, and this effect was significantly prevented by the concomitant presence of rapamycin (**Figure 3**, panels on the left). Brefeldin-A exposure markedly increased the expression of all the assayed ER stress markers in human islets with respect to controls, and the concomitant stimulation of autophagy by rapamycin in part counteracted this effect (**Figure 3**, panels on the right).

#### Rapamycin and T2D Human Islets

Finally, we explored the effects of rapamycin in human islets isolated from T2D donors. In T2D islets beta cell apoptosis was significantly higher than in ND islets, in agreement with previous results (31, 34), and exposure to rapamycin significantly decreased the number of apoptotic beta cells (**Figure 4A**). Rapamycin also caused a significant increase of glucosestimulated insulin secretion T2D human islets (**Figure 4B**).

Quantitative morphometry confirmed the ultrastructural alterations previously observed in human islets isolated from T2D patients (31, 34): volume density of insulin granules was significantly decreased whereas volume density of both mitochondria and ER was significantly increased compared to ND islets (**Figure 4C**). Exposure of diabetic islets to rapamycin was able to significantly counteract these alterations (**Figure 4C**). These favorable changes induced by rapamycin were associated with decreased expression of PERK, CHOP, and BiP similar to those observed in ND islets exposed to palmitate (**Figure 4D**).

#### p62 in T2D Islets

In order to confirm whether rapamycin and 3-MA could modulate autophagy flux, we evaluated the amount of p62, an autophagy receptor that accumulates in cells when macroautophagy is inhibited (21). In ND islets, we found 1.15 ± 0.36 ng/ml of p62/µg total proteins. In T2D islets, p62 levels (3.62 ± 0.90 ng/ml of p62/µg total proteins) were higher with respect to non-diabetic islets (p < 0.05). Exposure to rapamycin significantly reduced the amount of p62 (**Figure 5**), indicating that the drug could increase the autophagy flux in T2D islets. Conversely, the use of 3-MA, an autophagy inhibitor, significantly increased the levels of the receptor (**Figure 5**).

## DISCUSSION

The synthesis, modification and delivery of proteins to their target sites occur in the ER. Several physiological and pathological conditions are able to impair these processes leading to ER stress, and in the last years a large body of evidences has

been provided indicating that ER stress may be implicated in beta cell dysfunction and death in diabetes (22, 23, 35–42).

Recently, autophagy has emerged as a crucial protective mechanism during ER stress (43, 44). An early study reported that cells undergoing ER stress, as indicated by ER expansion, showed a concomitant increase in autophagosome abundance (44). This activation of autophagy was interpreted as a survival mechanism to prevent ER stress-induced toxicity (43, 44). In mammals, ER stress inducers generally act as on-off switches for mTOR-regulated cell growth, survival and energy balance, through the downregulation of AKT1, which induces the activation of autophagy by decreasing mTOR activity (45). Autophagy is a degradation mechanism that can be induced by starvation or other form of nutrient deprivation to supply substrates for cellular energy generation (46). Autophagy also serves as a catabolic pathway to recycle excessive or damaged

panels) in the presence (black bars) or absence (white bars) of 10 ng/ml rapamycin. N = 4–5 for both treatments. C, control; P = 0.5 mmol/l palmitate; B = 0.1 ng/ml brefeldin; Statistical analysis: \*p < 0.05 vs. all the other groups; §p < 0.05 vs. C, after Bonferroni correction.

intracellular organelles such as mitochondria (47). Therefore, it can act as a housekeeping mechanism in the absence of stress, while under stress conditions it exerts a crucial protective role (48).

The aim of this study was to explore the effects of the modulation of autophagy on the ER stress-induced beta-cell dysfunction in isolated human pancreatic islets. We induced ER stress in cultured human islets by their

N = 9 (300–400 cells, two islet preparations analyzed). Statistical analysis: \*p < 0.05 vs. T2D. (B) Glucose-stimulated insulin secretion of type 2 diabetes human islets in the presence (black bar) or absence (white bar) of 10 ng/ml rapamycin. N = 5. Statistical analysis: \*p < 0.05 vs. T2D. (C) Quantitative morphometric analysis of volume density of insulin granules, mitochondria, and endoplasmic reticulum in beta cells of type 2 diabetes human islets in the presence (black bar) or absence (white bar) of 10 ng/ml rapamycin. N = 9 (300–400 cells, two islet preparations analyzed). Statistical analysis: \*p < 0.05 vs. T2D. (D) Expression of several ER stress markers in type 2 diabetes human islets in the presence (black bars) or absence (white bar) of 10 ng/ml rapamycin. N = 5. Statistical analysis: \*p < 0.05 vs. T2D.

exposure to a metabolic (0.5 mmol/l palmitic acid for 5 days) or a chemical (0.1 ng/ml brefeldin for 1 day) stressor. The increased levels of fatty acids, commonly associated with obesity, can induce insulin resistance and beta-cell dysfunction, making them good candidates to explain the link between obesity and the development of T2D (49–51). It has been proposed that palmitateinduced ER stress may arise from the disruption of protein processing and trafficking (49), or from incorrect Ca++ regulation (52). Brefeldin A is a macrocyclic lactone antibiotic which is synthetized from palmitate by several fungi and has been shown to induce ER stress through the inhibition of ADP ribosylation factor (ARF) and the consequent disruption of the ER-Golgi vesicular transport (53, 54).

Here, we confirm that lipotoxic conditions (such as palmitate exposure) and chemically-induced (brefeldin) ER stress are associated with alterations of beta cell survival and function, and show that the modulation of autophagy influences these effects. In particular, rapamycin, an inducer of autophagy through inhibition of mTORC1 complex (25), was able to prevent beta cell apoptosis induced by palmitate or brefeldin, and restore a proper insulin secretion in response to glucose which was altered by the presence of the fatty acid. These data are in line with those previously published on the effect of rapamycin in Akita mice (55), where the drug attenuated cellular stress and apoptosis that, conversely, were exacerbated by autophagy inhibitors in conditions of accumulation of misfolded proinsulin. Conversely, 3-MA and concanamycin-A were used as autophagy inhibitors. The first blocks an early stage of autophagy by inhibiting the class III PI3K (21), while concanamycin causes an increase in lysosomal/vacuolar pH, and, ultimately, blocks fusion of autophagosomes with the vacuole by inhibiting V-ATPase (25). In our study, the presence of 3-MA caused a significant increase of betacell apoptosis in human islets exposed to palmitate, whereas concanamycin A did not show any effect on the cytotoxicity induced by palmitate or brefeldin. Our data are in agreement with previously published results in other cell types (56), where it was observed that inhibition of autophagy at an early stage, but not at a late stage, potentiated chemosensitivity, increasing caspase 3/7 activation, especially in conditions of high levels of autophagy. More in detail, 3-MA could exert this proapoptotic action making available beclin-1 (a master component of the PI3K-III complex inhibited by the drug, playing a role also in apoptosis) for caspase 8 cleavage in order to elicit cell death (57–60). However, other studies exist reporting a more deleterious effect of late autophagy blockade on cell survival (61).

At the ultrastructural level, we found that in both palmitate and brefeldin-treated human islets a significant decrease in the volume density of insulin granules, and a significant increase in volume density of mitochondria and ER was present. Coexposure with rapamycin was able to prevent these alterations mainly in the islets treated with palmitate. These beneficial effects on ultrastructure were associated with a reduction in the expression of some ER stress-related genes. The ultrastructural alterations found in our experiments were similar to those previously observed in other studies. In particular, they look like those observed in beta cells isolated from Atg1β−cell mice with beta cell-specific deletion of Atg7 (autophagy-related 7) (19) indicating that autophagy could be necessary to maintain beta-cell homeostasis (62, 63) and in human islets isolated from T2D patients (20, 23, 27) confirming a pathogenetic role played by ER stress (50, 64). However, the role of autophagy in diabetes pathophysiology has not been fully elucidated. Recently, it was suggested that impaired autophagy could lead to accumulation of dysfunctional organelles such as mitochondria (65) and that in type 2 diabetic pancreatic beta cells, a massive overload of autophagic vacuoles and autophagosomes might contribute to the loss of beta-cell mass (20). In addition, some authors reported that rapamycin improved insulin resistance and hepatic steatosis in T2D rats via activation of autophagy (66). Our results in T2D islets showed that the promotion of autophagic process by rapamycin, as evaluated by the clearance of the p62 protein, is associated with amelioration of function, survival and ultrastructure possibly due to reduction of ER stress. Recently, it was hypothesized that mTORC1 (one of the major signaling complex in beta cells, where it is responsible for nutrient sensing and beta cell growth) (67), if short term and transiently activated, regulates beta cell replication, anabolic growth and insulin secretion under physiological conditions (67). Conversely, its sustained activation (as in presence of chronic excess of nutrients) is associated with impaired insulin release, ER stress and reduced beta cell survival. In this regards, several authors showed that chronic exposure to high glucose and/or high free fatty acid concentrations could activate mTORC1 in beta cells and that its inhibition could prevent gluco- lipotoxicinduced beta cell derangement (68). In particular, Yuan and colleagues (69) showed that islets isolated from T2D organ donors had increased mTORC1 activity in comparison with nondiabetic islets and that this augmentation was present in beta but not alpha cells. Moreover, mTORC1 genetic or chemical inhibition was associated with restoration of insulin release in T2D islets. Differently from them, in our study, 10 ng/ml rapamycin (corresponding to around 11 nmol/l) were able to promote beta cell function, survival and ultrastructure in T2D islets. Similarly, it has been observed that the defective autophagic flux associated with a lysosomal dysfunction observed in T2D islets can be restored also by a GLP-1 receptor agonist (70). It should be underlined that the beneficial effects observed with rapamycin in presence of ER stress modulators, especially those related to insulin release, are exerted in a short-term setting. Additional experiments should be performed to assess the role of a chronic exposure to the drug. In this regards, some authors have observed how some pharmacological agents, currently used in diabetes treatment, show deleterious effects in beta cells when chronically administered in presence of rapamycin (71).

In conclusion, this study provides information on how conditions of metabolically or chemically induced stress on human beta cells associate with reduced beta cell survival, impaired beta cell function, and ultrastructural alterations, which are also mediated by ER stress. More importantly, our data suggest that promotion of autophagy at the beta cell level, in some context, might be helpful to protect beta cell health.

## ETHICS STATEMENT

The study was approved by the Ethics Committee of the University of Pisa.

## AUTHOR CONTRIBUTIONS

MB, PM, and VD conceived the study and wrote the manuscript. MB, SM, FG, MS, LM, UB, PD, DE, and MC researched data. All the authors contributed to discussion and reviewed the manuscript. VD is the guarantor of this study.

### FUNDING

This study has received funding from the Italian Ministry of University and Research (PRIN 2015, grant number 2015373Z39\_008) and from the European Union Project T2DSystems in the Horizon 2020 Programme of the European Commission, grant number 667191.

## SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fendo. 2019.00052/full#supplementary-material

#### REFERENCES


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Bugliani, Mossuto, Grano, Suleiman, Marselli, Boggi, De Simone, Eizirik, Cnop, Marchetti and De Tata. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Palmitic Acid Reduces the Autophagic Flux and Insulin Sensitivity Through the Activation of the Free Fatty Acid Receptor 1 (FFAR1) in the Hypothalamic Neuronal Cell Line N43/5

María Paz Hernández-Cáceres 1,2‡, Lilian Toledo-Valenzuela1,2‡, Francisco Díaz-Castro2,3 , Yenniffer Ávalos 1,2†, Paulina Burgos 1,2, Carla Narro1,2, Daniel Peña-Oyarzun2,4,5 , Jasson Espinoza-Caicedo1,2, Flavia Cifuentes-Araneda1,2, Fernanda Navarro-Aguad<sup>6</sup> , Cecilia Riquelme<sup>6</sup> , Rodrigo Troncoso2,3,4, Alfredo Criollo2,4,5 and Eugenia Morselli 1,2 \*

<sup>1</sup> Laboratory of Autophagy and Metabolism, Department of Physiology, Faculty of Biological Sciences, Pontificia Universidad Católica de Chile, Santiago, Chile, <sup>2</sup> Autophagy Research Center, Santiago, Chile, <sup>3</sup> Research Laboratory of Nutrition and Physical Activity, Institute of Nutrition and Food Technology, Universidad de Chile, Santiago, Chile, <sup>4</sup> Advanced Center for Chronic Diseases and Center for Molecular Studies of the Cell, Universidad de Chile, Santiago, Chile, <sup>5</sup> Facultad de Odontología, Instituto de Investigación en Ciencias Odontológicas, Universidad de Chile, Santiago, Chile, <sup>6</sup> Laboratory of Differentiation and Pathology, Department of Cell and Molecular Biology, Faculty of Biological Sciences, Pontificia Universidad Católica de Chile, Santiago, Chile

Chronic consumption of high fat diets (HFDs), rich in saturated fatty acids (SatFAs) like palmitic acid (PA), is associated with the development of obesity and obesity-related metabolic diseases such as type II diabetes mellitus (T2DM). Previous studies indicate that PA accumulates in the hypothalamus following consumption of HFDs; in addition, HFDs consumption inhibits autophagy and reduces insulin sensitivity. Whether malfunction of autophagy specifically in hypothalamic neurons decreases insulin sensitivity remains unknown. PA does activate the Free Fatty Acid Receptor 1 (FFAR1), also known as G protein-coupled receptor 40 (GPR40); however, whether FFAR1 mediates the effects of PA on hypothalamic autophagy and insulin sensitivity has not been shown. Here, we demonstrate that exposure to PA inhibits the autophagic flux and reduces insulin sensitivity in a cellular model of hypothalamic neurons (N43/5 cells). Furthermore, we show that inhibition of autophagy and the autophagic flux reduces insulin sensitivity in hypothalamic neuronal cells. Interestingly, the inhibition of the autophagic flux, and the reduction in insulin sensitivity are prevented by pharmacological inhibition of FFAR1. Our findings show that dysregulation of autophagy reduces insulin sensitivity in hypothalamic neuronal cells. In addition, our data suggest FFAR1 mediates the ability of PA to inhibit autophagic flux and reduce insulin sensitivity in hypothalamic neuronal cells. These results reveal a novel cellular mechanism linking PA-rich diets to decreased insulin sensitivity in the hypothalamus and suggest that hypothalamic autophagy might represent a target for future T2DM therapies.

Keywords: saturated fatty acids, G-protein coupled receptor 40, central nervous system, AKT, insulin resistance, autophagy, glucose uptake

#### Edited by:

Maria Ines Vaccaro, Universidad de Buenos Aires, Argentina

#### Reviewed by:

Angelo Poletti, University of Milan, Italy Franck Oury, Institut National de la Santé et de la Recherche Médicale (INSERM), France

#### \*Correspondence:

Eugenia Morselli emorselli@bio.puc.cl

#### †Present Address:

Yenniffer Ávalos, Departamento de Biología, Facultad de Química y Biología, Universidad de Santiago de Chile, Santiago, Chile

> ‡These authors have contributed equally to this work

#### Specialty section:

This article was submitted to Endocrinology of Aging, a section of the journal Frontiers in Endocrinology

Received: 20 August 2018 Accepted: 01 March 2019 Published: 26 March 2019

#### Citation:

Hernández-Cáceres MP, Toledo-Valenzuela L, Díaz-Castro F, Ávalos Y, Burgos P, Narro C, Peña-Oyarzun D, Espinoza-Caicedo J, Cifuentes-Araneda F, Navarro-Aguad F, Riquelme C, Troncoso R, Criollo A and Morselli E (2019) Palmitic Acid Reduces the Autophagic Flux and Insulin Sensitivity Through the Activation of the Free Fatty Acid Receptor 1 (FFAR1) in the Hypothalamic Neuronal Cell Line N43/5. Front. Endocrinol. 10:176. doi: 10.3389/fendo.2019.00176

## INTRODUCTION

Obesity is currently considered a global epidemic both in developed and developing countries. Excessive accumulation of body fat promotes obesity-associated metabolic dysfunctions, such as insulin resistance, type II diabetes mellitus (T2DM), cardiovascular diseases, neurodegeneration, and certain cancers (1).

Consumption of high fat diets (HFDs), and especially of diets high in saturated fatty acids (SatFAs) such as the palmitic acid (PA), increases body weight and is associated with the onset of several metabolic diseases, such as insulin resistance and T2DM (2, 3). Previous studies from our and other groups show that SatFAs accumulate in the central nervous system (CNS), specifically in the hypothalamus (4–7). Interestingly, fatty acid (FA) sensing neurons reside in the hypothalamus and play important roles in feeding behavior, as well as lipid and glucose metabolism. Due to their critical role in the regulation of energy homeostasis, these neuronal populations have been implicated in development of obesity and T2DM (8, 9).

Macroautophagy, hereafter referred to as "autophagy," is a conserved adaptive mechanism that maintains the balance between synthesis, degradation, and recycling of cellular components (10). Autophagy supports cellular homeostasis and studies in multiple cell types have demonstrated autophagic dysregulation following chronic consumption of HFDs (11). Previous studies from our and other groups have shown that hypothalamic autophagic flux is defective in rodents chronically fed with HFDs (12, 13). Consistently, previous work demonstrates that the inhibition of autophagy in proopiomelanocortin (POMC) neurons leads to metabolic disorders, including obesity and insulin resistance (14– 16); however the mechanism by which this occurs is still largely unknown.

Evidence exists of crosstalk between autophagy and insulin signaling pathways. For example, hyperinsulinemia reduces the autophagic response in different peripheral tissues and cell types (17–20), whereas inhibition of autophagy (by autophagy related gene 7 -ATG7- tissue-specific knockout) promotes insulin resistance in liver tissue (21) and reduces insulin secretion in pancreatic β-cells (22, 23). However, whether inhibition of autophagy directly reduces neuronal insulin sensitivity is currently unknown.

The Free Fatty Acid Receptor 1 (FFAR1), also known as G protein-coupled receptor 40 (GPR40), is a seven-transmembrane domain receptor activated by medium and long-chain fatty acids, including PA (24, 25). Previous studies in GPR40−/<sup>−</sup> mice show that this receptor plays a significant role in the chain of events linking obesity and metabolic disorders, as GPR40−/<sup>−</sup> mice are protected from obesity-induced hyperinsulinemia, hyperglycemia, and glucose intolerance (26). Moreover, FFAR1 is required to mediate the insulin response to FA in the pancreas (27–29). However, its function in the hypothalamus, specifically in the context of autophagy and insulin sensitivity, is not known.

Here, we demonstrate that exposure to the SatFA PA inhibits the autophagic flux and reduces insulin sensitivity in N43/5 cells, a hypothalamic neuronal cell line. Furthermore, our findings show that autophagy malfunction promotes insulin resistance in hypothalamic neuronal cells. In addition, our data suggest FFAR1 may link PA with reduced insulin sensitivity and impaired autophagic flux in hypothalamic neuronal cells. Altogether, these results suggest activation of autophagy should be considered as treatment for insulin resistance.

## MATERIALS AND METHODS

#### Cell Culture and Treatments

N43/5 cells (Cellutions Biosystems) were cultured in Dulbecco's modified eagle medium (DMEM) high glucose (11995-040, Gibco, USA) supplemented with 10% of fetal bovine serum (FBS) (10437028, Gibco), 100 U/ml penicillin streptomycin (15140122, Gibco) and maintained at 37 ◦C with 5% CO2. To evaluate the changes in the autophagic flux in response to PA exposure, cells were incubated with DMEM high glucose supplemented with 2% of FBS 24 h before treatments and then exposed to 100µM PA (P0500, Sigma-Aldrich, St. Louis, MO, USA) conjugated to fatty acid-free bovine serum albumin (BSA) (152401, MP Biomedicals, Santa Ana, CA, USA). BSA treatment was used as control. To assess the autophagic flux and insulin response, cells were incubated at the time points indicated in each experiment with the autophagic flux inhibitor Bafilomycin A1 (BafA1, 100 nM) (B1793, Sigma-Aldrich) or with its vehicle DMSO (BM-0660, Winkler). To determine the effect of 6-h treatment with PA or BafA1 on insulin signaling, cells were serum starved in medium DMEM/F-12 (11330-32, Gibco) overnight prior to treatments. Then, cells were co-treated with insulin (1 nM) (I0516, Sigma-Aldrich) or phosphate-buffered saline (PBS). To evaluate the involvement of FFAR1 in PA-mediated inhibition of the autophagic flux or reduction in insulin signaling, cells were pre-incubated with the FFAR1 antagonist GW1100 (1µM; CAS 306974-70-9, Calbiochem, San Diego, CA, USA) for 20 min or with its vehicle (DMSO), followed by co-incubation with PA or BSA at the indicated time point, depending on the readouts.

#### Animals

Animal care and procedures were approved by the Ethical Committee of the Pontificia Universidad Católica de Chile. Male C57BL/6 mice were housed in a temperature-controlled environment in groups of two to four at 22–24◦C using a 12 h light/dark cycle. Mice were fed a standard chow (Prolab <sup>R</sup> RMH3000).

#### Animals Gavage

Mice (n = 3–5/group) received an intragastric gavage of nutritionally complete diet as described by Benoit et al. (30). Briefly, the PA diet contains 20 g fat/100 g diet (19 g of ethyl palmitate dissolved in medium-chain triglyceride and 1 g of soybean oil to provide essential fatty acids), while the control diet contains 3 g ethyl palmitate and 1 g soybean oil/100 g diet. This liquid diet was injected into the stomach of the mice using a gavage-feeding needle. Mice were given 3 equally sized feedings daily at 7:00 AM, 12:00 PM, and 5:00 PM for 3 days. At the end of day 3, mice were sacrificed 3 h after the final gavage. Brains were removed, hypothalami dissected and stored in RNAlater (Ambion, Life Technologies) at 4 ◦C. 24 h later, tissues were homogenized in 1 ml of TRIzol (Ambion, Life Technologies) and RNA was extracted using the RNeasy Kit (Qiagen Sciences, Inc., Germantown, MD, USA) according to the manufacturer's instructions. Total RNA (1 µg) was reverse transcribed using the SuperScript III First-Strand Synthesis System (Invitrogen) according to the manufacturer's instructions and cDNA was used for RT-PCR assays.

#### siRNA Transfections

Cells were cultured in six-well plates and transfected at 50% confluence with siRNAs targeting murine Beclin 1 (BECN1) (SASI\_Mm01\_00048143, Sigma-Aldrich) or murine autophagy related gene 7 (ATG7) (SASI\_Mm01\_00044616, Sigma-Aldrich). Transfection was performed using Lipofectamine RNAiMAX <sup>R</sup> Transfection Reagent (Invitrogen, Carlsbad, CA, USA) according to the manufacturer's instructions. As negative control, cells were incubated with Lipofectamine RNAiMAX <sup>R</sup> Transfection Reagent only. 48 h after siRNA transfection, cells were treated as indicated or directly lysed for protein or RNA extraction.

#### Western Blot Analysis

Cells were lysed in RIPA buffer and 30–40 µg of denatured proteins from each sample were resolved in 8–12% SDS-PAGE. Gels were transferred to nitrocellulose membranes and incubated with 5% BSA (BM-0150, Winkler, RM, Chile)-trisbuffered saline-0.1% Tween-20 (TBS-T) to block nonspecific binding. Membranes were incubated with the primary antibodies anti LC3A/B (4108, Cell Signaling Technology, Danvers, MA, USA), SQSTM1 (H00008878-M01, Abnova, Jhouzih St., Taipei, Taiwan), p-Insulin Receptor β (Tyr1150/1151) (3024, Cell Signaling Technology), IR (ab131238, Abcam), p-AKT (Ser473) (9271, Cell Signaling Technology), AKT (9272, Cell Signaling Technology), BECN1 (H-300; sc-11427, Santa Cruz Biotechnology, Inc., Dallas, Texas, USA), at dilution of 1:1,000 in 5% BSA-TBS-T overnight on a rocking platform at 4 ◦C. Then, membranes were washed 3 times for 10 min in TBS-T and revealed with the appropriate horseradish peroxidase-labeled secondary antibodies (Goat Anti-Mouse IgG (H + L)-HRP Conjugate, 1706516; Goat Anti-Rabbit IgG (H + L)-HRP Conjugate, 1706515; Bio-Rad, CA, USA) and the chemiluminescent substrate. GAPDH (1:1000; sc-365062, Santa Cruz Biotechnology, Inc.) and β-actin (1:10,000; A1978, Sigma-Aldrich) were used as loading control. To evaluate insulin signaling the same samples were run on parallel gels, one for p-AKT and the other for AKT. Analysis of data was performed by comparing p-AKT vs. β-actin and AKT vs. β-actin. The obtained ratios were analyzed.

#### Immunofluorescence and Fluorescence Microscopy

For fluorescence microscopy determinations in N43/5 cells, cells cultured on coverslips were fixed with cold methanol (−20 ◦C) for 10 min. Cells were blocked in 3% BSA in PBS for 1 h and then incubated with the following primary antibodies overnight at 4 ◦C. The primary antibodies used are LC3A/B (1:250; Cell Signaling Technology), p62/SQSTM1 (1:300; Abnova). Primary antibodies staining was followed by conjugation with its respective secondary antibody (1:300; Alexa Fluor <sup>R</sup> , Life Technologies) for 1 h at room temperature. Nuclei were counterstained with Hoechst 33342 (10 mg/ml) (Molecular Probes, Eugene, OR, USA) or with ProLong <sup>R</sup> Gold Antifade Mountant (P36931, Molecular Probes, Eugene, OR, USA). Images were taken in an inverted fluorescence microscopy (Nikon Eclipse Ti, Tokio, Japon).

For fluorescence microscopy determinations on brain sections, mice were anesthetized and perfused with 10% formalin. Brains were dissected and post-fixed in 10% formalin for 24 h followed by treatment with 30% sucrose in PBS. Brain sections were cut at 30µm using a Thermo Scientific HM 450 sliding microtome (Thermo Scientific). The sections were permeabilized in 0.01% Triton and blocked in 3% bovine BSA (BM-0150 Winkler) for 1 h. Brain sections were incubated overnight at 4 ◦C with the primary antibody against GPR40 (Y17; sc-28416, Santa Cruz Biotechnology, Inc.) in combination with either a marker of neurons (NeuN; MAB377, Millipore, Burlington, MA, USA), astrocytes (GFAP; G9269, Sigma-Aldrich) or microglia (IBA-1; 019-19741, Wako, USA), followed by conjugation with the respective secondary antibodies (Alexa Fluor, Life Technologies) for 1 h at room temperature. Sections were placed on gelatinized slides, mounted with VECTASHIELD anti-fading medium with DAPI (Vector Laboratories), and coverslipped. Pictures containing the arcuate nucleus (ARC) region of the hypothalamus were taken in a confocal fluorescence microscopy (Carl Zeiss, LSM700, Oberkochen, Germany).

## Real Time PCR (RT-PCR)

For analysis of gene expression, mice were anesthetized and decapitated. Tissues were stored in RNAlater (Ambion, Life Technologies) at 4◦ C, and 24 h later, hypothalami dissected and then homogenized in 1 ml of TRIzol (Ambion, Life Technologies). RNA was extracted using the RNeasy Kit (Qiagen Sciences, Inc., Germantown, MD, USA) according to the manufacturer's instructions. Total RNA (1 µg) was reverse transcribed using the SuperScript III First-Strand Synthesis System (Invitrogen) according to the manufacturer's instructions.

For analysis of Atg7 gene expression in N43/5 cells, RNA was extracted using the E.Z.N.A. <sup>R</sup> Total RNA Kit (OMEGA Bio-Tek, Norcross, GA, USA) according to the manufacture's indications. cDNA was synthesized using iScriptTM cDNA kit (Bio-Rad) from 1 µg of total RNA.

Quantitative PCR reactions were carried out on a Step One System Real Time PCR (Applied Biosystems) using Fast SYBR Green Master Mix (4385370, Life Technologies). Specific forward and reverse primers sequences used to evaluate gene expression are indicated in **Supplemental Table 1**. Hprt1 was used as housekeeping gene. The 11C<sup>T</sup> method was used for relative quantification analysis.

#### Intracellular Calcium Measurement

Intracellular calcium (Ca2+) was determined using a spectrofluorometric technique described by Darling R.A. and collaborators (31). N43/5 cells were loaded with 1.5µM Fura 2-AM (Molecular Probes) for 45 min at 37 ◦C. Cells were

excited at a wavelength of 340 and 380 nm in a Nikon Diaphot microscope equipped with a Photon Technology International spectrofluorometer (Lawrenceville, NJ). To confirm the presence of a functional FFAR1 in our cellular model, cells were preincubated with its antagonist GW1100 (1µM) for 20 min. After 10 min of stabilization, the intensity ratio was continuously recorded and cultures were co-stimulated for 210 seconds with 100µM PA or BSA. Fluorescence emission intensity was detected with a photomultiplier and analyzed using a fluorescence analysis program (FELIX version 1.1).

#### 2-NBDG Uptake

To assess the insulin-dependent glucose uptake, N43/5 cells were stimulated with insulin 1 nM for 30 min and incubated with the fluorescent analog of glucose (2-NBDG, 300 mM) for 15 min at 37 ◦C as previously described (32). To evaluate the involvement of FFAR1 in PA-mediated inhibition of insulindependent glucose uptake, cells were pre-incubated with the FFAR1 antagonist GW1100. Cells were transferred to an inverted Nikon Ti Eclipse microscope equipped with 40X oil objective [numerical aperture, N.A. 1.3]. A Xenon lamp was coupled to the monochromator device (Cairn Research Ltd, Faversham, UK). Digital images were acquired by means of a cooled CCD camera (Hamamatsu ORCA 03, Japan). Images were quantified by ImageJ software (NIH, Bethesda, MD).

#### Results and Statistical Analysis

Results are shown as mean ± SEM from at least 3 independent experiments. Two groups were compared using two-tailed Student's t tests. For more than two groups, one- or twoway ANOVA was used, as appropriate, followed by post hoc adjustment. All analyses were performed with GraphPad software (San Diego, CA, USA). P value of < 0.05 was considered statistically significant.

#### RESULTS

## Palmitic Acid Inhibits the Autophagic Flux in N43/5 Cells

We and others have previously shown that autophagy is dysregulated in vivo in the hypothalamus of male mice chronically fed with HFDs (12, 13, 33). Interestingly, we also demonstrated that in the brain of mice fed with the same diet for 16 weeks, SatFAs are significantly increased (4, 6, 7). Among those PA represents the most abundant SatFA, thus, we decided to evaluate whether its increase affects basal autophagy in hypothalamic neurons. We exposed N43/5 cells, a model of POMC neurons (34), to 100µM PA, a concentration of PA similar to the one identified in the brain of obese mice chronically exposed to HFDs (7). As indicated by western blot, the levels of the autophagy receptor Sequestosome 1/p62 (SQSTM1), which is degraded during the autophagic process, increased compared to control (BSA vs. PA: ∗∗p < 0.01; **Figures 1A,D**), suggesting PA inhibited autophagic flux. This result was supported by measuring levels of microtubule-associated protein light chain 3 II (LC3II) in presence and absence of Bafilomycin A1 (BafA1), a classic autophagic flux inhibitor which prevents fusion of autophagosomes and lysosomes (**Figures 1A–C**). PA exposure increased the levels of LC3II; however, addition of BafA1 had not effect (BSA vs. PA: ∗∗p < 0.01; BafA1 vs. BafA1 + PA: non-significant; **Figures 1A–C**). Consistent with this, immunofluorescence studies in N43/5 cells stained against LC3 showed that treatment with PA stimulated the formation of autophagic puncta per se already 1 h following PA exposure (**Figures 1E,F** and **Supplemental Figures 1A,B**). However, exposure to BafA1 did not further enhance the PA-triggered induction of LC3 puncta (**Figures 1E,F** and **Supplemental Figures 1A,B**). Finally, in addition to LC3 puncta, PA treatment also increased the number of SQSTM1 puncta, already 1 h following PA treatment (**Figures 1E,G** and **Supplemental Figures 1A,C**). Importantly, the levels of Atg5, Atg7, Atg16, Beclin 1, Fip200, and Gabarap are not increased by PA exposure, and Lc3 and Sqstm1 levels, even if affected, are significantly increased only 6 and 4 h, respectively, following PA exposure (**Supplemental Figures 1D–K**), thus suggesting PA treatment specifically affects the autophagic flux.

Importantly, as PA is not the only SatFA increased in the brain of male mice chronically fed with HFDs, to determine if the inhibition of the autophagic flux is specific to PA or not, we evaluated the effect of stearic acid (SA) on autophagy on N43/5 cells. The results we obtained show also SA inhibits the autophagic flux (**Supplemental Figure 2**). Indeed, as indicated by western blot, the levels of SQSTM1 are significantly increased when compared to BSA (BSA vs. SA: <sup>∗</sup>p < 0.05; **Supplemental Figures 2C,D**). Consistent with this, the levels of LC3II, which are increased by SA per se, were not affected by BafA1 (BSA vs. SA: ∗∗p < 0.01, BafA1 + BSA vs. BafA1 + SA: non-significant; **Supplemental Figures 2A,B**). Accordingly, SA stimulated the formation of autophagic puncta, as assessed by immunofluorescence (**Supplemental Figures 2E,F**) and exposure to BafA1 did not further enhance the SA-triggered induction of LC3 puncta (**Supplemental Figures 2E,F**). Finally, in addition to LC3 puncta, SA treatment also increased the number of SQSTM1 puncta (**Supplemental Figures 2E,G**). Altogether, these data indicate that SatFAs inhibit the basal rate of autophagy in N43/5 hypothalamic neurons.

### Palmitic Acid Reduces Insulin Sensitivity in N43/5 Cells

Consumption of HFDs causes hypothalamic insulin resistance in rodents (35). Importantly, insulin activates neurons located in the arcuate nucleus of the hypothalamus (such as POMC neurons), thereby reducing food intake. In conditions of chronic caloric excess (as occurs in HFDs feeding), insulin resistance develops and contributes to pathological weight gain (36). Moreover, consumption of HFDs increases PA levels in the hypothalamus (4, 5), and intracerebroventricular injection of PA blunts hypothalamic insulin signaling (30). Here, we evaluated the sensitivity of our cellular model to insulin, and whether PA exposure could reduce insulin sensitivity in the same cell type. Following exposure to insulin, levels of AKT phosphorylation (p-AKT) increased (**Figures 2A,B**), showing that N43/5 hypothalamic neuronal cells are, indeed, sensitive

\*\*\*\*p < 0.0001. n = 3.

to insulin. Critically, PA treatment reduced phosphorylation levels (BSA + Ins vs. PA + Ins: ∗∗p < 0.01; **Figures 2A,B**) and consistently, PA exposure inhibited insulin-induced glucose uptake (BSA + Ins vs. PA + Ins: ∗∗p < 0.01; **Figures 2C,D**). Importantly, these effects seem not be specific of PA, as also SA exposure reduces insulin sensitivity, as indicated by p-AKT levels in presence and absence of insulin (BSA + Ins vs. SA + Ins: ∗∗∗∗p < 0.0001; **Supplemental Figure 3**). Altogether these results indicate SatFAs reduce insulin sensitivity in N43/5 cells.

#### Inhibition of Autophagy Reduces Insulin Sensitivity in N43/5 Cells

Our data indicate PA inhibits the autophagic flux and reduces insulin sensitivity. In the next series of experiments, we determined if this response is specific to PA or if other compounds that inhibit the autophagic flux also affect the insulin response. First, we exposed N43/5 cells to BafA1 or vehicle for 6 h and added insulin during the last 3 or 15 min. While insulin treatment increased p-Insulin Receptor (IR) and p-AKT

levels (3 and 15 min following insulin treatment, respectively), again suggesting N43/5 cells are sensitive to insulin, pretreatment with BafA1 prevented the increase in their phosphorylation (Ins vs. BafA1 + Ins: ∗∗p < 0.01 for p-IR and <sup>∗</sup>p < 0.05 for p-AKT; **Figures 3A–C**). Importantly, the reduction in insulin sensitivity is confirmed by the reduction in glucose uptake, which is blunted by BafA1 treatment (Ins vs. BafA1 + Ins: <sup>∗</sup>p < 0.05; **Figures 3D,E**). No differences were seen in Glut4 expression, suggesting the effect of BafA1 does not occur at the transcriptional level (**Supplemental Figure 4**). In addition, BafA1 exposure increased the levels of LC3II and SQSTM1 (**Figure 3A**), suggesting that inhibition of the autophagic flux reduces the insulin response in hypothalamic neuronal cells.

Then, we assessed if inhibition of autophagy, and not only the inhibition of the autophagic flux, could affect the response to insulin in our cellular model. To do this, we downregulated the expression of two autophagy essential genes Beclin1 and Atg7 (**Figures 3F,J**) and evaluated the levels of IR and p-AKT in response to insulin treatment. Interestingly, downregulation of both autophagy essential genes significantly reduced the levels of p-IR and p-AKT following insulin treatment (Ins vs. siBeclin + Ins: <sup>∗</sup>p < 0.05 for p-IR and ∗∗p < 0.01 for p-AKT. Ins vs. siAtg7 + Ins: ∗∗p < 0.01 for p-IR and p-AKT; **Figures 3F–L**). Glut4 expression levels were increased following Atg7 downregulation, while no differences were identified following BafA1 exposure or Beclin1 downregulation (**Supplemental Figure 4**). Altogether these results indicate inhibition of autophagy, by downregulation of autophagy essential genes, reduces insulin sensitivity of N43/5 cells in vitro.

## Palmitic Acid Activates FFAR1 in N43/5 Cells

Given the increased PA in the hypothalamus of HFDs animals (5), we addressed whether PA affects the levels of FFAR1. Consistent with previous work (37, 38), we identified FFAR1 in hypothalamic neurons but not astrocytes or microglia (**Figures 4A,B**). To determine the effect of PA on FFAR1 levels we gavaged male mice with PA. Importantly, this experiment was performed only in male mice as the increase in PA concentration in the brain, and more specifically in the hypothalamus, was identified only in male mice (4–7). Our data showed Ffar1 mRNA levels increased following PA gavage (BSA vs. PA <sup>∗</sup>p < 0.05; **Figure 4C**). Furthermore, consistently with the results we obtained in N43/5 cells, Sqstm1 levels increased, while Atg genes were not affected (**Supplemental Figure 5**).

The activation of FFAR1 by FAs triggers the phospholipase C (PLC)/ Inositol trisphosphate (IP3) signaling pathway leading to Ca2<sup>+</sup> release from the endoplasmic reticulum (24, 39). Thus, to determine if PA exposure activates FFAR1 in our cell model, we assessed intracellular Ca2<sup>+</sup> levels following PA treatment. Intracellular Ca2<sup>+</sup> levels were significantly increased in N43/5 cells after exposure to PA relative to vehicle (BSA vs. PA ∗∗∗p < 0.001; **Figures 4D,E**). Importantly, this increase was significantly reduced in cultures exposed to GW1100, a specific FFAR1 antagonist, prior to PA treatment (PA vs. GW1100 + PA <sup>∗</sup>p < 0.05; **Figures 4D,E**). These data suggest that PA, among other factors, activates FFAR1 in N43/5 cells.

FIGURE 3 | Inhibition of autophagy reduces insulin sensitivity in N43/5 cells. (A) Representative western blots of the indicated proteins of N43/5 cells pre-incubated with BafA1 (100 nM) or its vehicle (DMSO) during 6 h and then stimulated with insulin for 3 min to evaluate IR phosphorylation (Tyr1150/1115) or 15 min to evaluate AKT phosphorylation. (B,C) Quantification of IR (Tyr1150/115) and AKT (Ser473) phosphorylation in cells exposed to the treatments as indicated in (A). (B,C) Data are presented as mean ± SEM. Two ways ANOVA followed by Sidak's multiple comparisons test. \*p < 0.05, \*\*p < 0.01, \*\*\*p < 0.001. n = 3. (D) Representative images and quantification (E) of 2-NBDG uptake in N43/5 cells pre-incubated with BafA1 (100 nM) or its vehicle (DMSO) during 6 h and then stimulated with insulin for 30 min. Size bar: 10µm. (E) Data are presented as mean ± SEM. One-way ANOVA followed by Holm-Sidak's multiple comparisons test. \*p < 0.05. n = 3. Representative blot of the indicated proteins of N43/5 cells transfected with siRNA against BECN1 (F) or ATG7 (I) followed by insulin or PBS treatment for 3 min to evaluate IR phosphorylation or 15 min to evaluate AKT phosphorylation, with its respective quantifications (G,H,K,L). (G,H,K,L) Data are presented as mean ± SEM. Two ways ANOVA followed by Sidak's multiple comparisons test. \*p < 0.05, \*\*p < 0.01. n = 3. (J) mRNA levels of Atg7 in N43/5 cells transfected with a siRNA to downregulate ATG7. As control condition, cells were incubated with Lipofectamine RNAiMAX reagent only (Mock). Data are presented as mean ± SEM. Unpaired t test. \*\*p < 0.01. n = 3.

absence of FFAR1 antagonist GW1100 (1µM). (E) Area under the curves (AUC) represented in (D). Data are shown as mean ± SEM. One-way ANOVA followed by

## FFAR1 Pharmacological Inhibition Restores Autophagy and Insulin Sensitivity

Holm-Sidak's multiple comparisons test. \*p < 0.05, \*\*\*p < 0.001. n = 4.

Our data indicate that PA inhibits the autophagic flux and activates FFAR1 (**Figures 1**, **4**); thus, we determined if the activation of FFAR1 by PA is mechanistically involved in the inhibition of the autophagic flux. We exposed N43/5 cells to the FFAR1 antagonist GW1100 prior to PA treatment. GW1100 pretreatment reduced the levels of LC3II in cells pre-exposed to PA (PA vs. GW1100+PA <sup>∗</sup>p < 0.05; **Figures 5A,B**). As we previously showed that PA inhibits the autophagic flux (**Figure 1**), these results suggest FFAR1 inhibition restores the autophagic flux in cells treated with PA (**Figure 5C**).

Interestingly, pharmacological inhibition of FFAR1 activity in cells exposed to PA also restored insulin sensitivity in N43/5 cells. Indeed, PA exposure reduced p-AKT levels when cells were stimulated with insulin (**Figures 2**, **5D,E**). However, p-AKT levels in cells treated with insulin were similar between control cells (treated with BSA) and cells pretreated with GW1100, prior to PA exposure (PA + Ins vs. GW1100 + PA + Ins <sup>∗</sup>p < 0.05; **Figures 5D,E**). Consistently, glucose uptake, significantly reduced by PA treatment, is restored by GW1100 pretreatment (**Figures 5F,G**). Again, no significant differences were seen in Glut4 mRNA levels following treatments (**Supplemental Figure 6**). These data indicate FFAR1 pharmacological inhibition promotes insulin sensitivity in N43/5 cells exposed to PA.

#### DISCUSSION

Here, we demonstrate that exposure to the SatFA PA inhibits the autophagic flux and reduces insulin sensitivity in N43/5 hypothalamic neurons (**Figures 1**, **2**). Furthermore, we show that inhibition of autophagy and the autophagic flux reduces insulin sensitivity in the same cellular model (**Figure 3**). Lastly, our data indicate that PA activates FFAR1 in hypothalamic neuronal cells (**Figure 4**) and that the inhibition of the autophagic flux and the reduction in insulin sensitivity are prevented by pharmacological inhibition of FFAR1 (**Figure 5**).

Chronic consumption of HFDs promotes the accumulation of PA in the hypothalamus of male mice (4–7); thus, we decided to determine whether this increase in PA might affect autophagy, which also appears dysfunctional in the hypothalamus following chronic consumption of HFDs. Importantly, previous studies indicate autophagy malfunction in hypothalamic POMC neurons contributes to obesity-associated metabolic dysfunctions, including increased plasma insulin

\*p < 0.05, \*\*p < 0.01. n = 3. (F) Representative images of 2-NBDG uptake in N43/5 cells incubated with vehicle (BSA), GW1100, PA and then stimulated with insulin 1 nM for 30 min, with its representative quantification (G). Size bar: 10µm. Data are shown as mean ± SEM. One-way ANOVA followed by Holm-Sidak's multiple comparisons test. \*p < 0.05. n = 3.

levels, hyperglycemia, glucose intolerance, impaired lipolysis, and leptin resistance (14–16). Taken together, these data suggest that HFDs-induced SatFAs accumulation in the CNS in male mice promotes a vicious cycle where autophagy inhibition might further enhance obesity-associated diseases. In this context, our data indicate that, at least in our cellular model, exposure to PA inhibits the autophagic flux, a condition that also affects insulin sensitivity. Future research, should confirm these data in in vivo models. Special attention should be paid to assess the possible sexual dimorphism is this response, especially considering previous studies indicating the sex-dependent function of POMC neurons in the regulation of energy homeostasis (40, 41).

The existence of a crosstalk between autophagy and insulin sensitivity has been previously suggested and identified in peripheral metabolic tissues (22, 23, 42). Interestingly, it was recently demonstrated that autophagy degrades insulincontaining vesicles in β-cells of autophagy-hyperactive mice, whereas in insulin-sensitive cells, autophagy enhances insulin response (42). Indeed, induction of autophagy, by different means, in skeletal muscle, hepatocytes, podocytes and adipocytes (43–45), stimulates insulin sensitivity; suggesting increased autophagy might be a general mechanism to boost insulin response. Despite this, there are no reports that determine if this crosstalk also occurs in the CNS, or specifically, in the hypothalamus where insulin sensitive neurons key in the regulation of food intake and peripheral glucose homeostasis reside. In the present study, we evaluated if inhibition of autophagy or autophagic flux blockade in a hypothalamic neuronal cell line affects insulin response. Our data indicate that this might be the case, as downregulation of different autophagy essential genes (Atg7 and Beclin1), as well as inhibition of the autophagic flux using a classic autophagy inhibitor (Bafilomycin A1), reduced the ability of the neuron to respond to insulin, as indicated by IR and p-AKT level and the reduction in glucose uptake following Bafilomycin A1 exposure. How this might be occurring has not been elucidated; however, a possibility is that, by inhibiting autophagy we prevent the degradation of negative regulators of the insulin signaling pathway, such as the protein

increase in AKT and IR phosphorylation levels stimulated by insulin.

phosphatase and tensin homolog (PTEN), which can be degraded by the autophagic/lysosomal pathway (46). In addition, based on these results, it is tempting to speculate that the maintenance of the autophagic balance in hypothalamic neurons might be key in the regulation of peripheral glucose homeostasis, as well as food intake. Future studies in neuron-specific transgenic mouse models need to be performed to confirm this hypothesis in vivo.

Sensing lipids is a mechanism through which the hypothalamus controls energy balance. Brain lipid sensing is important to control feeding behavior, hepatic glucose production, and insulin secretion (8). Fatty acids sensitive neurons can increase/decrease their activity in response to FAs (8). Different types of proteins can act as receptors of fatty acids, such as fatty acid transporters, fatty acid binding proteins, and fatty acid receptors. Among those, we focused our attention on FFAR1, the activation of which in pancreatic β-cells, leads to insulin secretion (25, 47), suggesting that this receptor might be involved in the chain of events that regulate the insulin response. Importantly, Steneberg et al. demonstrated that FFAR1 overexpression impairs β-cells function, promoting hypoinsulinemia and diabetes following HFDs feeding (26). In addition, in immortalized and primary pancreatic β-cells, FFAR1 activation induces lipotoxicity (48). Altogether these studies suggest that, in pancreatic β-cells, the effects of FFAR1 activation appear to be dual, since it is involved in both homeostatic and pathologic effects. In addition, FFAR1 overexpression causes hypoinsulinemia and autophagy hyperactivation reduces insulin release in β-cells, suggesting that an additional crosstalk might be occurring between FFAR1 and autophagy, specifically in pancreatic β-cells. Considerably less information is available regarding FFAR1 function in the CNS, where its role is still elusive. Our data, supported by previously published studies (37), show that, within the hypothalamus, FFAR1 is only expressed by neurons and not by astrocytes and microglial cells. Dragano et al. also confirmed FFAR1 is expressed in hypothalamic neurons (both in neuropeptide Y (NPY) and POMC neurons) and acts to maintain whole body energy homeostasis by decreasing energy efficiency and reducing hypothalamic inflammation when FFAR1 is chemically activated by receptor specific agonists (38). Additional studies demonstrate that polyunsaturated fatty acids (PUFAs)-mediate FFAR1 activation in the hypothalamus to produce an anti-inflammatory and neuroprotective effect (37, 49–52). This apparent discrepancy in the effects of FFAR1 is most likely attributable to a ligand-dependent activation of distinct allosteric sites on the receptor by the different molecular structures (53). Furthermore, differences between SatFAsmediated activation of the receptor in hypothalamic neurons in vitro vs. in vivo have not been investigated. In this study, we show PA-mediated FFAR1 activation in N43/5 hypothalamic neuronal cells reduces insulin sensitivity and the autophagic

#### REFERENCES

1. Hotamisligil GS. Inflammation, metaflammation and immunometabolic disorders. Nature. (2017) 542:177–85. doi: 10.1038/ nature21363

flux. Future research, using in vivo models, should evaluate if hypothalamic stimulation of this receptor may lead to the development of metabolic diseases, by affecting whole-body insulin sensitivity.

In summary, our study identifies a new role of autophagy in N43/5 hypothalamic neuronal cells in the regulation of insulin sensitivity. In addition, we identified PA as negative modulator of the autophagic flux in the same cell type. Finally, our data suggest FFAR1 mediates the ability of PA to reduce insulin sensitivity promoted by autophagic flux inhibition in hypothalamic neuronal cells (**Figure 6**).

These results may help understanding the cellular mechanisms that drive to insulin resistance induced by PA in hypothalamic neurons.

#### AUTHOR CONTRIBUTIONS

EM and MH-C conceived and planned the experiments. MH-C, LT-V, YÁ, FD-C, CN, JE-C, and DP-O carried out the experiments. EM took the lead in writing the manuscript. All the authors contributed to the interpretation of the results and provided critical feedback and helped shape the research, analysis, and manuscript.

#### FUNDING

This work was founded by: grants from Fondo Nacional de Desarrollo Cientifico y Tecnologico, FONDECYT 1160820 to EM, 1171075 to AC, 1150584 to CR and 3180209 to YÁ; by the CONICYT PIA 172066 to EM, RT, and AC; by CONICYT PhD fellowships 21151609 to MH-C, by CONICYT PCI REDI170147 to YÁ, by the International Centre for Genetic Engineering and Biotechnology, ICGEB (CRP/CHL16-06 to EM) and FONDECYT-VID: EN29/18 to RT.

## ACKNOWLEDGMENTS

The authors acknowledge the services provided by UC CINBIOT Animal Facility funded by PIA CONICYT∗ECM-07∗Program for Associative Research, of the Chilean National Council for Science and Technology. The authors thank Aaron P. Frank for critically reading and revising the manuscript. In addition, we thank everyone in the Morselli lab for constructive discussions and criticisms.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fendo. 2019.00176/full#supplementary-material

2. Milanski M, Degasperi G, Coope A, Morari J, Denis R, Cintra DE, et al. Saturated fatty acids produce an inflammatory response predominantly through the activation of TLR4 signaling in hypothalamus: implications for the pathogenesis of obesity. J Neurosci. (2009) 29:359–70. doi: 10.1523/JNEUROSCI.2760-08.2009


GPR40 as a novel pain regulatory system. Brain Res. (2012) 1432:74–83. doi: 10.1016/j.brainres.2011.11.012


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Hernández-Cáceres, Toledo-Valenzuela, Díaz-Castro, Ávalos, Burgos, Narro, Peña-Oyarzun, Espinoza-Caicedo, Cifuentes-Araneda, Navarro-Aguad, Riquelme, Troncoso, Criollo and Morselli. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Polycystin-2 Is Required for Starvation- and Rapamycin-Induced Atrophy in Myotubes

Catalina Kretschmar 1,2†, Daniel Peña-Oyarzun1,2†, Cecilia Hernando1,2 , Nadia Hernández-Moya1,2, Alfredo Molina-Berríos <sup>1</sup> , María Paz Hernández-Cáceres <sup>3</sup> , Sergio Lavandero2,4,5, Mauricio Budini 1,6, Eugenia Morselli 3,6, Valentina Parra2,4,6 , Rodrigo Troncoso2,6,7 \* and Alfredo Criollo1,2,6 \*

#### Edited by:

Maria Ines Vaccaro, University of Buenos Aires, Argentina

#### Reviewed by:

James Harper, Sam Houston State University, United States Michela Battistelli, University of Urbino Carlo Bo, Italy

#### \*Correspondence:

Rodrigo Troncoso rtroncoso@inta.uchile.cl Alfredo Criollo alcriollo@u.uchile.cl

†These authors have contributed equally to this work

#### Specialty section:

This article was submitted to Endocrinology of Aging, a section of the journal Frontiers in Endocrinology

Received: 15 September 2018 Accepted: 16 April 2019 Published: 08 May 2019

#### Citation:

Kretschmar C, Peña-Oyarzun D, Hernando C, Hernández-Moya N, Molina-Berríos A, Hernández-Cáceres MP, Lavandero S, Budini M, Morselli E, Parra V, Troncoso R and Criollo A (2019) Polycystin-2 Is Required for Starvation- and Rapamycin-Induced Atrophy in Myotubes. Front. Endocrinol. 10:280. doi: 10.3389/fendo.2019.00280 <sup>1</sup> Facultad de Odontología, Instituto de Investigación en Ciencias Odontológicas, Universidad de Chile, Santiago, Chile, <sup>2</sup> Facultad Ciencias Químicas y Farmacéuticas and Facultad Medicina, Advanced Center for Chronic Diseases, Universidad de Chile, Santiago, Chile, <sup>3</sup> Departamento de Fisiología, Facultad de Ciencias Biológicas, Pontificia Universidad Católica de Chile, Santiago, Chile, <sup>4</sup> Facultad de Medicina, Centro de Estudios en Ejercicio, Metabolismo y Cáncer, Universidad de Chile, Santiago, Chile, <sup>5</sup> Cardiology Division, Department of Internal Medicine, University of Texas Southwestern Medical Center, Dallas, TX, United States, <sup>6</sup> Autophagy Research Center, Santiago, Chile, <sup>7</sup> Laboratorio de Investigación en Nutrición y Actividad Física, Instituto de Nutrición y Tecnología de los Alimentos, Universidad de Chile, Santiago, Chile

Muscle atrophy involves a massive catabolism of intracellular components leading to a significant reduction in cellular and tissue volume. In this regard, autophagy, an intracellular mechanism that degrades proteins and organelles, has been implicated with muscle breakdown. Recently, it has shown that polycystin-2 (PC2), a membrane protein that belongs to the transient receptor potential (TRP) family, is required for the maintenance of cellular proteostasis, by regulating autophagy in several cell types. The role of PC2 in the control of atrophy and autophagy in skeletal muscle remains unknown. Here, we show that PC2 is required for the induction of atrophy in C2C12 myotubes caused by nutrient deprivation or rapamycin exposure. Consistently, overexpression of PC2 induces atrophy in C2C12 myotubes as indicated by decreasing of the myogenic proteins myogenin and caveolin-3. In addition, we show that inhibition of mTORC1, by starvation or rapamycin is inhibited in cells when PC2 is silenced. Importantly, even if PC2 regulates mTORC1, our results show that the regulation of atrophy by PC2 is independent of autophagy. This study provides novel evidence regarding the role of PC2 in skeletal muscle cell atrophy.

Keywords: atrophy, polycystin-2, myotubes, mTOR, starvation, rapamycin

## INTRODUCTION

Atrophy defined as a decrease in the mass and size of tissues or cells, is caused by a massive loss of proteins, cytoplasm, and organelles. Muscle cells respond to different pathophysiological stimuli by activating pathways involved in protein degradation. Stimuli such as cancer pharmacological treatments, AIDS, sepsis, heart failure, burn injury, and multiple sclerosis among others can induce severe muscle atrophy (1–4). The preservation of the homeostasis in muscle cells is crucial, not only because of the maintenance of an optimal muscle performance, but also because muscle represents an important source of amino acids and nutrients, which can be metabolized by

**86**

different organs such as brain, heart and liver (5). A severe or aggressive episode of atrophy can aggravate other colateral diseases, and seriously increase morbidity and mortality. Importantly, massive macroautophagy, hereafter referred as autophagy, is one of the mechanisms involved in muscle cell atrophy (6). Autophagy is a fundamental intra cellular process for degrading and recycling components such as proteins, organelles, and cytoplasm. Autophagy is characterized by the formation of autophagosomes, which fuse with the lysosome to form the autolysosome where the intravacuolar material is degraded. Studies in vitro and in vivo have shown that there is a strong relationship between autophagy and skeletal muscle atrophy (7– 9). Studies in C2C12 myotubes and a murine cancer model showed that activation of autophagy contributes to muscle wasting in cancer cachexia (10). Consistently, other studies have shown that the mechanistic target of rapamycin complex 1, mTOR, a constitutive kinase protein that inhibits autophagy, blocks atrophy in muscle and other cell types (11). In this regard, muscle-specific mTOR knockout mice present a severe muscle atrophy phenotype. Consistently, the insulin- or IGF-1-induced overactivation of mTOR blunts atrophy through the inhibition of autophagy in cardiac muscle (12, 13). Furthermore, inhibition of mTOR, induced by nutrient restriction or treatment with rapamycin (which inhibits mTORC1), causes autophagy and atrophy in skeletal muscle (11, 12, 14, 15). These evidences indicate the existence of a cross-talk between mTOR, autophagy and atrophy in the control of metabolism and cell and tissue size. Although different signaling pathways have been elucidated during the process of atrophy, the molecular mechanisms by which they can be modulated are still unknown. Recently, we have showed that the protein polycystin-2, PC2, a member of the transient receptor potential family which acts as a nonselective cation channel, is required for nutrient deprivationinduced autophagy in vivo and by hypertonicity, rapamycin and starvation in vitro in different types of cells (16, 17). Others studies also supported the role of PC2 in the regulation of autophagy. Indeed in renal epithelial cells and human embryonic stem cell-derived cardiomyocytes, PC2 is required for fluid flow- and glucose starvation-induced autophagy, respectively (18, 19).

PC2 regulates autophagy by activating different pathways; among those are listed classic (mTOR-dependent) and mTOR independent pathways, such as Ca2+-dependent modulation of autophagy. Interestingly, as PC2 is not an ubiquitous protein and since PC2 can regulate autophagy thanks to its Ca2<sup>+</sup> channel function, the modulation of autophagy by PC2 can be cell, tissue and stimuli dependent (16–19).

Here, we show that PC2 is required for starvation- and rapamycin- induced atrophy and inactivation of mTOR in C2C12 myotubes, without affecting the modulation of autophagy in the same cell type.

#### RESULTS

#### Classical Autophagic Inducers Causes Atrophy in C2C12 Myotubes

C2C12 myotubes were exposed to classical autophagy inducers such as nutrient deprivation, by treatment with Earle's Balanced Salt Solution (EBSS) or by pharmacological inhibition of mTORC1 with rapamycin, 0.1µM, at different time points (0– 6 h) (**Figure 1**). Protein levels of myogenic markers Myogenin (MYO) and Caveolin-3 (CAV3), which are down-regulated in different in vitro and in vivo models of atrophy, decrease in C2C12 myotubes following EBSS or rapamycin exposure (**Figures 1A–J**). In addition, as expected, both EBSS and rapamycin induce autophagy in C2C12 myotubes as assessed by the analysis of LC3 I to LC3 II conversion by western blotting, reaching the highest level at 0.5 and 1 h post-treatments (**Figures 1A,D,F,I**). We also analyzed the levels of p62/SQSTM1, a protein that binds specifically to LC3 and thus is degraded in the autolysosome (20). Consistently, the level of p62/SQSTM1 decreases when cells are exposed to EBSS or treated with rapamycin (**Figures 1A,E,F,J**), confirming that autophagy is upregulated in C2C12 myotubes following the aforementioned treatments. Altogether these data show that nutrient deprivation and rapamicyn not only induce autophagy but also atrophy in C2C12 myotubes.

#### PC2 Deficiency Prevents Atrophy in an Autophagy-Independent Manner in C2C12 Myotubes

PC2 is a member of the TRP channels protein family and studies have shown that PC2 is involved in mechanisms of mechanotransduction mostly in renal epithelial cells (21, 22). In addition, recently we and others have shown that PC2 regulates autophagy in different cell types (16, 17, 19). Indeed, we showed PC2 is required for hyperosmotic stress-induced autophagy in human cervical and colon cancer cell lines, HeLa and HCT116, respectively (16). Furthermore, it has been shown that PC2 is required for autophagy induction also in primary rat neonatal cardiomyocytes submitted to starvation or exposed to rapamycin (17), as well as in kidney mouse epithelial cells exposed to changes in fluid flow (18). In addition, animals knocked-out for PC2 in cardiomyocytes are resistant to starvation-induced autophagy, specifically in the heart (17). Given that PC2 regulates autophagy, and that autophagy is enhanced by action of different atrophy inducers in skeletal muscle cells, we evaluated if PC2 is required for atrophy in C2C12 myotubes exposed to nutrient deprivation or rapamycin. Our results showed that down regulation of PC2, by the use of specific siRNAs, prevented atrophy as indicated by the myotube diameter in C2C12 cells submitted to starvation (**Figures 2A,B**). In addition, western blot assays revealed that the decrease in the levels of MYO and CAV3 in cells exposed to nutrient deprivation or rapamycin for 4 and 6 h, respectively, is inhibited in C2C12 myotube cultures transfected with a specific siRNA against PC2 (siPC2) (**Figures 2C–H**). Despite the effects of PC2 in the regulation of atrophy, we did not observe inhibition of starvation- and rapamycin-induced autophagy,

**Abbreviations:** AU, arbitrary units; Baf A1, Bafilomycin A1; mTOR, mechanistic target of rapamycin; PC2, polycystin-2; PKD, polycystic kidney disease; VPS34, vacuolar protein sorting 34; 4EBP1, 4E-binding protein 1; ULK1, 51-like kinase 1 protein; Cav3, Caveolin-3; Myo, Myogenin; LC3, light chain.

(A–E) or treated with rapamycin 0.1µM (F–J) by 0, 0.5, 1, 2, 4, and 6 h. Subsequently, starvation- and rapamycin- induced atrophy were evaluated by western blot, by evaluating the levels of myo (A,B,F,G) and Cav3 (A,C,F,H). Autophagy was evaluated by western blot by assessing the levels of LC3 I and II (A,D,F,I). GAPDH was used as loading control. Representative gels are showed in (A) and (F) and gels quantifications are depicted in (B–E) and (G–J), respectively (mean± S.E.M., n = 3, \*p < 0.05, \*\*p < 0.01, and \*\*p < 0.001).

evaluated by the conversion of LC3 I to LC3 II, in cells downregulated for PC2 (**Supplementary Figure 1**). Altogether these results show that PC2 is required for starvation- and rapamycininduced atrophy in an autophagy-independent manner in C2C12 myotubes.

## PC2 Overexpression Induces Atrophy in C2C12 Myotubes

Our results showed that PC2 is required for starvation- and rapamycin-induced atrophy in C2C12 myotubes (**Figure 2**). However, if PC2 overexpression is sufficient to induce atrophy in

FIGURE 2 | PC2 is required for atrophy but not autophagy in C2C12 myotubes. PC2 was downregulated in C2C12 myotube cultures using a specific siRNA against PC2 (siPC2). An unrelated siRNA (siUNR) was used as control. Subsequently, 48 h post-transfection cells were subjected to nutrient deprivation by incubation in EBSS for 0 and 4 h (A–D) or treated with rapamycin 0.1µM for 0 and 6 h (F–H). (A,B) Cells were fixed and immunostaining against caveolin-3 (Cav3) was performed to evaluate myotube diameter by fluorescent microscopy. Nuclei were stained with 1µg/mL Hoechst 33342. Representative pictures are showed in (A) and myotube diameter quantification is represented in (B) (mean± S.E.M., n = 3, \*p < 0.05). (C,D) Whole lysates were resolved by western blot and Myogenin (Myo), Cav-3, LC3 I, and LC3 II levels were evaluated by the use of specific antibodies. GAPDH was used as a loading control. Representative gels are showed in (C,F) and gels quantifications are depicted in (E,D,H,G) (mean± S.E.M., n = 3, \*p < 0.05, \*\*p < 0.01).

C2C12 myotubes remains elusive. To this aim, we overexpressed PC2 by the use of the adenovirus, Ad PC2, or control adenovirus, Ad Co, and we evaluated atrophy in C2C12 myotubes. Our data showed that over expression of PC2 induced a reduction in myotubes diameter (**Figures 3A,B**), which correlated with a decrease in the levels of MYO and CAV3 (**Figures 3C–E**). Furthermore, PC2 overexpression did not induce autophagy, as evaluated by LC3 I to LC3 II turnover, indicating that regulation of autophagy is not involved as mechanism by which PC2 regulates atrophy in C2C12 myotubes (**Supplementary Figures 2A,B**). Given that C2C12 myotubes are a syncytium formed by the fusion of multiple cells, it is possible that they might be more resistant to autophagy. Thus, we also evaluated if PC2 induces autophagy in myoblasts. Our results showed that, similar to the results obtained in myotubes, overexpression of PC2 did not induce autophagy in myoblasts (**Supplementary Figures 2C,D**). These data, together with the results depicted in **Figure 2**, indicate that PC2 induces atrophy in C2C12 skeletal muscle cells by an autophagyindependent mechanism.

## PC2 Modulates mTOR Pathway in C2C12 Myotubes

It is known that the mTOR and AKT axis not only regulates hypertrophy, but also atrophy in skeletal muscle in vivo and in vitro (11, 15, 23, 24). Given that it is well-known that starvation affects mTOR (14) and that our results indicate PC2 is required for starvation-induced atrophy, we evaluated if PC2 modulates mTOR signaling in C2C12 myotubes. To this aim, myotubes were submitted to siRNA-mediated down regulation of PC2 or transfected with an unrelated siRNA, siUNR. Then, cells were subjected to starvation with EBSS medium at different time points. Our results show that down regulation of PC2 prevents the decrease in the phosphorylation of the downstream-mTOR proteins, S6, and 4EBP1, when atrophy is induced by starvation (**Figures 4A–C**). Another signaling pathway implicated in the induction of atrophy is the inactivation of AKT with the subsequent translocation from the cytoplasm to the nucleus of the forkhead box O (FoXO) proteins (25, 26). In this regard, our data showed that downregulation of PC2 does not regulate starvation-induced dephosphorylation of AKT on Ser473 (**Supplementary Figures 3A,B**). Altogether, these results suggest that PC2 regulates starvation-induced atrophy via mTORC1 in C2C12 myotubes, independently of AKT (**Figure 4A–C** and **Supplementary Figures 3A,B**).

## DISCUSSION

Severe and progressive muscle atrophy is observed in different human pathological conditions and several of the molecular mechanism that control atrophy remains elusive (1–4). Previous studies of our group demonstrated that PC2 positively regulates autophagy and that depending on the stimuli, cell type and tissue the modulation of mTOR is implicated in the process (16, 17). Furthermore, downregulation of PC2 inhibits autophagy induced by nutrient deprivation, rapamycin, and hypertonicity in different types of cells, while its overexpression triggers autophagy (16, 17). The requirement of PC2 for autophagy induction has also been demonstrated in vivo, in mouse heart tissues, where starvation-induced autophagy was inhibited in the heart mouse knocked-out for PC2 in cardiomyocytes (17). Given that massive autophagy has been observed in different models of atrophy and that there are no evidences about the role of PC2 in skeletal muscle cell models, we evaluated here if PC2 is regulating atrophy in C2C12 myotubes induced by classical stimuli of autophagy, such as nutrient deprivation or treatment with rapamycin. We observed that starvation and rapamycin induced atrophy in C2C12 myotubes and that PC2 is necessary in this process (**Figures 1**, **2**). Although, other studies have showed that PC2 has a role in the control of autophagy, specifically in human embryonic stem cell-derived cardiomyocytes, neonatal cardiomyocytes, and renal epithelial cells (16, 18, 19) we did not observe that PC2 regulates autophagy in C2C12 myotubes, as assessed by the conversion of LC3 I to LC3 II. Importantly, previous studies that identified a role for PC2 in the regulation of autophagy in renal epithelial cells showed that it is related with pathways activated by fluid flow-induced mechanical stress, where the role of primary cilium has been shown to be key. In this regard, we can speculate that the intracellular location of PC2 as well as the type of stimuli or condition that affects autophagy might differentially affect PC2 function and/or autophagy. Here, our data showed that modulation of atrophy by PC2 is autophagy-independent given that PC2 did not modulate rapamycin-induced LC3 I to LC3 II conversion (**Supplementary Figure 1**) and consistently, autophagy was not altered following PC2 over expression-induced atrophy conditions (**Figure 3** and **Supplementary Figure 2**).

mTOR has a pro-myogenic role given that it positively regulates protein synthesis (14). Indeed, skeletal muscle-specific ablation of Raptor, an adapter protein of the mTORC1, or deletion of the mTORC1 substrate S6 kinase, causes muscle dystrophy, and suppresses muscle growth adaptations to nutrient availability, respectively (27). Given the pro-myogenic relevance of mTOR, we evaluated if PC2 regulates atrophy by modulation of the mTOR pathway. Our results showed that downregulation of PC2 inhibited starvation-induced dephosphorylation of the protein S6, a key downstream target of mTOR, suggesting that regulation of atrophy by PC2 can be mediated by the modulation of mTOR signaling (**Figure 4**). Studies have shown that the role of mTOR not only has been related with the modulation of the size of the muscle fibers but its function has been also observed in other types of cells. Indeed, MDCK cells show a dramatically increase in cell size when the mTOR pathway is activated (28). Interestingly, our results showed that although the down regulation of PC2 attenuates the effects of starvation on the mTOR pathway, however, despite this it was unable to inhibit autophagy induced by rapamycin (**Supplementary Figure 1**), indicating that PC2, by modulating mTOR, and not autophagy, regulates atrophy in C2C12 myotubes.

To elucidate the mechanism by which PC2 regulates atrophy, we evaluated the phosphorylation status of AKT, which controls

the activation of the transcription factor FoXO1, a known regulator of atrophy-related genes expression. Even if the status of phosphorylation of AKT was sensitive to starvation, it was unaffected by PC2 downregulation, indicating that AKT pathway does not participate in the regulation of atrophy by PC2. In this regard, other studies have also shown events of atrophy independent of AKT/FOXO1 pathway, especially in glucocorticoid-induced muscle atrophy (29).

gels are showed in (C) and gels quantifications are depicted in (D,E) (mean ± S.E.M., n = 3, \*p < 0.05).

Regarding the function of PC2, it is a Ca2+-permeable ion channels with a relevant role in the maintenance of the cytosolic Ca2<sup>+</sup> (30). Indeed, mutations in PC2 lead to impaired calcium homeostasis in cardiac muscle which predispose cardiomyopathies in vivo (31). Thus, given that PC2 was required for starvation- and rapamycin-induced atrophy and that autophagy was not implicated in the mechanism by which PC2 regulates atrophy, it is possible that calcium microdomains controlled by PC2 may be also required to induce atrophy in C2C12. In fact, agents such as angiotensin II, tumor necrosis factor-α (TNFα) and lipopolysaccharide not only induce muscle atrophy but also a rise in Ca2<sup>+</sup> which is necessary for both proteolysis and decreasing in protein synthesis in muscle cells (29, 32–34). This work did not evaluate the role of Ca2+. However, it could be addressed in future studies.

In conclusion, all these findings reveal a novel role of PC2 on the regulation of atrophy, which is mediated by the modulation of mTOR in C2C12 myotubes.

## MATERIALS AND METHODS

#### Cell Culture and Treatments

C2C12 myoblasts were purchased in Sigma-Aldrich (Cat N◦ 91031101-1VL). Cells were grown in DMEM (glucose 4.5 g/L) containing L-glutamine, 110 mg/L sodium pyruvate, 10% FBS, and 10 mM HEPES. Differentiation of C2C12 myoblast to myotubes was performed by culturing cells in DMEM (glucose 4.5 g/L) containing L-glutamine, 110mg/L sodium pyruvate, 10 mM HEPES and 2% horse serum by 1 week. Media, supplements and reagents for cell culture were purchased from Gibco-Invitrogen (Carlsbad, USA). Cells were submitted to nutrient deprivation by culture of cells in Earle's Balanced Salt Solution medium, EBSS (Sigma-Aldrich, St. Louis, USA). Rapamycin and Bafilomycin A1 were purchased in Sigma-Aldrich. All experiments were independently repeated at least three times.

#### siRNA Transfection and Adenovirus Infection

siRNAs were purchased by Sigma-Aldrich Corporation. An unrelated-siRNA sequence was used as negative control. Lipofectamine iMax (Invitrogen) and Optimem culture medium were used for siRNA transfections. Thirty-six hours after transfection cells were stimulated. Protein quantification of the targeted protein was used to evaluate the efficiency of the different siRNAs. For adenovirus-mediated protein

overexpression, cells were incubated for 12 h with the AdPC2 adenovirus.

## Western Blot Analysis

3, \*\*p < 0.01, \*\*\*p < 0.001).

Protein samples of C2C12 myoblasts cells were prepared in M-PER lysis buffer (Thermo Scientific) supplemented with protease and phosphatase inhibitors (ROCHE). Aliquots of the extracted proteins (∼30 µg/lane) were resolved in 10 or 12% SDS-PAGE gels and then subjected to immunoblotting using antibodies specific for Myogenin (mouse monoclonal IgG clone F5D, cat. n◦ sc-12732; Santa Cruz Biotechnology), Caveolin-3 (Mouse monoclonal IgG clone 26, cat. n◦ 610420; BD Bioscience LaboratoriesTM), GAPDH (mouse monoclonal IgG, cat n◦ MAB274; Chemicon International), 4EBP1(rabbit polyclonal IgG cat n◦ 9452; Cell Signaling Technology), P-4EBP1 Thr37/46 (rabbit polyclonal IgG clone 236B4, cat n◦ 2855; Cell Signaling Technology), LC3 I and II (rabbit polyclonal IgG, cat n◦ 9748; Cell Signaling Technology), p62/SQSTM1 (rabbit polyclonal IgG, cat n◦ , NBP1-42822; Novus Biologicals), PC2 (rabbit polyclonal IgG clone H-280, cat. n◦ sc-25749; Santa Cruz Biotechnology), AKT/PKB (mouse monoclonal IgG cat n◦ 2966; Cell Signaling Technology), P-AKT/PKB Ser473 (rabbit polyclonal IgG cat n ◦ 4060; Cell Signaling Technology), S6 (mouse monoclonal IgG clone 54D2, cat n◦ 2317; Cell Signaling Technology), P-S6 Ser235/236 (rabbit polyclonal IgG clone 236B4, cat n◦ 2211; Cell Signaling Technology). Then, membranes were incubated with secondary goat anti-mouse or anti-rabbit IgG conjugated to horseradish peroxidase (SouthernBiotech, Birmingham, USA) prior to revelation by means of ECL Detection Kit (Amersham Pharmacia, Pittsburgh, USA). Gels were analyzed and quantified with the software ImageJ (http://rsb.info.nih.gov/ij/).

#### Confocal and Fluorescence Microscopy

Following treatments cells were washed twice with ice-cold PBS, fixed in paraformaldehyde (4% w/v) for 15 min, permeabilized with Triton 0.1%, PBS for 10 min and blocked in 3% BSA-PBS for 1 h. Nuclei were counterstained with Hoechst 33342 (1µg/mL) (Molecular Probes). Fluorescence and confocal fluorescence images were captured using an IRE2 microscope equipped with a DC300F camera (both from Leica Microsystems GmbH, Wetzlar, Germany) and an LSM 510 microscope (Carl Zeiss, Jena, Germany). Images were analyzed with the software ImageJ (http://rsb.info.nih.gov/ij/).

#### Myotube Diameter Measurement

Myoblasts were differentiated into myotubes by culturing cells in DMEM containing 2% horse serum for 1 week. Then, myotubes were used to evaluate the cell diameter. Briefly, images were obtained with an epifluorescent microscope (Nikon Eclipse TI) and analyzed by a transverse line across the myotube. Myotubes with more than three nuclei were used for diameter measurements. We draw the line of distance across the myotube, which represents the myotube diameter (µm). At least three diameters per myotube were measured and at least 100 myotubes per well were analyzed using ImageJ Software. Data is presented as fold changes relative to control levels.

#### RESULTS AND STATISTICAL ANALYSIS

Results are shown as mean ± S.E.M. from at least three independent experiments. Statistical analyses were performed using Student's t–test when analyzing two independent groups, one-way ANOVA for more than two independent groups and two-way ANOVA for two independent variables followed by a Sidak post-hoc test (GraphPad Software Inc.). P < 0.05 was considered to be statistically significant.

#### AUTHOR CONTRIBUTIONS

CK, DP-O, CH, NH-M, and AM-B performed the experiments. MH-C performed experiments and statistical analysis in the second round of revisions. SL and MB contributed to the experimental design and manuscript preparation. EM contributed to the experimental design and image analysis. VP performed mitochondrial morphology studies in the second round of revisions. RT and AC conceived the project and contributed to manuscript preparation.

#### FUNDING

Fondo Nacional de Desarrollo Científico y Tecnológico (FONDECYT); 1190743 to VP, 1160820 to EM, 1161123 to MB, and 1171075 to AC. Programa de Investigación Asociativa (PIA)-CONICYT 172066 to MB, EM, VP, RT, and AC. Fondo de

#### REFERENCES


Financiamiento de Centros de Investigación en Áreas Prioritarias (FONDAP); 15130011 to VP, RT, AC, and SL. Concurso Nacional de Inserción de Capital Humano Avanzado en la Academia 79150007 (PAI) to VP. Enlace FONDECYT-VID EN29/18 to RT. Comisión Nacional de Investigación Científica y Tecnológica (CONICYT); Ph.D. Fellowships program; 21140458 to DP-O and 21151609 to MH-C. Ph.D. Fellowships Programa de Doctorado en Ciencias Biomédicas, Facultad de Medicina, Universidad de Chile to CK. PEW Latin American Fellows Program in the Biomedical Science; 00002991 to AC. International Centre for Genetic Engineering and Biotechnology (ICGEB); CRP/CHL18- 04 to VP, CRP/CHL16-06 to EM and CRP/CH113-04RG to MB.

### ACKNOWLEDGMENTS

We sincerely thank everyone in the AC and RT laboratories for discussion and constructive criticism.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fendo. 2019.00280/full#supplementary-material

Supplementary Figure 1 | PC2 is not required for rapamycin-induced autophagy. PC2 was downregulated in C2C12 myotubes using a specific siRNA against PC2 (siPC2) (A,B). Unrelated siRNA (siUNR) was used as control. C2C12 myotubes downregulated for PC2 were treated with rapamycin 0.1µM (A–B) for 0 and 6 h. Whole lysates were resolved by western blot and Polycystin-2 (PC2), LC3 I and LC3II were evaluated. GAPDH was used as loading control. Representative gel is showed in (A) and the relative levels of LC3 II/LC3 I are depicted in the graph in (B) (mean ± S.E.M., n = 3, ∗∗p < 0.01, ∗∗∗p < 0.001).

Supplementary Figure 2 | Overexpression of PC2 does not regulate autophagy both in C2C12 myotubes and myoblasts. PC2 was overexpressed in C2C12 myotubes (A,B) or myoblast (C,D) for 24 h by using an adenovirus for PC2 (Ad PC2). Empty adenovirus was used as control (Ad Co). Whole lysates were resolved by western blot and Polycystin-2 (PC2), LC3 I, and LC3II were evaluated. GAPDH was used as loading control. Representative gels for myotubes and myoblasts are showed in (A) and (C), respectively. Relative levels of LC3 I to LC3 II turnover in myotubes and myoblasts are depicted in the graph in (C) and (D), respectively (mean ± S.E.M., n = 3).

Supplementary Figure 3 | PC2 does not regulate AKT/PBK pathway. PC2 was downregulated in C2C12 myotubes by transfection with a specific siRNA against PC2 (siPC2). Unrelated siRNA (siUNR) was used as control. Then, cultures were submitted to starvation with EBSS (A,B) for 0, 5, 10, 20, 30, and 40 min. C2C12 myotubes whole lysates were resolved by western blot and total AKT/PKB and P-AKT/PKB Ser473 levels were evaluated by the use of specific antibodies. GAPDH was used as loading control. Representative gels are showed in (A) and relative levels of P-AKT/AKT are depicted in the graphs in (B) (mean ± S.E.M., n = 3).


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Kretschmar, Peña-Oyarzun, Hernando, Hernández-Moya, Molina-Berríos, Hernández-Cáceres, Lavandero, Budini, Morselli, Parra, Troncoso and Criollo. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Autophagic and Proteasomal Mediated Removal of Mutant Androgen Receptor in Muscle Models of Spinal and Bulbar Muscular Atrophy

#### Edited by:

Maria Ines Vaccaro, University of Buenos Aires, Argentina

#### Reviewed by:

Haruhiko Banno, University of Massachusetts Medical School, United States Masahisa Katsuno, Nagoya University Hospital, Japan Heather Lynn Montie, Philadelphia College of Osteopathic Medicine, United States

#### \*Correspondence:

Angelo Poletti angelo.poletti@unimi.it

†These authors have contributed equally to this work

#### Specialty section:

This article was submitted to Endocrinology of Aging, a section of the journal Frontiers in Endocrinology

Received: 30 April 2019 Accepted: 05 August 2019 Published: 20 August 2019

#### Citation:

Cicardi ME, Cristofani R, Crippa V, Ferrari V, Tedesco B, Casarotto E, Chierichetti M, Galbiati M, Piccolella M, Messi E, Carra S, Pennuto M, Rusmini P and Poletti A (2019) Autophagic and Proteasomal Mediated Removal of Mutant Androgen Receptor in Muscle Models of Spinal and Bulbar Muscular Atrophy. Front. Endocrinol. 10:569. doi: 10.3389/fendo.2019.00569 Maria Elena Cicardi 1†, Riccardo Cristofani 1†, Valeria Crippa<sup>1</sup> , Veronica Ferrari <sup>1</sup> , Barbara Tedesco<sup>1</sup> , Elena Casarotto<sup>1</sup> , Marta Chierichetti <sup>1</sup> , Mariarita Galbiati <sup>1</sup> , Margherita Piccolella<sup>1</sup> , Elio Messi <sup>1</sup> , Serena Carra<sup>2</sup> , Maria Pennuto3,4,5, Paola Rusmini <sup>1</sup> and Angelo Poletti 1,6 \*

<sup>1</sup> Dipartimento di Scienze Farmacologiche e Biomolecolari (DiSFeB), Dipartimento di Eccellenza 2018-2022, Centro di Eccellenza sulle Malattie Neurodegenerative, Università degli Studi di Milano, Milan, Italy, <sup>2</sup> Dipartimento di Scienze Biomediche, Metaboliche e Neuroscienze, Centro Interdipartimentale di Neuroscienze e Neurotecnologie (CfNN), Università degli Studi di Modena e Reggio Emilia, Modena, Italy, <sup>3</sup> Department of Neurosciences, Neuromuscular Center, University of Padova, Padova, Italy, <sup>4</sup> Department of Biomedical Sciences, University of Padova, Padova, Italy, <sup>5</sup> Dulbecco Telethon Institute, Centre for Integrative Biology (CIBIO), University of Trento, Povo, Italy, <sup>6</sup> Centro InterUniversitario sulle Malattie Neurodegenerative, Università degli Studi di Firenze, Milan, Italy

Spinal and bulbar muscular atrophy (SBMA) is an X-linked motoneuron disease (MND) caused by a mutant androgen receptor (AR) containing an elongated polyglutamine (polyQ) tract. ARpolyQ toxicity is triggered by androgenic AR ligands, which induce aberrant conformations (misfolding) of the ARpolyQ protein that aggregates. Misfolded proteins perturb the protein quality control (PQC) system leading to cell dysfunction and death. Spinal cord motoneurons, dorsal root ganglia neurons and skeletal muscle cells are affected by ARpolyQ toxicity. Here, we found that, in stabilized skeletal myoblasts (s-myoblasts), ARpolyQ formed testosterone-inducible aggregates resistant to NP-40 solubilization; these aggregates did not affect s-myoblasts survival or viability. Both wild type AR and ARpolyQ were processed via proteasome, but ARpolyQ triggered (and it was also cleared via) autophagy. ARpolyQ reduced two pro-autophagic proteins expression (BAG3 and VCP), leading to decreased autophagic response in ARpolyQ s-myoblasts. Overexpression of two components of the chaperone assisted selective autophagy (CASA) complex (BAG3 and HSPB8), enhanced ARpolyQ clearance, while the treatment with the mTOR independent autophagy activator trehalose induced complete ARpolyQ degradation. Thus, trehalose has beneficial effects in SBMA skeletal muscle models even when autophagy is impaired, possibly by stimulating CASA to assist the removal of ARpolyQ misfolded species/aggregates.

Keywords: autophagy, chaperones, protein aggregation, androgen receptor, motoneuron disease

## INTRODUCTION

Spinal and bulbar muscular atrophy (SBMA) is an inherited X-linked motoneuron disease (MND) linked to a CAG triplet repeat expansion present in the exon 1 of the gene coding for the androgen receptor (AR) (1). Because of that, the AR protein carries an elongated polyglutamine (polyQ) tract in its N-terminus. In normal individuals, the polyQ tract is comprised between 9 and 37 Qs with an average value of 22, but SBMA patients have a polyQ tract longer than 38 Qs (ARpolyQ) with a maximum of 68 Qs observed so far in some patients affected by a pathology characterized by an unusual early onset (2, 3). The physiological role of the polyQ tract is still largely debated, but the region could act as a transcriptional regulatory domain (4– 6). Other eight totally unrelated proteins presenting expanded CAG/polyQ repeats have been involved in neurodegenerative diseases (CAG/polyQ diseases) (7). Thus, the polyQ expansion is likely to confer a gain of neurotoxic function(s) to these mutant proteins. Indeed, the polyQ tract induces the acquisition of aberrant protein conformation (misfolding) to the host proteins making them prone to aggregate. Misfolded proteins affect the protein quality control (PQC) system functioning and, in SBMA, this event occurs in the cells expressing high levels of ARpolyQ. In fact, AR is abundantly expressed in motoneurons located in the anterior horns of the spinal cord and in the brain stem, as well as in sensory neurons of the dorsal root ganglia. These neurons degenerate in SBMA leading to atrophy of bulbar, facial and limb muscles, and in sensory function alterations (8– 14). Also, non-neuronal cells, like the motoneuron-controlled skeletal muscle cells, are directly affected by mutant protein toxicity. Indeed, even if originally classified as a typical MND, due to the relevant involvement of muscle tissue, SBMA has been reclassified as neuromuscular disease (15–26). The involvement of the muscle cells is rather complex. The atrophy of muscle cells may result from the loss of innervation arising from affected motoneurons and/or may be a direct consequence of ARpolyQ proteotoxicity on skeletal muscle cells. In fact, like spinal cord motoneurons and dorsal root ganglia neurons, also skeletal muscle cells are post-mitotic cells highly sensitive to the presence of misfolded species of ARpolyQ (27). Several evidences obtained initially in SBMA animal models support the direct involvement of muscle in the pathogenesis of SBMA, since the inhibition of AR production selectively in muscle correlates with an amelioration of the phenotype in mice; this notion has been proposed to be valid also in SBMA patients (24, 28, 29). Even human wild type AR (wtAR) overexpressed in mouse skeletal muscle induces several alterations normally observed in SBMA (16, 18, 30, 31). In addition, the downregulation of ARpolyQ levels specifically in skeletal muscle, by mean of antisense oligonucleotides (ASOs), results in prolonged survival in different SBMA mouse models, proving ARpolyQ direct action on muscle (25, 26, 32, 33); also, the restricted overexpression of ARpolyQ in muscle cells determines a delay of the SBMA onset in mouse models. Moreover, muscle samples from SBMA patients show dysregulation of several important pathways such as mitochondrial turnover, or the neuromuscular transmission at birth with an increased expression of the neonatal isoform of acetylcholine receptor (34). A major aspect of SBMA is that castration completely rescues SBMA phenotype in male mice, ascribing SBMA onset to testosterone activation of ARpolyQ (35–38), even if some early symptoms could appear in an androgen-independent manner (39). Muscle is the typical direct target of the anabolic androgenic activity of the AR (40, 41), and, thus, testosterone-triggered ARpolyQ toxicity may sensitize skeletal muscle cells to "toxic" ARpolyQ conformations, which cause ARpolyQ aggregation. In addition, testosterone induces the translocation of misfolded ARpolyQ into the nucleus where the protein exerts most of its toxicity (42, 43). These aggregates may not be toxic per se (44), but their presence in cell environment can lead to many cellular dysfunctions. However, misfolded ARpolyQ are likely to be formed soon after the release from HSPs which occurs in the cell cytoplasm, and there is the possibility to clear them as soon as they are formed prior to their migration into the cell nuclei. A typical cytoplasmic degradative process which may prevent misfolded ARpolyQ accumulation, or aberrant nuclear migration, is autophagy. Unfortunately, in the cytoplasm the ARpolyQ protein may block the autophagic flux due to misfolded proteins overload (45–51). Autophagy is considered one of the most important degradative system in cells, since its impairment in neurons leads to their death (52, 53). Autophagy is based on the formation of autophagosomes that entrap the waste material which will be then degraded when autophagosomes fuse with lysosomes (54). Indeed, by using trehalose, a well-known activator of the autophagy master regulator transcription factor EB (TFEB) (55, 56), to restore a normal autophagic flux in SBMA neuronal models, we found an improved clearance of misfolded ARpolyQ and the prevention of its aggregation (49, 51, 56), particularly in motoneuron (57–59).

The importance of a functional autophagy flux in SBMA is also sustained by several studies performed in animal and cell models of SBMA (50, 60, 61). In particular, autophagy is dysregulated in muscles of AR113Q knock-in SBMA mice (19, 22, 26), and this dysregulation includes alteration of TFEB, and its physiological antagonist ZKSCAN3 (22), as well as TFEB-target genes (coding for LC3, VPS11, VPS18 and LAMP1), both in mice and in patients (22). Notably, the inhibition of BECN1/Beclin1 mediated autophagy activation in AR113Q knock-in SBMA mice reduces skeletal muscle atrophy, extends survival and improves the phenotype, while over-activation of autophagy worsens phenotype (19). Thus, a crucial point when considering autophagy is that its levels of activation must be finely tuned, and, thus, any autophagy stimulator must be able to prevent accumulation of harmful material preserving cell's functionality. In this scenario, autophagic clearance of ARpolyQ in skeletal muscle, and how this is related to alternative degradative systems could have a high relevance. However, autophagy works in conjunction with the ubiquitin-proteasome system (UPS) in the removal of misfolded ARpolyQ, and its aggregated forms.

Interestingly, skeletal muscles of SBMA mice also display a high activation of Tgfb1, Ppargc1a, Pax7, Myog, E2-ubiquitin ligase Ube2q1, but not of Myod, and of two E3-ligases (Trim63/Murf-1 and Cul3). We found that the skeletal muscle of SBMA mice are characterized by a dramatic perturbation of several components of the autophagic pathways (Becn-1,

Atg10, Sqstm1/p62, Lc3), particularly those involved in the peculiar autophagic process now recognized as chaperoneassisted selective autophagy (CASA) (56, 61–72), like the CASA complex components: the small heat shock protein (HSP) B8 and BAG3, which in cooperation with the co-chaperone BAG1 control the correct routing of misfolded proteins to clearance (61). The Hspb8, Bag3 and Bag1 gene are all iper-induced in skeletal muscle of SBMA mice, and the Bag3:Bag1 ratio is increased in these muscles (73). Of note, the equilibrium between UPS and autophagy is critical to maintain the regular misfolded ARpolyQ clearance in SBMA (61). The molecular players regulating the equilibrium that re-routes substrates to UPS or autophagy are BAG1, which mediates UPS clearance of clients, and BAG3 which controls autophagic clearance of clients (46, 48, 49, 61, 68, 72, 74). BAG3 interacts (in a 2:1 ratio) with HSPB8, and the complex reduces ARpolyQ aggregation, by enhancing its solubility and clearance acting as an autophagy facilitator (49, 61). In this process HSPB8/BAG3 complex needs to interact with HSC70/CHIP dimer and the client misfolded protein, allowing its ubiquitination for SQSTM1/p62-mediated insertion into autophagosomes (63, 65). Only few studies aimed to unravel the involvement of the HSPB8-BAG3 and BAG1 systems in SBMA skeletal muscle (73), but the identification of specific autophagy related molecular target might represent a therapeutic valuable strategy for counteracting ARpolyQ toxicity (73). Notably, both HSPB8 and BAG3 mutations have been linked to neuromuscular disorders suggesting that they may be deeply involved in the regulation and in the control of the proteotoxic response of muscle cells (70, 71, 75–79).

For all these reasons, in this study, we have provided an extensive characterization of the autophagic activation, the role of the CASA complex and the HSPB8/BAG3 machinery as well as of the BAG1 co-chaperone in the PQC system response in a SBMA muscle cellular model.

#### MATERIALS AND METHODS

#### Chemicals

Testosterone; Z-Leu-Leu-Leu-al or MG132; Bafilomycin A1 from Streptomyces griseus; D-(+)-Trehalose dihydrate were all obtained from Sigma-Aldrich (St. Louis, MO, USA).

Cell Cultures, Treatments, and Transfection

Immortalized mouse myoblast C2C12, stably transfected, respectively, with cDNA encoding the human full length wt AR (with 24 Qs = ARQ24), or the mutant AR (with elongated polyQ of 100 Qs = ARQ100), were obtained by infection with the Lentiviral vector #945.PCCL.sin.cPPT.SV40ployA.eGFP.minCMV.hPGK.deltaLN -GFR containing the human cDNA encoding the ARQ24 or the ARQ100 (s-myoblasts) (80). After transfection, cells were sorted using the GFP fluorescence to identify positive cells. Sorted cells of both lines were cultured with DMEM high glucose medium (Euroclone, Pero, MI, Italy) supplemented with 1 mM glutamine (Euroclone), (30µg/mL) penicillin [SERVA, Electrophoresis GmbH, Heidelberg, Germany (64µg/mL)] streptomycin (SERVA), and 10% charcoal-stripped fetal bovine serum (CS-FBS) (GIBCO, Thermo Scientific Life Sciences Research, Waltham, MA, USA), to deplete hormones contained in the serum. Basal C2C12 cells were grown in medium containing unstripped serum. Testosterone was added in presence of CS-FBS. Cells were regularly maintained at 37◦C, with 5% CO2, and propagated after trypsin (Euroclone) dissociation as previously described (81). Cells were treated with testosterone (10 nM) for 48 h (ethanol was used as control); MG132 (10µM) for 16 h (DMSO was used as control); Bafilomycin A1 from Streptomyces griseus (100µM) for 16 h (DMSO was used as control); D-(+)-Trehalose dihydrate (100 mM) for 48 h (diluted directly in the culture medium), as detailed in figure legends.

Lipofectamine <sup>R</sup> 2000 Transfection Reagent (Thermo Scientific Life Sciences Research) was used to transfect cells, using 2 µL for transfecting 1 µg of DNA. 12-well plates were transfected with 1 µg of DNA, while 24-well plate were transfected with 0.5 µg of DNA. After 5 h, medium was replaced.

The following plasmids were used: p5HBhARQ112 (kindly provided by Dr. A.P. Lieberman, University of Michigan, Ann Harbor) here referred as ARQ112; pARQ161HA, pARQ1121HA (kindly provided by Dr. Diane Merry, Thomas Jefferson University, Philadelphia); pCI-HSPB8 encoding human HSPB8, pCI-neo-6xHisBAG3 encoding the full-length form of human BAG3 and pCDNA/HA-BAG1 encoding the human BAG1, were all kindly provided by Prof. H. H. Kampinga (Groeningen University, Groeningen, The Netherlands); pEGFP-N1 (Clontech-Takara Bio, Saint-Germain-en-Laye, France) was utilized to determine transfection efficiency.

## PBS and NP-40 Protein Extraction

PBS extracts: cells were plated in 12-well plate at a density of 65,000 cells/well, and the day after plating, cells were transfected and/or treated. At the end of the experiment, cells were harvested, centrifuged (100 x g; 6 min; 4◦C), and diluted in 60µL of PBS (Euroclone) added of protease inhibitor cocktail (Sigma-Aldrich), containing individual components including AEBSF at 104 mM, Aprotinin at 80µM, Bestatin at 4 mM, E-64 at 1.4 mM, Leupeptin at 2 mM and Pepstatin A at 1.5 mM. After slight sonication using Bandelin Sonoplus Ultrasonic Homogenizers –HD 2070, protein content of each sample was quantified by bicinchoninic acid (BCA) assay (Euroclone).

NP-40 extracts: cells were plated in 6-well plate at a density of 130,000 cells/well. After treatments, cells were harvested, centrifuged (100 x g; 6 min; 4◦C), and diluted in 65 µL in NP-40 extraction buffer (composition: 150 mM NaCl (Sigma-Aldrich); 20 mM TrisBase (Sigma-Aldrich); 0.5% Nonidet P-40 (NP-40) (Sigma-Aldrich); 1,5 mM MgCl2 (Sigma-Aldrich); 3% Glycerol (Sigma-Aldrich), pH 7.4), added of protease inhibitors [complete EDTA-free Tablet 25X (Sigma-Aldrich)], and 1 mM 1,4-Dithiotreitol (Sigma-Aldrich). Cells were lysed by passage in syringe (27 gauges). Samples were then centrifuged (16,000 x g; 15 min; 4◦C). Supernatants were transferred in new tubes, and the pellets were rinsed in 65 µL of NP-40 extraction buffer. Protein content of the NP-40 soluble fraction was quantified by BCA assay (Euroclone). The insoluble fraction was sonicated following the same protocol described above.

#### Filter Retardation Assay

Filter retardation assay (FRA) was performed using Bio-Dot SF Microfiltration Apparatus (Bio-Rad, Hercules, CA, USA). Six micrograms of both PBS and NP-40 soluble extracts were loaded on a cellulose acetate membrane with pores of 0.22µm. For NP-40 insoluble extracts, the amount to be loaded was calculated as equal volume to NP-40 soluble extracts. After loading, the samples onto the cellulose acetate membrane, vacuum was applied at the apparatus and protein suspension was filtered. Proteins were fixed at the membrane using a 20% methanol solution, and the membrane was incubated for 1 h at RT in blocking solution [5% non-fat dried milk (Euroclone) in TBS-T 1X]. The membrane was then incubated with rabbit polyclonal anti-AR antibody (AR-H280, Santa-Cruz, sc-13162; dilution 1:1,000 in blocking solution) for at least 2 h at RT. After two washes with 1X TBS-T, the membrane was incubated for 1 h at RT with goat anti-rabbit HRP-conjugate secondary antibody (Santa Cruz Biotechnology, sc-2004; dilution 1:5,000 in 1X TBS-T). After three washes in 1X TBS-T signal was revealed with ClarityTM Western ECL Blotting Substrate (Bio-Rad) and optical density was acquired by ChemiDoc XRS System (Bio-Rad). Results were analyzed using Prism 5.0. Sample variations were related alternatively to ARQ24 (EtOH) or ARQ100 (EtOH). Statistical differences were obtained applying the twoway ANOVA test followed by Bonferroni post-hoc test. Each experiment was replicated three times, and each bar represents mean ± SEM of three independent biological replicates.

#### Western Blot Analysis

Western blot experiments were performed using 10% polyacrylamide gels. To visualize AR protein, 15 µg of each PBS extract or 30 µg of each NP-40 soluble and insoluble extracts were loaded on gels. After electrophoresis, proteins were transferred over night at 4◦C on nitrocellulose membrane (Bio-Rad). Membrane was then incubated 1 h at RT in blocking solution, and then overnight at 4◦C with primary antibody diluted in blocking solution (5% dried non-fat milk (Euroclone) in 1X T-BST). After two washes with 1X TBS-T, the membrane was incubated 1 h at RT with secondary antibody diluted in 1X TBS-T. Signal was revealed using ClarityTM Western ECL Blotting Substrate (Bio-Rad) and images were acquired by ChemiDoc XRS System (Bio-Rad) as described for FRA. The following primary antibody were used: rabbit polyclonal AR-H280 antibody (Santa-Cruz Biotechnology, sc-13162; dilution 1:1,000) rabbit polyclonal anti-LC3-B antibody (Sigma-Aldrich, L8918; dilution 1:1,000), rabbit polyclonal anti-p62/SQSTM1 antibody (Abcam, Cambridge, UK, ab91526; dilution 1:3,000), home-made rabbit polyclonal anti-HSPB8 (kindly provided by Dr. Landry, Centre of Recherche Cancerologie, University of Laval, Canada; dilution 1:2,000), rabbit polyclonal anti-GAPDH (Santa Cruz Biotechnology, sc-32233; dilution 1:1,000), goat polyclonal anti-ACTIN (Santa Cruz Biotechnology, sc1615; dilution 1:1,000), mouse monoclonal anti-α-TUBULIN (Sigma-Aldrich, T6199; dilution 1:3,000). The following secondary antibodies were used: goat anti-rabbit HRP-conjugate secondary antibody (Santa Cruz Biotechnology, sc-2004; 1:10,000), goat anti-mouse HRP-conjugate secondary antibody (Santa Cruz Biotechnology, sc-2005; 1:10,000), donkey anti-goat HRPconjugate secondary antibody (Santa Cruz Biotechnology, sc-2020; 1:10,000).

#### Immunostaining and Confocal Microscope Analysis

Cells were seeded on coverslips at a density of 25,000 cells/well (in 24-well plate), and the day after plating were transfected and/or treated. After treatments, cells were fixed at 37◦C for 25 min using a solution 1:1 of 4% paraformaldehyde (Sigma-Aldrich) in PB 0.2 M [a solution made of KH2PO4 (0.06M) and Na2HPO4 (0,26M)] and 4% sucrose (Sigma-Aldrich) in PB 0.2 M. Then, fixing solution was removed and iced methanol was added for 10 min to complete the fixation. Cell permeabilization was performed using a solution of 0.2% TRITON X100 (Sigma-Aldrich) followed by incubation for 1 h in blocking solution (5% dried non-fat milk in 1X T-BST). Incubation with the primary antibody was kept o/n at 4◦C. Incubation with the fluorescenttagged secondary antibody was preceded by three washes with PBS, to remove the excess of primary antibody. Nuclei were stained with DAPI (Sigma-Aldrich). The following primary antibodies were used: rabbit polyclonal AR-H280 antibody (Santa-Cruz Biotechnology, sc-13162; dilution 1:500), rabbit polyclonal anti-LC3 antibody (Sigma-Aldrich, L8918; dilution 1:500), rabbit polyclonal anti-p62/SQSTM1 antibody (Abcam, ab91526; dilution 1:500). The following secondary antibodies were used: goat anti-rabbit Alexa 594 (Life technologies, Thermo Scientific, A-11012; dilution 1:1,000). All the primary and secondary antibodies were diluted in blocking solution. Coverslips were mounted on a glass support using MOWIOL and images were acquired using an Axiovert 200 microscope (Zeiss Instr., Oberkochen, Germany) combined with a Photometric Cool-Snap CCD camera (Ropper Scientific, Trenton, NJ, USA) or using Eclipse Ti2 (Nikon, Netherlands) confocal microscope equipped with A1 plus camera (Nikon) and processed with the NIS-Elements software (Nikon) or using LSM510 Meta system confocal microscope (Zeiss, Oberkochen, Germany) and processed with the Aim 4.2 software (Zeiss).

## Real Time PCR

Cells were plated in 6-well plate at a density of 130,000 cells/well, and the day after plating were transfected and/or treated. At the end of the experiment, cells were harvested, centrifuged (100 x g; 6 min; 4◦C) and lysed using TRI Reagent (Sigma-Aldrich). RNA was extracted following manufacturer instructions and quantified using NanoDrop 2000 spectrophotometer (Thermo Scientific). After DNA removal using DNase I (Sigma-Aldrich), 0.5 µg of the total mRNA was reverse-transcribed using High-Capacity cDNA Archive Kit (Thermo Scientific). mRNA levels were assayed using iTaq SYBR Green Supermix (Bio-Rad) on CFX 96 Real-Time System (Bio-Rad). All results were normalized to RplP0 used as control. All the primers used were obtained by Eurofins Genomics, sequences of primers have been previously reported (73). The following primers were newly designed: Vcp FW- 5 ′ -TGCCATCCTAAAAGCCAATC-3′ RV-5 ′ -TCAGCTCCAGAAAAGCCATT-3′ .

#### Statistical Analysis

Statistical analysis has been performed by using Student's ttest to compare two groups and analysis of variance (ANOVA) to compare three or more groups. Two-Way ANOVA was used to compare the effect of two independent variables. Analyses were performed with the PRISM (version 5) software (GraphPad Software).

#### RESULTS

#### AR Aggregation in Muscle Cells

Here we used immortalized C2C12 myoblasts, that are widely used as model to mimic muscle cells in culture; this cell line has been infected with viral vectors expressing ARQ24 or ARQ100, subcloned and stabilized in culture (s-myoblast). We initially performed a characterization of the AR biochemical properties in s-myoblasts to assure that the viral expression of this protein was retained even after several passages in culture. Immunofluorescence (IF) analysis showed that ARQ24 and ARQ100 have similar fluorescence intensity, and are both localized in the cell cytoplasm in basal conditions; as expected, upon testosterone treatment they both translocated into the nucleus (**Figure 1A**) and signal intensity also increased upon testosterone treatment. No visible aggregates or inclusions were seen by IF in C2C12-ARQ100 cell line, even after testosterone treatment. Western blot (WB) correctly showed ARQ24 with a higher SDS-PAGE motility than ARQ100, because of the presence of the polyQ tract of different length which results in different molecular weights (MW) of the two AR proteins. Moreover, both ARQ24 and ARQ100 expression was stabilized by testosterone treatment which also induced a mild upshift of the band possibly linked to AR phosphorylation during activation process (27, 82) (**Figure 1B**, upper panel). No high MW (HMW) forms were observed in the stacking gel in all the tested conditions (not shown), suggesting that the AR does not form SDS-resistant insoluble species in s-myoblasts. A low intensity band, possibly related to ARpolyQ fragmentation (or to the endogenous mouse AR) appeared to be mildly increased in ARpolyQ testosterone-treated samples. Interestingly, filter retardation assay (FRA) showed that, after testosterone exposure, ARQ100 formed aggregated species that can be retained on cellulose acetate membrane (with size exclusion of 0.22µm) ( ∗∗∗p < 0.001 vs. testosterone-treated C2C12-ARQ24; ∗∗p < 0.01 vs. untreated C2C12-ARQ100) (**Figure 1B**, lower panel). In this analysis, ARQ24 immunoreactivity was very low even after its activation with testosterone. To better characterize the ARQ100 aggregated species identified in FRA (but not visible in IF), we performed a detergent fractionation assay using NP-40 extraction on cell lysates. In WB, we found that large amounts of ARQ24 and ARQ100 were present in the NP-40 soluble fraction of testosterone activated ARs samples, which were considerably higher than those found for the corresponding untreated ARs samples (**Figure 1C**, upper panel). This confirmed testosterone stabilization of AR protein (83). Of note, in the NP-40 insoluble extracts we found a much more abundant amounts of testosterone-treated ARQ100 compared to testosteronetreated ARQ24, and to untreated controls (**Figure 1D**, upper panel). Using FRA analysis, we found that testosterone treatment triggered the formation of NP-40 soluble, and NP-40 insoluble aggregates retained on the cellulose acetate membrane of ARQ100, while these species were not formed by ARQ24 (∗∗p < 0.01 vs. ARQ24 cell line) (**Figures 1C,D**, lower panels). Despite these data, we found no differences in cell viability, or cell survival in cells expressing ARQ24 or ARQ100, even after testosterone treatment (data not shown), suggesting that s-myoblasts are not sensitive to ARpolyQ toxicity.

Collectively these data suggest that testosterone induces the formation of ARQ100 aggregates detectable in FRA. These species are present both in PBS extracts and in NP-40 soluble, and insoluble extracts. Surprisingly, no aggregates were observed in IF. It might be possible that their size is lower than the detection sensitivity of IF as in the case of small oligomeric species.

### The Impact of the Modulation of the Protein Quality Control System on AR Aggregation in Muscle Cells

We next investigated which degradative pathway is specifically responsible for ARQ24 and ARQ100 degradation in s-myoblasts by inhibiting the UPS or autophagy, using MG132 or bafilomycin A1, respectively. Proteasome inhibition resulted in an increase of the accumulation of the total amounts of ARQ24 in FRA, which is normally processed via this degradative pathway (47) (**Figure 2A**, lower inset), showing that a high concentration of wtAR inside cells (associated to its impaired clearance) could lead to its accumulation in HMW species. In s-myoblasts, proteasome inhibition resulted in a dramatic increase of the accumulation of mutant ARQ100 in FRA independently from its activation, as we already reported for immortalized motoneurons (47). We performed detergent fractionation assay, and we found no difference in the levels of ARQ24 species after proteasome inhibition (**Figures 2B,C**), suggesting a variability in the response of normal (ARQ24) cells to UPS inhibition. Conversely, both NP-40 soluble and insoluble testosteroneinduced ARQ100 aggregates, which are retained on cellulose acetate membrane, were increased after proteasome inhibition (**Figures 2B,C**, lower insets).

With regards to autophagy, we found no involvement of this pathway in the clearance of the wtAR (ARQ24) in s-myoblasts, while the perturbation of autophagosome and lysosome fusion with bafilomycin A1 resulted in a robust increase of PBS soluble form of ARQ100 in presence of testosterone (**Figure 2A**) in FRA. Bafilomycin A1-mediated inhibition of autophagy resulted also in a dramatic increase of both ARQ100 NP-40 soluble and insoluble species, independently from testosterone treatment (**Figures 2B,C**).

These data suggest that in s-myoblasts proteasome is the main mediator of the clearance of both wt and mutant AR, while autophagy appears to be predominantly involved in the clearance of the mutant ARpolyQ.

Next, we evaluated whether the presence and activation of mutant ARpolyQ have an impact on the expression of genes involved in the PQC system. We found no variation in the expression of Tfeb, Becn1, Bag1, Hspb8, Sqstm1/p62, Lc3 in all

conditions tested (**Figure 3A**). We found that the expression of mutant ARpolyQ correlated with a reduction in the expression of Bag3 and Vcp (another autophagy associated proteins found to be involved in motoneuron diseases), but these changes were not linked to the presence of testosterone (**Figure 3A**).

We then analyzed whether the levels of ARpolyQ species entrapped in FRA could be modulated by the overexpression of components required to route misfolded proteins to either UPS or autophagy. The data shown in **Figure 3B** indicate that the formation of testosterone-induced aggregated species of mutant ARpolyQ in transiently transfected C2C12 (ARQ112 HMW aggregates) can be counteracted by the overexpression of HSPB8 and BAG3. These two proteins are essential components of the CASA complex, which delivers misfolded proteins to the microtubule organization center where aggresomes are formed before their engulfment into nascent autophagosomes. Notably, both HSPB8 alone, and BAG3 alone preserve their pro-autophagic activity even if the CASA complex required both proteins in association with HSP70 and CHIP. This suggests that may be both considered limiting factor for the CASA complex activity. Interestingly, also the overexpression of BAG1 resulted in a great reduction of the accumulation of testosterone-induced aggregated species of mutant ARpolyQ measured in FRA (**Figure 3B**). It must be noted that BAG1 exerts its activity by preventing HSP70 and CHIP to become part of the CASA complex (61, 84–86), thus routing misfolded proteins to UPS degradation as an alternative to autophagy. Since, it has been demonstrated that testosterone induces the formation of ARpolyQ aggregates via the generation of a N-terminal caspase-3 cleaved fragment containing the polyQ stretch, which is highly prone to aggregate, we wanted to test whether the routing system may also be involved in the removal of this highly neurotoxic AR species. The results (**Figure 3C**) clearly demonstrated that both the overexpression of HSPB8 and BAG3, as well as that of BAG1, are capable to revert the accumulation in FRA of HMW aggregates of a highly

FIGURE 2 | bafilomycin A1. Optical densitometry quantification of FRA (lower inset). (\*\*\*p < 0.001 vs. relative control conditions –T/+T; \*p < 0.05 vs. ARQ24 -T). (C) FRA (upper inset) of NP-40 insoluble extracts of cells treated with testosterone, MG132 and bafilomycin A1. Optical densitometry quantification of FRA (lower inset) (\*\*\*p < 0.001 vs. relative control conditions +T). For each panel, FRA images derive from the same membranes with identical exposure time to permit direct comparison of wtAR and ARpolyQ levels. Two-way ANOVA followed by Bonferroni post-hoc test was used. Each experiment was independently replicated three times. Graphs show quantification of three independent biological samples (n = 3).

neurotoxic caspase-3 released N-terminal fragment of ARpolyQ ARQ1121HA (87–90).

Collectively, these data suggest that, by modulating specific components of the PQC system, the ARpolyQ and its highly neurotoxic aggregate-prone caspase-3 released fragment can be eliminated from muscle cells using both the proteasome and the autophagy system, when they are still normal and functioning as in our cell line [ARQ100 does not greatly affect proteasome and autophagy machinery (**Figure 3A**)].

## Pharmacological Induction of the Autophagic System Reduces ARpolyQ Accumulation and Aggregation

Based on these data, we hypothesized that compounds capable of activating autophagy may serve to enhance the ARpolyQ clearance from muscle cells. We use a well-known autophagy activator, trehalose, which acts in a mTOR-independent manner. We recently described that trehalose causes a transient lysosomal damage, which in turn activates TFEB and, consequently, promotes autophagosome and lysosome assembly and fusion (56). We found that in s-myoblasts, trehalose retained its capability to activate autophagy, as demonstrated by the conversion of LC3 from its LC3-I diffuse form to the LC3- II lipidated form associated to autophagosomes in its punctate status (**Figure 4A**, left insets) or by the relocalization of SQSTM1/p62 into p62 bodies (**Figure 4A**, right insets). These data were also corroborated by the mRNA expression analysis showing that trehalose induced the de novo expression of several pro-autophagic genes, including Foxo3, Tfeb, Becn1, Bag3, Bag1, Hspb8, Lc3, Sqstm1/p62, Vcp, and AchR (**Figure 4B**).

The effects of trehalose were then tested on the accumulation of ARQ100 in s-myoblasts and the data showed that this autophagy activator reduced the levels of monomeric soluble ARQ100, and fully counteracted the accumulation of testosterone-induced HMW aggregated species of ARQ100 (**Figure 4C**). Trehalose effect on mutant ARpolyQ was fully blocked by bafilomycin A1, proving that its pro-degradative activity is mediated by autophagy. The effects of trehalose on ARpolyQ clearance were also maintained when the mutant protein was transiently overexpressed in basal C2C12, since this autophagy activator significantly reduced both the monomeric soluble ARQ112 evaluated in WB (**Figure 4D**, upper inset), and testosterone-induced aggregated species of ARQ112 evaluated in FRA (**Figure 4D**, lower inset). Trehalose activation of autophagy was tested in basal C2C12 expressing ARQ112

independent biological samples (n = 3).

only by LC3 conversion in WB, assuming that the effects observed by RT-qPCR and IF in s-myoblasts expressing ARQ100 were recapitulated also in basal C2C12, and the data confirmed that trehalose acts as a potent autophagy inducer as previously shown (51, 56, 91, 92). Importantly, testosteronedependent ARQ112 inclusions observed in IF (**Figure 4E**) were found to be fully degraded after trehalose treatment. Finally, we found that activation of autophagy with trehalose counteracted the accumulation also of the aggregated species of fragmented ARQ1121HA retained in FRA (**Figure 4F**), and the ARQ1121HA inclusions evaluated in IF (**Figure 4G**).

## DISCUSSION

In this study, we characterized the biochemical behavior of the mutant ARpolyQ in s-myoblasts, and we compared it to the one of the wtAR. We found that, in these cells, the mutant ARpolyQ does not form inclusions visible by microscopy, or detectable by immunoblotting as SDS-insoluble aggregates in WB, even after its testosterone activation. Instead, we found that ARpolyQ generated testosterone-inducible aggregated species readily detectable in FRA, which were resistant to NP-40 solubilization. Notably wtAR insoluble species were detected only in the PBS resistant fraction, but not in NP-40 soluble or insoluble fractions, suggesting that even if formed they remain largely soluble, while those formed by mutant ARpolyQ becomes detergent-insoluble. Both the wtAR and the mutant ARpolyQ are processed via the proteasome, while only mutant ARpolyQ is cleared by autophagy, since autophagy inhibition resulted in a robust accumulation of ARpolyQ insoluble species in FRA. Despite this, the presence of ARpolyQ was insufficient to trigger an autophagic response, since no variation were found in the expression of classical autophagy related genes (e.g., Tfeb, Becn1, Bag1, Hspb8, Sqstm1/p62, Lc3) even after testosterone treatment.

Of note, we found that ARpolyQ activation in s-myoblasts correlated with a reduced expression of two pro-autophagic proteins such as BAG3 and VCP. Mutations in BAG3 and VCP genes are responsible for late onset degenerative diseases affecting skeletal muscle (93, 94), suggesting that these proteins might play an important role in the maintenance of muscle

three independent biological samples (n = 3).

DAPI. Magnification: 63X. Microscope: confocal LSM510 Zeiss. Each experiment was independently replicated three times (n = 3). Graphs show quantification of

cell homeostasis. This phenomenon is not correlated to the presence of testosterone, but still suggestive of a decreased autophagic response in s-myoblasts in presence of ARpolyQ. Overall these results point to the fact that activated ARpolyQ does not greatly affect the functionality of the PQC system in our muscle cell model. Anyway, it might be possible that the mildly reduced autophagic potential causes ARpolyQ aggregation only after testosterone activation, slowing down ARpolyQ clearance via autophagy. To facilitate the degradation of ARpolyQ, we overexpressed BAG3 or its partner HSPB8, essential components of the CASA complex, showing that they are both able to enhance ARpolyQ clearance even in presence of testosterone. Also, the overexpression of BAG1, the co-chaperone which routes the HSP70/CHIP/misfolded protein complex to UPS (95–97), exerted a similar effect on ARpolyQ clearance. Overexpression of these chaperones was effective also against aggregates formed by the N-terminal ARpolyQ fragment, physiologically formed upon testosterone treatment. Thus, the modulation of the PQC could be viewed as a potential target to ameliorate the removal of toxic ARpolyQ from our muscle cell models. Indeed, by adopting a pharmacological treatment with trehalose, which is a mTOR independent autophagy activator, we have clearly shown that the insoluble species of ARpolyQ disappeared, both using the ARQ100 in stably infected cells as well as with ARQ112 or the caspase-3 released N-terminal fragment transiently transfected in s-myoblasts. The involvement of autophagy in mediating the pro-degradative activity of trehalose in s-myoblasts was proved by the fact that treatment with bafilomycin A1 fully reverted the protection exerted by trehalose against ARpolyQ accumulation.

In our view, these data acquire particular relevance keeping in mind that SBMA, regarded for years as a MND (3, 98), has now been defined as a neuromuscular disease (25). Muscle tissue is a primary site for SBMA toxicity as muscle atrophy often precedes motoneuron loss and the onset of SBMA is rescued by specific repression of ARpolyQ in muscle cells (33, 50). Even if these studies were carried out in murine models and findings remain to be confirmed in human cell lines, they support the notion of a direct muscle involvement in SBMA onset and progression. Here, we found that autophagy activation, or facilitation, prevents ARpolyQ accumulation in our muscle cell models, suggesting that autophagy could be a specific pathway for the degradation of testosterone activated ARpolyQ insoluble species. In addition, autophagy appears to be partially impaired, making it an important target to facilitate misfolded ARpolyQ clearance in SBMA. Studies performed in SBMA mouse models indicated that at later stage of disease, autophagy is altered in skeletal muscle (22, 32, 50, 73, 99), but its role is still largely debated. One of the major problem linked to these analysis is that the specific time window in which the mice are analyzed (pre-symptomatic, symptomatic or end stage of disease) may influence the relative involvement of autophagy in response to mutant ARpolyQ due to several compensatory mechanisms triggered during muscle atrophy progression. In addition, the mouse SBMA models utilized significantly differ in term of level of protein expression, and its tissue distribution. However, the analysis of TFEB activity, measured by evaluating its target genes, showed that autophagy is enhanced in presence of ARpolyQ in muscle tissue (22). Using the same SBMA model, we also confirmed and extended the activation of TFEBmediated autophagy (73). In addition, in the same animals, we demonstrated that, at the symptomatic stage, also the expression of genes involved in CASA-complex (e.g., Hspb8 and Bag3) resulted upregulated. Thus, in skeletal muscle autophagy is activated during disease progression, and its upregulation might be an attempt to respond to ARpolyQ toxicity, or to mediate the catabolic activity induced by muscle atrophy associated to the chronic exposure to ARpolyQ (100). This may suggest that the autophagy response observed in the skeletal muscle of the SBMA mice is an adaptive mechanism related to both the presence of the misfolded ARpolyQ, and the muscle atrophy. In any case, we cannot exclude that aberrant autophagic upregulation contribute to SBMA progression. Overall, these data suggest that restoration of physiological autophagic function might represent an important therapeutic target for SBMA. Our smyoblast model may be particularly relevant for the screening of compounds that may modulate autophagy dysregulation in muscle cells. In addition, our s-myoblast model will permit to evaluate the acute response to ARpolyQ activation by testosterone. Indeed, acute ARpolyQ expression leads to a mild autophagy response. Misfolded ARpolyQ production is insufficient to induce a de novo expression of all gene tested, with the exception of Bag3 and Vcp. Our s-myoblast SBMA model may thus contribute to understand the different events taking place in skeletal muscle cells exposed to misfolded ARpolyQ allowing to discriminate between early and/or adaptive response. A possible limitation of this model is the fact that s-myoblasts do not show modification in cell viability induced by ARpolyQ. Despite this, they are characterized by the specific accumulation of testosterone-induced ARpolyQ NP-40 insoluble species (not detectable in the case of wtAR); these species clearly represent a biochemical form of misfolded ARpolyQ which play a role in SBMA pathogenesis recapitulating the disease phenotype. As it has been published (80, 101, 102), the formation of the insoluble ARpolyQ species, might be a valuable biomarker to follow the progression of muscle degeneration.

Importantly, the systems here described, including some chaperones and autophagy/proteasome, are highly conserved and work in a similar manner in neuronal, muscular and non-neuronal cells. Therefore, boosting them could provide protection by enhancing the clearing capacities, and maintaining protein homeostasis in different cell types affected by the disease.

In this context, targeting autophagy could be an efficient strategy to reduce the accumulation of ARpolyQ. Trehalose not only activates the basal autophagy process (e.g., TFEB activation enhanced SQSTM1/p62 and LC3 expression), but also increased the expression of key factors of the CASA-complex, like HSPB8, that probably helps in the recognition of selected cargo avoiding the uncontrolled degradation of every intracellular element.

Even if C2C12 SBMA cell model does not show a reduction in cell viability induced by ARpolyQ NP-40 insoluble species, it might be helpful to understand molecular mechanisms responsible for muscle degeneration observed in SBMA patients as it has been published in other publications (80, 101, 102). Since in this model AR NP-40 insoluble species are polyQ and testosterone-dependent, recapitulating the disease phenotype, these C2C12 cell lines could be used to co-culture skeletal muscle and motoneurons in order to study if ARpolyQ expression in myoblast can alter motoneuron functionality and viability.

Overall these results show that ARpolyQ aggregation may occur also in muscle cells, and that targeting aggregation of ARpolyQ could be beneficial in SBMA, since the permanence of inclusions in the cells could cause the damage of several pathways and the recruitment of other soluble proteins, impairing other pathways. Concluding, trehalose plays beneficial effects against ARpolyQ aggregation and autophagy appears as a valuable pathway for the degradation of insoluble ARpolyQ species. In parallel and supporting this study, there are several ongoing studies that are testing, in in vivo models, novel compounds that will address the PQC system, to reduce the presence of the misfolded toxic proteins.

#### DATA AVAILABILITY

The datasets generated for this study are available on request to the corresponding author.

#### ETHICS STATEMENT

This study is not subjected to ethical committee approval, since no animal and human data have been collected.

#### AUTHOR CONTRIBUTIONS

MEC and RC performed most of the experiments. VC, VF, BT, EC, and MC contributed to perform experiments and critically revised the manuscript. MG, MPi, and EM contributed to design the experiments and critically discussed the data. SC provided expertise on BAG1-BAG3 functions and revised the manuscript. MPe provided the stably infected cells and assisted in experiment design with the expertise on SBMA. PR provided expertise on trehalose experiments and critically revised the manuscript. AP supervised the entire study and wrote the paper.

### REFERENCES


#### FUNDING

The following grants are gratefully acknowledged: Fondazione Telethon, Italy (n. GGP14039 to AP, GGP19128 to AP and MP); Kennedy's disease association (2018 grant to RC); Fondazione Cariplo, Italy (n. 2014-0686 to AP and SC; n. 2017\_0747 to VC); Fondazione AriSLA, Italy (n. ALS\_HSPB8 to AP and SC; ALS\_Granulopathy to AP and SC; MLOpathy to AP and SC; Target-RAN to AP); Association Française contre les Myopathies, France (AFM Telethon n. 16406 to AP); Università degli Studi di Milano e piano di sviluppo UNIMI - linea B (to VC and PR); Italian Ministry of University and Research (MIUR), PRIN - Progetti di ricerca di interesse nazionale (n. 2015LFPNMN to AP and SC; n. 2017F2A2C5 to AP, SC, and MP); Fondo per il Finanziamento delle Attività Base di Ricerca (FFABR) (MIUR, to MG, EM, and to PR); Agenzia Italiana del Farmaco (AIFA) (Co\_ALS to AP and SC); Italian Ministry of Health (n. GR-2011-02347198 to VC and SC); Fondazione Regionale per la Ricerca Biomedica (FRRB) (Regione Lombardia, TRANS\_ALS, project nr. 2015-0023, to AP); European Molecular Biology Organization (EMBO), short term fellowship (n. 537 - 2015 to RC). International Brain research organization (InEurope short stay grant to MEC). This is an EU Joint Programme - Neurodegenerative Disease Research (JPND) project.

#### ACKNOWLEDGMENTS

The authors are grateful to Dr. A. P. Lieberman (University of Michigan, Ann Harbor) for having provided the plasmid p5HBhARQ112. Dr. D. Merry (University of Philadelphia, Thomas Jefferson Hospital) for having provided the plasmids pARQ161HA and pARQ1121HA. Prof. H. H. Kampinga (Groeningen University, Groeningen, The Netherlands) for having provided the plasmids pCI-HSPB8, pCI-neo-6xHisBAG3, pCDNA/HA-BAG1. Dr. Landry (Centre of Recherche Cancerologie, University of Laval, Canada) for having provided the home-made HSPB8 antibody.


(Kennedy's syndrome): an electrophysiological study. J Neurol. (1996) 243:388–92.


of the phosphorylation sites. J Biol Chem. (2002) 277:29304–14. doi: 10.1074/jbc.M204131200


and frontotemporal dementia spectrum. J Clin Neurosci. (2019) 64:8–10. doi: 10.1016/j.jocn.2019.03.063


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Cicardi, Cristofani, Crippa, Ferrari, Tedesco, Casarotto, Chierichetti, Galbiati, Piccolella, Messi, Carra, Pennuto, Rusmini and Poletti. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Secretory Autophagy and Its Relevance in Metabolic and Degenerative Disease

#### Claudio Daniel Gonzalez 1,2, Roxana Resnik 1,2 and Maria Ines Vaccaro1,2 \*

*<sup>1</sup> Department of Pathophysiology, Institute of Biochemistry and Molecular Medicine (UBA-CONICET), School of Pharmacy and Biochemistry, University of Buenos Aires, Buenos Aires, Argentina, <sup>2</sup> CEMIC University Institute, Buenos Aires, Argentina*

Proteins to be secreted through so-called "conventional mechanisms" are characterized by the presence of an N-terminal peptide that is a leader or signal peptide, needed for access to the endoplasmic reticulum and the Golgi apparatus for further secretion. However, some relevant cytosolic proteins lack of this signal peptides and should be secreted by different unconventional or "non-canonical" processes. One form of this unconventional secretion was named secretory autophagy (SA) because it is specifically associated with the autophagy pathway. It is defined by ATG proteins that regulate the biogenesis of the autophagosome, its representative organelle. The canonical macroautophagy involves the fusion of the autophagosomes with lysosomes for content degradation, whereas the SA pathway bypasses this degradative process to allow the secretion. ATG5, as well as other factors involved in autophagy such as BCN1, are also activated as part of the secretory pathway. SA has been recognized as a new mechanism that is becoming of increasing relevance to explain the unconventional secretion of a series of cytosolic proteins that have critical biological importance. Also, SA may play a role in the release of aggregation-prone protein since it has been related to the autophagosome biogenesis machinery. SA requires the autophagic pathway and both, secretory autophagy and canonical degradative autophagy are at the same time, integrated and highly regulated processes that interact in ultimate cross-talking molecular mechanisms. The potential implications of alterations in SA, its cargos, pathways, and regulation in human diseases such as metabolic/aging pathological processes are predictable. Further research of SA as potential target of therapeutic intervention is deserved.

Keywords: unconventional protein secretion, IL-1β, aggregate-prone proteins, macroautophagy, ATG (autophagyrelated) proteins

## AUTOPHAGY

Autophagy is an evolutionarily conserved cellular process induced by nutrient starvation or lack of growth factors that sequester and delivers cytoplasmic components to the lysosome for degradation (1). The classical functions of autophagy are nutrient recycling functions by bulk sequestration from the cytoplasm (2). It is also involved in the cytoplasmic component quality control by removing

#### Edited by:

*Alessandro Cellerino, Normal School of Pisa, Italy*

#### Reviewed by:

*Shanya Jiang, University of New Mexico, United States Maurizio Renna, University of Cambridge, United Kingdom*

#### \*Correspondence:

*Maria Ines Vaccaro mvaccaro@ffyb.uba.ar; maria.vaccaro@gmail.com*

#### Specialty section:

*This article was submitted to Endocrinology of Aging, a section of the journal Frontiers in Endocrinology*

Received: *04 October 2019* Accepted: *09 April 2020* Published: *05 May 2020*

#### Citation:

*Gonzalez CD, Resnik R and Vaccaro MI (2020) Secretory Autophagy and Its Relevance in Metabolic and Degenerative Disease. Front. Endocrinol. 11:266. doi: 10.3389/fendo.2020.00266*

**109**

specifically damaged or aging organelles, such as depolarized mitochondria (3). Autophagy is also important in proteostasis, sequestering and degrading long-life proteins and invading microbes (4). However, recent reports found that autophagy also presents non-canonical functions (5), especially regulating unconventional secretory processes. Thus, a novel nondegradative role of autophagy has emerged, raising the concept of Secretory Autophagy (SA) (6–8).

In the classical view, according to the pathway that cargo follows to reach the lysosomal compartment, there are three mayor types of canonical degradative autophagy. These types are: microautophagy/endosomal microautophagy (9, 10), chaperonemediated autophagy (CMA) (11, 12), and macroautophagy. The last one is characterized by the engulfment of cytoplasmic contents by a double membrane vesicle, named autophagosome. Therefore, macroautophagy (hereafter mentioned as autophagy) is distinguished by the formation of the autophagosome as its characteristic and representative intracellular organelle (13).

The autophagic process involves the fusion of the outer membrane of the autophagosomes with lysosomes to deliver the inner vesicle with its cargo to the degradation compartment forming the autolysosome. In the autolysosome, the inner vesicle is degraded, and its products recycled. So far, more than 100 molecules have been related to autophagy regulation and were named ATG molecules (14). Two signaling pathways are associated with autophagy induction: those involve mTOR and AMPK activation. These signaling pathways can sense the environmental, nutritional and energetic status of the cell and promote autophagy through the ULK1-complex, which is the first member of the core molecular machinery in the autophagosome biogenesis [reviewed in (15, 16)]. In brief: Following ULK1 complex activation, the transmembrane protein VMP1 (17) recruits on the ER surface contact site (18) where the first structure in the autophagosome biogenesis, called omegasome, is formed. VMP1 also interacts with the BH3 domain of BECN1 recruiting the kinase complex PI3KC3-C1 to the autophagosome membrane (19). The events that lead to the initial structures (isolation membrane) are followed by the BECN1-PI3K complex activity that phosphorylates the autophagosome membrane and two ubiquitin-like systems ATG12 and LC3 that promote the proper recognition of PI3P (20). In this way, cytoplasmic ATG12 is covalently attached to a C-terminal glycine of ATG5. Furthermore, the ATG5-ATG12 complex promotes LC3 conjugation to phosphatidylethanolamine (PE) on the autophagosomal membrane and this process is mediated by ATG16L, which interacts with ATG5 to eventually form the ATG12-ATG5-ATG16L complex [(21); **Figure 1**].

It is well-known that LC3 plays a central role in autophagy being involved in vesicle elongation, maturation, fusion of autophagosome-lysosome, and even as an adaptor to cargo recognition (22, 23). The lipidated LC3, (LC3B) is present at the isolation membrane and in the autophagosome, in both sides of the membrane. The arrival of autophagosome to the lysosome, is a fusion dependent mechanism of the HOPS complex, through STX17 (24), and RAB7 (25). Thus, LC3 from the inner membrane of the autophagosome is degraded with the cargo. LC3 localized in the external membrane is cleaved from the PE by ATG4B and then recycled (26–28).

Apart from its physiological/homeostatic function, autophagy is also considered as a cell adaptation-to-stress process, which frequently starts as a consequence of organelle damage caused by oxidative species and other stress conditions. Any specific sequestration of a selected type of cargo by autophagy for its delivery to the lysosome is called selective autophagy (29). Selective autophagy has a role in intracellular homeostasis, mediating the specific degradation of cytoplasmic material such as aggregated proteins or damaged mitochondria (30). Interactions between autophagy receptors and ubiquitin-like proteins constitute the molecular basis of selective autophagy. In selective autophagy, a cargo-receptor-protein, such as p62, makes the connection between the selected cargo and LC3 in the autophagosomal membrane (31). Importantly, selective degradative autophagy is involved in the cellular response to complex diseases, such as metabolic/aging pathological processes, by the specific degradation of aggregation-prone or aggregated proteins (30, 32) and organelles. These well-studied aspects of degradative autophagy are widely considered an attractive target for therapeutic strategies (33).

#### SECRETORY AUTOPHAGY

In most cases, especially in exocrine glands and neurons, proteins are secreted by exocytosis (34). The amino-terminal signal peptide (leader sequence) leads eukaryotic secretory proteins into the endoplasmic reticulum (ER), following a well-defined secretory pathway via the Golgi apparatus and eventually progress to the cell surface through vesicular flow. However, some relevant cytosolic proteins lack of this signal peptides and are not able to enter the endoplasmic reticulum (ER). Therefore, they should be secreted by different unconventional or "non-canonical" processes that differ from the classical ER-Golgi pathway (35–37). The autophagy machinery participates in at least one of these pathways. Thus, as mentioned above, this autophagy-dependent secretion pathway is also referred to as SA (6–8).

SA is becoming of increasing relevance to explain the secretion of a series of peptides that have critical biological importance. Interestingly, SA has been shown to play a role in the release of aggregation-prone proteins. This highlights the pathophysiological relevance of this novel, but still not fully elucidated autophagy mediated secretory pathway (38, 39). Autophagy has been also involved in extracellular export of cytosolic organelles, such as mitochondria that can also be released by secretory autophagy (40). Furthermore, different types of non-canonical autophagy have been involved in pathogen released from infected cells (41) and associated with the unconventionally trafficking of proteins to the plasma membrane (42).

Interleukin-1β (IL-1β) secretion is mediated by SA. LC3Bpositive carrier sequesters IL1β from the cytosol and fuses with

the plasma membrane to release this cytokine through a SA process (6–8, 43). IL-1β release seems to request the participation of the TRIM family proteins as receptors for cargo to be secreted. It has been reported that the TRIM family interacting with SEC22B, as well as some Qa-SNARE (syntaxins 3 and 4), and Qbc SNARE (SNAP 23 and 29) are needed to promote the secretory release of IL-1β. Also, these molecules are needed for other unconventional secretion processes, such as those involving Lysozymes, Cathepsin A, B, C, S, Z, and other dipeptidylpeptidases and Tubulin (7, 44). Other SA cargos that do not contain a signal peptide are IL-18 and HMGB1 (45, 46). SA is not restricted to inflammasome substrates and autophagic mediated secretion of other cytosolic proteins lacking leader peptide have been reported, such as Galectin-3, Ferritin, and Annexin-I (47).

It has been observed that SA is involved in α-Synuclein aggregates associated with Parkinson's disease (48–50). SA has been linked to the release of aggregates of amyloid-beta (Aβ) peptide associated with Alzheimer's disease (46). A decrease in Aβ secretion and extracellular Aβ plaque formation and an increase of intracellular Aβ aggregate in the perinuclear zone of neurons were reported in neuron-specific ATG7-deficient mice. Also, ATG7 was able to rescue the ability to release Aβ, whereas the induction of autophagy with rapamycin decreased the secretion of Aβ from wild-type primary neurons (46). Moreover, the secretion of Parkinson's disease- and cancerassociated protein Park7/DJ-1 is mediated by SA. Park7 secretion is induced by autophagy through AMPK and ULK1 activation and it is suppressed in Atg5, Atg9, or Atg16lL deficiency animals (51). On the other hand, the inhibition of degradative autophagy was reported to induce the unconventional secretion of α-Synuclein and Huntingtin protein (50). Therefore, growing evidence accumulates to point at a relevant role of SA in pathological protein aggregate secretion and their intracellular accumulation as mechanisms of cell response to degenerative diseases.

SA has a role in intestinal defense mechanisms being involved in the secretion of Lysozyme in Paneth cells, mediated by ATG161L (52). SA is triggered by bacteria-induced ER stress and it is disrupted in Paneth cells of mice harboring ATG16L1, which is a variant of ATG16L related to high risk for Crohn's disease. Another example is the secretion of CFTR (cystic fibrosis transmembrane conductance regulator): knockdown of ATG5 and ATG7, and treatment with autophagy inhibitors, such as wortmannin and 3-methyladenine, abolished the unconventional secretion of CFTR stimulated by ER stress and an ER-to-Golgi blockade (53).

Several ATG-proteins involved in the biogenesis of the autophagosome have been directly related to the molecular mechanisms in SA (**Table 1**). However, the participation of the autophagosomes in the membrane trafficking pathways of SA is not completely elucidated (37, 58). The analysis of the mechanisms underlying unconventional protein secretion includes a wide range of pathways that goes from different forms of plasma membrane translocation to the generation of extracellular vesicles (EVs).

EVs are a heterogeneous group of cell-derived membranous structures that can be originated from the endosomal system and micro-vesicles that are shed from the plasma membrane (59). A type of EVs with a diameter range between 40 and 120 nm were called exosomes. It is interesting to highlight that the secretion of exosomes was also related to the SA pathway (60). This finding suggests that exosomes can be another cargo for SA. Moreover, some cytoplasmic components, from single molecules to organelles, could be involved in the SA mechanism and may also work as SA cargos (61). However, the intracellular membrane trafficking involved in the exosome secretion pathway largely remains poorly understood. In the classic view, exosome biogenesis would start when early endosomes mature into late endosomes or MVBs (Multi Vesicle Bodies) that fuse with the plasma membrane to release the extracellular vesicles to the environment (62). Hence, during this process, the endosomal membrane invaginates to generate intraluminal vesicles (ILVs). These ILVs are finally released into the extracellular space as exosomes, after the fusion of MVBs with the plasma membrane. On the other hand, ATG proteins such as ATG5, participate in exosome production (63). The location of LC3B on the lumen side of the ILVs suggests that a lipidation event takes place at the MVB membrane, or membrane invaginations. Besides, the fact that intact LC3Bpositive EVs are eventually released, strongly suggests a secretory fate of selected autophagosomes by its fusion with MVBs in another vesicle called amphisomes (62). A recent article provided solid evidence regarding the involvement of SA in the EVs secretion. In this model, LC3-conjugation controlled the formation and secretion of EVs containing RNA-binding proteins (64). Nevertheless, the biological function of ATG5 and other ATG proteins in exosome production and release remains unclear (58). Finally, as it was recently reviewed, autophagy also participates in the control of conventional secretion, including selective types of autophagy such as such as ribophagy (65) and zymophagy (66, 67). It acts on the secretory apparatus at different steps, with selectivity between different secretory cell types (68).


\**Including: rheumatoid arthritis, inflammatory bowel disease; autoimmune thyroiditis; type 2 diabetes.*

#### THE RELEVANCE OF SECRETORY AUTOPHAGY IN DEGENERATIVE, ENDOCRINE, AND METABOLIC DISEASE

#### Disease Linked With Aggregation-Prone Proteins Deposition and Accumulation May Be Associated With Disturbances in Secretory Autophagy

The potential implication of alterations in autophagy mechanisms in human health is the subject of strong research interest. Parkinson's disease (PD), Alzheimer's disease (AD), and another severe neurological diseases might be at least partially associated with alterations in degradative autophagy and/or in the autophagy-based secretion of certain peptides. This may be also the case for endocrine diseases. In neurodegenerative diseases such as PD and AD, as well as in endocrine and metabolic disease such as Diabetes mellitus, the SA mechanism may be altered in a way that toxic products are secreted and accumulated outside the cell as detritus that in time can cause cell death. In AD, Amyloid beta (Aβ) aggregates and accumulates outside the cell causing neuron functional impairment, structural changes and eventually cell death. The complete process reflects a severe imbalance between production, secretion, aggregation, and clearance of Aβ which progresses until the most advanced stages of the disease. Some studies have shown that autophagy plays a critical role in Aβ secretion (57, 69). Degradative autophagy blockade is followed by Aβ intracellular accumulation which has been demonstrated to be toxic to neurons as well as degradative autophagy in microglia is deteriorated by long exposure to Aβ (69). Moreover, the secretion of Aβ depends on autophagy, since Aβ secretion and plaque formation are reduced in mice lacking ATG7 in the excitatory forebrain neurons (57). It has been observed that ATG7 deficient mice tend to accumulate Aβ in the Golgi (57). Interestingly, it seems that autophagy machinery mediates the transport of certain peptides from the Golgi to endosomes (57). Alterations in these processes may affect Aβ secretion and promote the intracellular accumulation of Aβ, which in turn, results in toxic effects for cells. On the other hand, the administration of rapamycin, an autophagy activator, is followed by a reduction in the intracellular content of Aβ and an improvement in cognitive function in mice (57). Some alterations in SA have also been proposed as playing a role in the pathophysiology of the disease. Dysfunctional insulin-degrading enzyme (IDE), secretion may serve as an example of these SA alterations affecting Aβ aggregation (55). IDE -a ubiquitous enzyme of the inverzincin family of peptidases, is involved in the clearance of insulin, insulin-like growth factors, glucagon, amylin, and other peptides. IDE also degrades some neurotransmitters/neurohormones, including transforming growth factor-α, somatostatin, and endorphins. IDE inactivates calcitonin gene-related peptide and seems to modulate inflammatory responses, and the production of some tumor-associated antigens (70). Cellular and oxidative stress, insulin concentrations, free-fatty acid, and starvation seem to modulate IDE expression. IDE dysfunction may be associated with some forms of type 2 diabetes (T2DM), in humans and several single nucleotide polymorphisms in non-coding regions of the IDE gene, are associated with the disease (71–75). However, many IDE inhibitors have resulted in non-conclusive effects in terms of their potential as glucose tolerance improvers (76). As this enzyme is part of the catabolic pathways of insulin and insulin-like growth factors, high renal and hepatic expression of IDE is not a surprise under normal conditions. However, other cells such as astrocytes and some neurons also display relatively high levels of IDE expression (55). Predominantly located in the cytosol, IDE is also found in other organelles including endosomes, peroxisomes and mitochondria, and it is also found at the cell surface. A close to C terminus Sly sequence motif prevents IDE from enzymatic degradation when present at its lysosomal location (70). Usually <10% of the enzyme is secreted, following a non-conventional pathway. IDE secretion in microglia is enhanced by HMG-CoA reductase inhibitors (statins) (77). Having no signal sequence, IDE secretory pathway is mediated by autophagy. Son et al. (55) found that simvastatin is able to induce the degradation of extracellular Aβ40, which depended on IDE secretion from primary astrocytes. Simvastatin increased IDE secretion is mediated by the activation of autophagy through the LKB1- AMPK-mTOR signaling pathway in astrocytes. Importantly, IDE acts as a key protease for Aβ in the central nervous system. Also, Aβ induces IDE secretion involving the contribution of ATG genes. On the other hand, IDE secretion is associated with GORASP (Golgi reassembly and stacking protein) physiology (36). The dose dependent Aβ induced secretion of IDE is also mediated by GORASP and RAB8A. Moreover, the integrity of the autophagy flow is needed for this process. It also has been shown that Aβ injection in mice with alterations in ATG7 results in reduced expression and activity of IDE in the cerebrospinal fluid potentially associated with alterations in SA (55). Even when Aβ induced IDE secretion by astrocytes, it seems to fulfill many of the critical characteristics required to be considered an autophagy-mediated process. It is worth to be mentioned that some steps in this pathway are not completely understood. Interestingly, some epidemiological studies suggest that patients with T2DM are at a higher risk of developing AD. However, it is difficult to be certain about the nature of the link of both diseases with a dysfunctional SA. A standard oral agent for the treatment of type 2 diabetes (metformin), has shown some positive effects in experimental models of AD. Metformin decreases Beta-secretase 1 (BACE 1) activity which results in reduced production of Aβ (78). The inhibition of acetylcholinesterase (AChE) in the central nervous system, may explain at least in part, some beneficial effects on cognitive function, learning, and memory. Furthermore, Metformin reduces oxidative stress and exhibits anti-inflammatory properties (78). However, metformin is also an autophagy regulatory agent that may modulate both, degradative and secretory autophagy, and enhance autophagic clearance of intracellular neurofibrillary tangles formed by hyperphosphorylated tau protein (79). The real impact of these potential beneficial effects in clinical practice remains to be elucidated.

Amyloid Polypeptide and other aggregation-prone proteins accumulation surrounding the Langerhans Islet cells is a common finding in type 2 diabetes in humans. Islet Amyloid Polypeptide (IAPP) accumulation is long recognized as a phenomenon occurring in human type 2 diabetes. IAPP is cosynthetized in beta cells and secreted along with insulin, this is a 37-amino-acid residue polypeptide. Human IAPP forms oligomeric structures and fibrous extracellular precipitates that accumulate in the islets. The genetic background seems to be associated with the degree of IAPP deposition (80) and some nutritional factors (such as fat intake), would facilitate IAPP deposition (81). The implications of the IAPP accumulation in the evolution of human forms of diabetes is poorly understood. Furthermore, extracellular human IAPP can promote autophagy in beta cells and the reactive Oxygen Species (ROS), mediates part of the IAPP induced degradative autophagy by involving the AMPK pathway (82). On the other hand, it has been suggested that intracellular IAPP oligomer formation would be toxic for beta cells. In this way, IAPP oligomers may damage mitochondrial membranes as well as the endoplasmic reticulum (ER) (83). In humans, the augmented expression of IAPP in beta cells is followed by an increase of the autophagy flux (84). Moreover, lack of autophagy in hIAPP-expressing animals resulted in hIAPP oligomer and amyloid accumulation in pancreatic islets, leading to β cell death (83, 85–87). It has been suggested that IAPP oligomers may cross plasma membranes inducing damage by a "prion-like" effect. In this way, the intraperitoneal injection of IAPP aggregates from the pancreas homogenate to the transgenic mouse that expresses hIAPP, dramatically accelerates IAPP amyloid deposition, which was accompanied by abnormalities resembling T2DM (88, 89). In this process, degradative autophagy integrity seems to be of relevance. In general, increased degradative autophagy would protect against IAPP induced damage on beta cells. However, evidence on the potential role of SA remains less conclusive. Alterations in IDE secretion has been suggested as part of the pathophysiology of IAPP deposition. IAPP is a substrate of IDE. However, IDE inhibition does not seem to increase amyloid deposition of endogenous IAPP in vivo (90). The recently described expression of another aggregable protein, α-synuclein, in beta cells adds complexity to the landscape of aggregationprone proteins induced damage in type 2 diabetes. Alphasynuclein expresses in murine pancreatic islets and exogenous overexpression of α-synuclein reduces insulin secretion by INS1 cells (91). It has been proposed that α-synuclein interacts with the Kir6.2 subunit at the KATP channel located at the betacell membrane suppressing insulin secretion (91). In normal conditions, this mechanism may protect beta-cell from ER stress by down-regulating exaggerated insulin secretion. In the presence of type 2 diabetes, α-synuclein may play a negative role in insulin release contributing to hyperglycemia, oxidative stress, and glycosylation of protein (91). Steneberg et al. (56) showed that autophagic flux is reduced by increased levels of α-synuclein present in β-cells from IDE KO mice and in T2D patients. On the other hand, Sharma et al. (92), proposed that α-synuclein is able to activate IDE, while IDE inhibits amyloid formation by α-synuclein. However, it is unclear if these factors may result in increased IAPP deposition in T2DM in humans. Moreover, several reported changes may represent indirect evidence of the IAPP induced damage and cell adaptation. Also, these alterations

might precede or follow the intracellular accumulation of oligomers. Finally, direct or indirect alterations in SA may result in an increased concentration of proinflammatory mediators such as IL-1β increasing IAPP deposition, functional alteration, oxidative stress, and cell death.

In PD's Disease (PD), the accumulation of α-synuclein, a protein with high binding affinity for smaller vesicles (as synaptic vesicles, for instance), was reported. Alpha-synuclein mutation or over-expression impairs membrane trafficking including exocytosis and ER-to-Golgi transport (93, 94). These alterations are associated with ER stress, increasing oxidative stress affecting cell homeostasis. Moreover, lysosome impairment seems to play a critical role in the progression of the disease (95). It has been proposed that the Secretory Carrier Membrane Protein 5 (SCAMP5), promotes the secretion of α-synuclein and other neurotoxic proteins via exosomes (50). It has been demonstrated that SCAMP5 has some effects as an inhibitor of degradative autophagy, playing role in balancing several processes involved in cell homeostasis, including vesicle trafficking as well as constitutive, degradative and secretory autophagy (50). Also, several other alterations linked to autophagy have been identified in different models of PD. Alterations in the TMEM230 gene (Transmembrane Protein 230), have been recently associated with the pathophysiology of familial forms of the disease. TMEM230 regulates the autophagy-mediated clearance of αsynuclein and mediates Rab8a-associated SA. As a result, this protein regulates Golgi-derived vesicle secretion (96). Lewybodies formation, which is typically associated with cognitive dysfunction and dementia seen in some patients with advanced PD, may result from these alterations (96). PARK7/DJ-1 protein has been associated with PD and certain forms of cancer. PARK7/DJ-1 participates in the regulation of several cell processes, including anti-oxidative protection. Furthermore, an increased pro-oxidative environment is associated with degenerative damage along with the disease evolution. The autophagy pathway involving ATG molecules was associated with PARK7/DJ-1 secretion (51). Thus, impairment in autophagy might result in the defective secretion of this protein with potential implications in terms of anti-oxidative stress protection, playing a role in PD progression.

Amyotrophic lateral sclerosis (ALS), is a degenerative disorder that affects cortical, bulbar and spinal motor neurons and is mainly characterized by a progressive adult-onset. Like other neurodegenerative diseases, ALS is categorized as a "proteinopathy" since SOD1, TDP-43, and FUS are pathological proteins that accumulate, interfere, and impair neuronal functions leading to cell death. Sproviero et al. suggest that these toxic proteins are transported mainly by EVs that might play a role in prion-like propagation of ALS disease (97).

#### Impairment in SA May Affect the Effect and the Concentration of Secreted Peptides and Other Co-secreted Substances

Disruptions in SA may result in relevant changes in the concentrations of secreted peptides and other co-secreted molecules, as well as in their paracrine and endocrine effects.

Altered secretion of lysozyme is related to some forms of inflammatory bowel disease and relevant changes in host defense mechanism. Paneth cells are specialized secretory cells in the small intestine that have been related to Crohn disease through a mutation of ATG16L1 (98). It has been observed, that lysozyme and other critical secreted antibacterial proteins are rerouted by Paneth cells through SA as a reaction to bacterial invasion. Moreover, activation of innate lymphoid cells type 3 (ILC3) which secrete IL22 licenses Paneth cells to secrete lysozyme through SA (52, 99). These data suggest that Crohn disease may be associated, at least in part, to some genetically mediated alterations in SA.

Impaired secretion of other inflammation regulatory molecules by a disrupted SA has been reported. IL-1β is a major pro-inflammatory cytokine, that also doesn't follow the classical endoplasmic reticulum-to-Golgi route. IL-1β release is mediated by SA requiring the participation of the microtubule-associated protein EB1 (100). This autophagy-dependent, unconventional secretion pathway is of special interest (45). IL-1β transcription is induced by different stimuli ranging from microbiological agents to other cytokines and also growth factors (101). Moreover, neutrophils, macrophages, and microglia are relevant sources of IL-1β under infectious and other conditions and it was demonstrated that autophagy is responsible for IL-1β exocytosis under these challenges. As expected for autophagy-based secretory mechanisms, it was shown that siRNA-mediated knockdown of ATG5 reduces IL-1β secretion in neutrophils (54). In contrast, cell starvation increases the colocalization of IL-1β and LC3B promoting IL-1β secretion. Extending data including all data mentioned above strongly suggests that SA is involved in IL-1β secretion by neutrophils and other secreting cells.

IL-1β derives from an inactive precursor, pro- IL-1β, that requires cleavage by caspase-1 for activation. Activation of caspase 1 is mediated by the protein complex called 'inflammasome'. Release of mature IL-1β relays on SA and it has been suggested that GRASP (a Golgi apparatus associated factor) participates in IL-1β secretion (45). After release, a series of inhibitory molecules including IL-1Ra, sIL-1RI, sIL-1RII, and sIL-1RAcP regulates IL-1β mediated inflammation. IL-1β secretion has been demonstrated to be increased in several endocrine, metabolic and degenerative conditions, as well as some acute entities and infections. In fact, a key role in the defense against microbial pathogens and in tissue injury repair, is performed by IL-1β. Local and systemic responses to this cytokine are responsible for homeostatic effects. Exaggerated responses to IL-1β, however, are associated with potentially deleterious effects; and excessive IL-1β activity is associated to vasculitis and thrombosis. IL-1β also plays a role in rheumatoid arthritis and other inflammatory diseases. Part of the structural damage associated with PD, AD, and other degenerative entities seems to be linked to IL-1β effects. For instance, β cells exposure to increased concentrations of IL-1β is associated with functional deterioration and cell death. Moreover, IL-1β is linked to inflammation in obesity, type 2 diabetes, insulinresistance associated entities (e.g., polycystic ovary syndrome), atherosclerosis and other conditions. In addition, IL-1β seems to play a role in the pathogenesis of Autoimmune Thyroid Diseases (102). Regarding type 2 diabetes, only when basal levels of IL-1β mRNA are low, hyperglycemia induced IL-1β production in β cells can be observed (103). IL-1β is also produced by infiltrating immune cells in the pancreas (104). Taking all this information into account, it seems that IL-1β acts as a "metabolic sensor" aside of its wellrecognized role as pro-inflammatory mediator (105). IL-1β physiology is a highly illustrative example of the complexities of degradative and secretory autophagic processes. Hence, the induction of inflammasomes triggers autophagosome formation in macrophages (106). In some tissues, this effect may be part of a negative mechanism to control and limit the inflammatory response confronting challenges of infectious origin. It seems to be also linked to an increased cytokine-mediated antimicrobial defense.

In some cases, IL-1β was found to be increased in degradative autophagy and may be linked to tissue damage. In pancreatic acinar cells, IL-1β hyperactivity seems to be associated with increased endoplasmic reticulum stress-inducing autophagy. This mechanism may be related to an impaired autophagic flux leading to trypsin activation and pancreatic injury. It has been observed that Atg 5 alterations are followed by an increment in IL-1β plasma concentration though pro-IL-1β caspase-mediated cleavage (107). As mentioned before, autophagy is critical to IL-1β release through the SA mechanism that involves the AIM2 inflammasome. The interplay between different cytokines (many of them affecting autophagy in several manners), adds complexity to this homeostatic network. This effect may result in a chain of autophagic process modifications affecting the SA of regulatory peptides. Although precise mechanisms articulating the balance between the complex processes enumerated above remain unclear, many of them seem to be of major relevance for a better understanding of the immune-inflammatory components in degenerative and metabolic conditions.

Autoimmune thyroiditis (AIT) is among the main causes of hypothyroidism in human. An increased expression of some inflammasome components as well as IL-1β was observed in thyroid gland tissues from AIT patients (108). This phenomenon suggests an upregulated SA activity in these subjects and may play a role in the pathophysiology of the disease. Other autoimmune endocrinopathies may also reflect a certain degree of dysfunctional SA with dysregulated secretion of IL-1β and other cytokines. Elevated levels of pro-osteoporotic cytokines including IL-1β have been found in patients with different forms of hyperthyroidism (109). IL-1β secreted by peripheral monocytes induces IL-6 secretion by stromal cells and osteoblast, IL-6 increments osteoclast proliferation and differentiation inducing increased bone resorption (110) A role of IL-1β hypersecretion cannot be excluded as a potential mediator of hyperthyroidism associated altered bone resorption (111).

Emerging role of secretory autophagy in carcinogenesis and endocrine tumor progression is a topic of significant clinical relevance. Secretory autophagy may be involved in the secretion of tumor-promoting proteins. It has been suggested, that low expression of Rasal2 gene was associated with the recurrence of luminal B breast cancer (112). Recently has been shown that the Rasal2 gene knock out induces secretory autophagy. Although the potential mediators are not elucidated, the increase in SA seems to be associated with luminal breast cancer proliferation (113). Altered secretory patterns of IL-1β, IL-6, IL-8, bFGF, and other growth factors have been demonstrated for certain tumor types (114). Interestingly, IDE is expressed in some human undifferentiated breast and ovarian types of cancer, as well as in retinoblastoma. A tumor-suppressing activity was suggested for this enzyme (115, 116). Dysfunctional SA of IDE might play a role in very aggressive tumors. Paracrine secretion between cancer-cell interactions would facilitate tumor initiation, growth, and spreading. It has been suggested that paracrine secretion of IL6, IL8, and bFGF induces autophagy in head and neck cancer-associated fibroblasts (114). Therefore, SA and degradative autophagy modulates critical interactions between tumor cells and the surrounding microenvironment and may result in significant metabolic modifications, with a strong impact in cancer cell adaptations by Warburg and reverse Warburg effects (117).

#### CONCLUSIONS

Macroautophagy is a complex cellular pathway characterized by the formation of the double-membrane vesicle called autophagosome that sequesters cytoplasmic contents to be delivered to the lysosomal compartment for degradation. However, increased experimental evidences strongly support that another fate of the autophagosome biogenesis molecular machinery exists, leading to a novel pathway of unconventional secretion. This last process was named SA to differentiate it from the canonical degradative autophagy (6, 61).

SA has been recognized as a novel mechanism to explain the secretion of a series of peptides which have critical biological importance. Although the molecular pathways and vesicular trafficking in SA are not fully elucidated, canonical autophagy machinery is mechanistically involved (7). Thus, the autophagyrelated proteins, such as ATG5 and several other components of the autophagosome biogenesis (13) play a central role in the secretion of critical disease-related proteins. Indeed, autophagosome formation seems to be involved in the eventual release of these relevant proteins (8). Interestingly, the fact that cargo recognition molecules, such as the TRIM family are mechanistically involved in this process (7) strongly suggests that SA could be considered as a non-canonical type of selective autophagy. In SA the process does not end in lysosomal degradation but in the secretion of the selected cargo. A wide range of proteins secreted by mammalian cells following different pathways are related to autophagy and associated with human diseases. This includes proteins related to some endocrine and aging associated diseases.

Degradative and secretory autophagy are two integrated multistep processes highly regulated by several physiological and disease-related factors. Starvation, oxidative stress, and hypoxia are well recognized stimuli for degradative autophagy. SA efficiency depends at least in part, on the integrity of the molecular machinery involved in the degradative processes. Changes in degradative and secretory autophagy may result in impairment of the release of different peptides that, in turn, may affect autophagy in other cells by paracrine or even endocrine ways. Implications of this interplay in the prevention, prognosis and treatment of several diseases are still to be elucidated.

Multiple neurodegenerative disorders have in common abnormal protein accumulation and aggregation (32). Autophagosomes or MVBs would be able to sequester and degrade cytosolic-protein aggregates through lysosomes. Externalization of protein aggregates may also be mediated by secretory autophagy. On the other hand, changes in the secretion of protein aggregates might decrease the proteotoxic stress in the releasing cells or reduce the spreading of protein aggregates to neighboring cells.

There is substantial experimental data that allow us to consider that canonical degradative autophagy and its related factors are mechanistically associated with secretion. It has been demonstrated that SA requires the autophagic pathway and both, secretory and degradative autophagy are integrated and highly regulated processes that interact in ultimate crosstalking molecular mechanisms. Impaired secretory autophagy may result in the aggregation-prone proteins deposition and accumulation, or severe alterations in the release and concentration of other secreted proteins. The relevance and the mechanisms involved in these interactions seem to be very important in metabolic and degenerative diseases. The pharmacological modulation of SA and its regulatory pathways might also be a clear target for drug research. Although there is available evidence that outline the potential relevance of the pharmacological control of SA, important gaps in the evidence remain to be filled. Metformin, for instance, may modulate SA by its demonstrated effects on degradative autophagy. It has been observed that in patients with type 2 diabetes, metformin upregulates mitophagy, and improves mitochondrial function in a glucose-lowering independent manner (118). By inhibiting macrophage activation and activating autophagic flux, metformin reduces pro-inflammatory cytokines release (including IL-1β) (119). However, a direct effect on SA mechanisms cannot be excluded. Statins (HMGCoA reductase inhibitors) increases IDE release by astrocytes in a dosedependent manner. These agents directly modulate SA at least in some tissues. However, the potential relevance of this mechanism on the beneficial and/or adverse events associated with statin use in practice remains obscure. Iron enhances Aβ stimulated IL-1β secretion in microglia (120). The impact of this finding in humans is still unknown. Many other agents seem to modulate IL-1β secretion by regulating SA processes in different ways, and in different cell types (121). Translation of these findings to drug development and application into clinical practice will take a while. However, the potential implication of impairment in secretory autophagy, its cargos, pathways, and regulation in human diseases such as metabolic/aging pathological processes is a clear focus of biomedical investigation. Further research on

secretory autophagy pathways as a potential target of therapeutic intervention is deserved.

#### AUTHOR CONTRIBUTIONS

MV selected the subject, wrote the 1st and 2nd sections, and the conclusions, selected part of the bibliography, reviewed the whole manuscript, and made the table. CG wrote the abstract and the third part of the manuscript and selected part of the bibliography

#### REFERENCES


and reviewed the manuscript. RR collaborated in the writing of the whole manuscript and made the figures.

#### FUNDING

This work was supported by grants from the University of Buenos Aires (UBACyT); National Council for Scientific Research and Technology (CONICET-PIP), and National Agency for Scientific and Technological Promotion (PICT), Argentina.

biogenesis by regulation of local PI3P synthesis. EMBO J. (2017) 36:2018– 33. doi: 10.15252/embj.201797006


controlled study. J Cell Mol Med. (2020) 24:2832–46. doi: 10.1111/jcmm. 14834


**Conflict of Interest:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2020 Gonzalez, Resnik and Vaccaro. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# A Novel E2F1-EP300-VMP1 Pathway Mediates Gemcitabine-Induced Autophagy in Pancreatic Cancer Cells Carrying Oncogenic KRAS

Alejandro Ropolo<sup>1</sup> \*, Cintia Catrinacio<sup>1</sup> , Felipe Javier Renna<sup>1</sup> , Veronica Boggio<sup>1</sup> , Tamara Orquera<sup>1</sup> , Claudio D. Gonzalez 1,2 and Maria I. Vaccaro<sup>1</sup>

*<sup>1</sup> Department of Pathophysiology, Institute of Biochemistry and Molecular Medicine (UBA-CONICET), School of Pharmacy and Biochemistry, University of Buenos Aires, Buenos Aires, Argentina, <sup>2</sup> CEMIC University Institute, Buenos Aires, Argentina*

#### Edited by:

*Giorgio Stassi, University of Palermo, Italy*

#### Reviewed by:

*Anna Tesei, Romagnolo Scientific Institute for the Study and Treatment of Tumors (IRCCS), Italy Elisa Giovannetti, VU University Medical Center, Netherlands*

> \*Correspondence: *Alejandro Ropolo aropolo@ffyb.uba.ar*

#### Specialty section:

*This article was submitted to Cancer Endocrinology, a section of the journal Frontiers in Endocrinology*

Received: *24 December 2019* Accepted: *22 May 2020* Published: *23 June 2020*

#### Citation:

*Ropolo A, Catrinacio C, Renna FJ, Boggio V, Orquera T, Gonzalez CD and Vaccaro MI (2020) A Novel E2F1-EP300-VMP1 Pathway Mediates Gemcitabine-Induced Autophagy in Pancreatic Cancer Cells Carrying Oncogenic KRAS. Front. Endocrinol. 11:411. doi: 10.3389/fendo.2020.00411* Autophagy is an evolutionarily preserved degradation process of cytoplasmic cellular constituents, which participates in cell response to disease. We previously characterized VMP1 (Vacuole Membrane Protein 1) as an essential autophagy related protein that mediates autophagy in pancreatic diseases. We also demonstrated that VMP1-mediated autophagy is induced by HIF-1A (hypoxia inducible factor 1 subunit alpha) in colon-cancer tumor cell lines, conferring resistance to photodynamic treatment. Here we identify a new molecular pathway, mediated by VMP1, by which gemcitabine is able to trigger autophagy in human pancreatic tumor cell lines. We demonstrated that gemcitabine requires the VMP1 expression to induce autophagy in the highly resistant pancreatic cancer cells PANC-1 and MIAPaCa-2 that carry activated *KRAS*. E2F1 is a transcription factor that is regulated by the retinoblastoma pathway. We found that E2F1 is an effector of gemcitabine-induced autophagy and regulates the expression and promoter activity of VMP1. Chromatin immunoprecipitation assays demonstrated that E2F1 binds to the *VMP1* promoter in PANC-1 cells. We have also identified the histone acetyltransferase EP300 as a modulator of VMP1 promoter activity. Our data showed that the E2F1-EP300 activator/co-activator complex is part of the regulatory pathway controlling the expression and promoter activity of VMP1 triggered by gemcitabine in PANC-1 cells. Finally, we found that neither VMP1 nor E2F1 are induced by gemcitabine treatment in BxPC-3 cells, which do not carry oncogenic KRAS and are sensitive to chemotherapy. In conclusion, we have identified the E2F1-EP300-VMP1 pathway that mediates gemcitabine-induced autophagy in pancreatic cancer cells. These results strongly support that VMP1-mediated autophagy may integrate the complex network of events involved in pancreatic ductal adenocarcinoma chemo-resistance. Our experimental findings point at E2F1 and VMP1 as novel potential therapeutic targets in precise treatment strategies for pancreatic cancer.

Keywords: pancreatic cancer, gemcitabine, autophagy, VMP1, E2F1

#### INTRODUCTION

Pancreatic ductal adenocarcinoma (PDAC) is one of the most aggressive human malignancies with 8–9% 5-year survival rate (1). Despite the progress in the knowledge of the disease, it remains a calamitous neoplasia. Up to 60% of patients have advanced pancreatic cancer at the time of diagnosis, and their median survival time is 3–6 months (2). Its poor prognosis has been attributed to a tendency regarding early vascular dissemination and spreading to regional lymph nodes, and to the incapacity to make a diagnosis while the tumor is still surgically removable (2). This is caused both by the aggressive nature of the disease, the lack of specific symptoms and early detection tools, and the refractory response to traditional cytotoxic agents and radiotherapy (3, 4). Furthermore, pancreatic cancer cells get more malignant and survive with an extremely low blood supply (2, 3). Up to now, contradictory data are available concerning autophagy activity and its regulation by specific autophagy related (ATG) proteins in pancreatic cancer cells. Experimental evidence places autophagy as a mechanism for survival of tumor cell under adverse environmental conditions, or as a defective mechanism of programmed cell death that promotes pancreatic cancer cell resistance to treatment (5–7).

At present, the first option for resectable tumors in pancreatic cancer is adjuvant chemotherapy before surgical resection (8, 9). However, most patients are in an advanced stage at the time of diagnosis and in these cases, chemotherapy is used as the first option. Regarding chemotherapy, gemcitabine used alone or in combination with nabpaclitaxel represent one of the most effective therapy (9, 10), despite its poor efficacy in terms of overall patient survival (11). Gemcitabine works by causing apoptosis of malignant cells in pancreatic cancer (12, 13). Intrinsic and acquired factors are involved in gemcitabine resistance. Several of them related to the transport and metabolism of gemcitabine (14) and associated with the tumor microenvironment, among others (15, 16).

Interestingly, recent studies highlight the importance of autophagic flux in acquiring resistance to gemcitabine in pancreatic cancer tumor cells (17–19). Macroautophagy (hereafter autophagy), is an evolutionarily conserved process that involves the sequestration and delivery of cytoplasmic components into the lysosome, where they are degraded and recycled (20). Autophagy is involved in the turnover of longlived proteins and other cellular macromolecules. It has also been involved in the physiological responses to exercise and aging and is implicated in different pathophysiological processes such as neurodegenerative disorders, cardiovascular, pulmonary diseases, and cancer (21–24). Autophagy correlates with poor patient outcome in pancreatic cancers (25), and it has been suggested that autophagy is required for tumor growth (6, 7). For instance, in mice with pancreas containing an activated oncogenic allele of KRAS proto-oncogene, GTPase (KRAS), the most frequent mutation in PDAC (26), a small number of pre-cancerous lesions are developed that become PDAC randomly over time (27). KRAS activates the expression of the Vacuole Membrane Protein 1 (VMP1) to induce and maintain autophagy levels in pancreatic tumor cells (28). Accordingly, mice lacking the essential autophagy genes ATG5 or ATG7 acquire pre-invasive low-grade pancreatic intraepithelial neoplasia lesions, but progression to high-grade pancreatic intraepithelial neoplasia lesions and PDAC is blocked (27). This evidence highlight the relevance of KRAS-induced autophagy in the malignant transformation of pancreatic tumor cells.

Autophagy involves the formation of double-membrane structure, autophagosomes, around the cellular components targeted for degradation, which include large structures such as organelles and protein aggregates (29). Autophagy is mediated by a set of evolutionarily conserved gene products (termed the ATG proteins) originally discovered in yeast (30). In mammalian cells, the sequential association of at least a subset of the ATG proteins, referred to as the core molecular machinery (29), leads to the autophagosome formation. VMP1 belongs to these essential ATG proteins. We have demonstrated that VMP1 expression triggers autophagy in mammalian cells even under nutrient-rich conditions (31, 32). By contrast, autophagy is completely blocked in the absence of VMP1 expression (31). VMP1 autophagyrelated function requires its hydrophilic C-terminal domain of 20 amino acids (VMP1-ATGD) (32). This domain binds directly to the Bcl-2 binding domain (BH3) motif of beclin 1 (BECN1) leading to the formation of a VMP1-BECN1-PI3KC3 (phosphatidylinositol 3-kinase catalytic subunit type 3) complex at the site where autophagosomes are generated (33, 34).

VMP1 is not expressed in normal pancreas, however its expression is early activated in pancreas suffering experimental diabetes mellitus, experimental and human pancreatitis, and in human pancreatic cancer cells (35–39). Interestingly, VMP1 prevents pancreatic cell death induced by acute pancreatitis (35). In previous studies, we found that VMP1 expression is induced by mutated KRAS in pancreatic tumor cells (28). KRAS is a member of the Ras family of GTP-binding proteins that mediate a wide variety of cellular functions including proliferation, differentiation, and survival. KRAS mutation is one of the earliest genetic events in human PDAC (40). Besides, it has been demonstrated that VMP1 down-regulation reduces cell resistance of pancreatic cells to chemotherapeutic drugs as Imatinib, Cisplatin, Adriamycin, Staurosporin, and Rapamycin (41). In colon cancer cells, we have recently shown that the HIF-1A-VMP1 autophagic pathway is involved in the resistance to photodynamic therapy in colon cancer cells (42). Therefore, we hypothesized that VMP1 is involved in the tumor cell response to chemotherapy in pancreatic cancer cells.

Here, we study the role of autophagy and its molecular mechanism involved in the pancreatic tumor cell response to chemotherapy. We identified a new regulatory pathway,

**Abbreviations:** ATG, related to autophagy; ChIP, chromatin immunoprecipitation; E2F1, E2F transcription factor 1; EGFP, enhanced green fluorescent protein; EP300, E1A binding protein p300; HIF1A, hypoxia inducible factor 1 subunit alpha; KRAS, Kirsten rat sarcoma viral oncogene homolog; LC3, microtubule-associated protein 1 light chain 3; PDAC, pancreatic ductal adenocarcinoma; PBS, phosphate-buffered saline; RLU, Relative light units; RFP, red fluorescent protein; shControl, short hairpin RNA with a scrambled sequence; shE2F1, short hairpin RNA targeting E2F1; shEP300, short hairpin RNA targeting EP300; shVMP1, short hairpin RNA targeting VMP1; TBS, TRIS-buffered saline; TTS, transcriptional starting site; VMP1, Vacuole Membrane Protein 1.

which is activated in high resistant pancreatic tumor cells, carrying oncogenic KRAS, under gemcitabine treatment but not in sensitive cells to chemotherapy. This molecular mechanism includes the activation of E2F transcription factor 1 (E2F1) that binds to VMP1 promoter to enhance VMP1-mediated autophagy. We also identified the histone acetyltransferase EP300 (E1A binding protein p300), as a modulator of this promoter activity. Our data show that the E2F1-EP300 activator/coactivator complex is part of the regulatory pathway controlling VMP1 expression triggered by gemcitabine. Together these data point at E2F1 as a regulatory factor modulating VMP1-mediated autophagy in human pancreatic cancer cells and integrate this degradative cellular process into the complex network of events involved in PDAC chemoresistance.

## MATERIALS AND METHODS

#### Mammalian Cell Lines, Transfections, and Treatments

Human pancreatic cancer cell lines with mutated KRAS, PANC-1 (KRASG12D), and MIAPaCa-2 (KRASG12C), and human pancreatic cancer cell line with wild type KRAS, BxPC-3 (43), and also a human HeLa cell line were obtained from American Type Culture Collection. PANC-1, MIAPaCa-2, and HeLa cells were cultured in Dulbecco's modified Eagle's medium (Biological Industries) containing 10% fetal bovine serum (Natocor). BxPC-3 cells were cultured in RPMI 1640 medium (Biological Industries) containing 10% fetal bovine serum (Natocor). All cell culture mediums were supplemented with 100 U µl <sup>−</sup><sup>1</sup> penicillin, and 100 µg µl −1 streptomycin (Biological Industries). All cell lines were maintained at 37◦C under a humidified atmosphere with 5% CO2. Mycoplasma contamination is periodically checked by PCR, each time a cell line enters the laboratory, and then monthly for each cell line currently in use. Cells were seeded 24 h before transfection and treatments to reach a 60% confluence. Cells were transfected using FuGENE-6 Transfection Reagent (Promega) as indicated by the manufacturer. Gemcitabine (Elli Lilly) and chloroquine (Sigma-Aldrich) were prepared according to the manufacturer's instructions. Cells were treated with 20µM gemcitabine (Elli Lilly) and/or 10µM chloroquine (Sigma) for different times when appropriate.

#### Expression Vectors

Plasmid pRFP-LC3 (microtubule-associated protein 1 light chain 3 fused to red fluorescent protein) was kindly provided by Dr. Maria I. Colombo [Universidad Nacional de Cuyo, Consejo Nacional de Investigaciones Científicas y Técnicas (CONICET), Argentina]. The vector pEGFP-VMP1 (enhanced green fluorescent protein fused to VMP1) was designed and constructed as previously reported (31). Expression vector for E2F1 was kindly provided by Dr. Cánepa (Facultad de Ciencias Exactas y Naturales, Universidad de Buenos Aires, Argentina). The expression vector Flag-EP300 was kindly provided by Dr. Donald Tindall (Mayo Clinic, Rochester, MN, USA). The shRNA targeting EP300 (shEP300) was obtained from Sigma-Aldrich (St. Louis, MO). VMP1 and E2F1 down-regulation was made using the plasmid pCMS3-1p-EGFP, which was kindly provided by Dr D. Billadeau (Department of Immunology, College of Medicine, Mayo Clinic, USA), and contains a separate transcriptional cassette for EGFP to identify transfected cells. A plasmid contains an VMP1 short hairpin RNA (shVMP1) construction (sense 5 ′ -GGCAGAAUAUUGUCCUGUG-3′ , and antisense 5′ -CACA GGACAAUAUUCUCUGCC-3′ ) for VMP1 down-regulation. Another plasmid contains an E2F1 short hairpin RNA (shE2F1) construction (sense 5′ -GACGTGTCAGGACCTTCGT-3′ , and antisense 5′ -CTGCACAGTCCTGGAAGCA-3′ ) for E2F1 downregulation. A vector with a scramble sequence was used as a control short hairpin RNA (shControl).

#### Cell Viability Assay by Trypan Blue Method

Cells were seeded into 6-well-plates at 1 × 10<sup>5</sup> cells per well with 3 ml growth medium. The following day, PANC-1, MIAPaCA-2, and BxPC-3 cells were treated with 20, 200, or 2,000µM gemcitabine for 24, 48, and 72 h. The number of viable cells was then determined by the Trypan blue method. Cells were raised by gentle pipetting into the growth medium, and 0.5 ml of the cell suspension was taken and mixed with 0.5 ml of a Trypan Blue 0.2% w/v in phosphate-buffered saline (PBS) solution and incubated for 3 min at room temperature. Colored (dead) and non-colored (live) cells were counted in a Neubauer chamber. The percentage of viable cells under gemcitabine treatment regarding to control was determined according to the following formula: % Viable cells = NT/NC × 100, where NC is the number of viable cells in the control and NT is the number of viable cells under treatment at the same time of incubation.

## RNA Extraction and Quantitative Real-Time PCR (qPCR)

Cells lines were grown in 6-well-plates at 3.5 × 10<sup>5</sup> cells per well with 3 ml growth medium. After the corresponding transfections and treatments, RNA extraction from cell cultures was performed with the TRIzol reagent (Invitrogen). The concentration of RNA was determined by Vision Life Science Spectrophotometer (Hoefer) and RNA was running in an agarose gel to check its quality. RNA purified (2 µg) were used and treated with 1 µl of RNase-free DNase I (Invitrogen) in a final volume of 10 µl containing 1X DNAse (deoxyribonuclease) buffer. The 10 µl of DNAse treatment was incubated with 100 nM random primers (N6) at 70◦C for 5 min. Then, 1X MMLV buffer, dNTPs (deoxynucleoside triphosphates) to 1 mM and 1 µl of the enzyme MMLV reverse transcriptase (Promega) were added in a final volume of 20 µl, incubated at 25◦C for 5 min and then at 37◦C for 1 h (RT). The RT-PCR reaction was performed on a Techne cycler. For the qPCR, the primers described below were used with 0.5 µl of RT per reaction in a final volume of 25 µl using the Master Mix Real Mix (Biodynamics). The qPCR was performed on RG 6000 cycler (Corvette). The cycling conditions were as follows: 95◦C for 2 min followed by 40 cycles of 95◦C for 20 s, 58◦C for 20 s, and 72◦C for 30 s. Transcript level of VMP1 mRNA was normalized by comparison with mRNA of β-actin and was calculated using the 2−11CT method (44, 45). Specific primers were used for VMP1: forward 5′ -GGTGCTGAACCA GATGATGA-3′ and reverse, 5 ′ -GCACCAAAGAAGGTCCAAA-3 ′ ; and for <sup>β</sup>-actin: forward 5′ -GACTTCGAGCAAGAGATGG-3 ′ and reverse, 5′ -GCACTGTGTTGGCGTACAG-3′ .

#### Western Blot Analysis

Cells were seeded into 60 mm culture dishes at 8 × 10<sup>5</sup> cells per dish with 5 ml growth medium. After different treatments and transfections, cells were lysed in ice-old RIPA buffer (150 mM NaCl, 1% Triton X-100, 1% sodium deoxycholate, 0.1% SDS, 10 mM Tris-HCl pH 7.2, 5 mM EDTA) containing Phosphatase and Protease Inhibitor Cocktail (Sigma–Aldrich). Protein concentration was determined using the bicinchoninic acid (BCA) protein assay reagent (Pierce). Equal amount of protein was analyzed on SDS-PAGE and transferred to polyvinylidene fluoride PVDF membranes (0.22µm pore size, Millipore). The membranes were blocked with Odyssey Blocking Buffer (LI-COR) at room temperature for 1 h and incubated with the corresponding primary antibodies overnight at 4◦C. The primary antibodies used were anti-VMP1 (1:1,000; rabbit mAb #12929, Cell Signaling Technology), anti-E2F1 (1:500; mouse mAb CS204394; Millipore), anti-Actin (1:4,000; rabbit polyAb A2066; Sigma–Aldrich). After incubation, the membrane was washed four times with PBS containing 0.1% Tween-20 (PBST) and twice with PBS, and then incubated with the corresponding IRDye secondary antibody (1:15,000, IRDye <sup>R</sup> 680LT Goat anti-Rabbit IgG or 1:10,000, IRDye <sup>R</sup> 800CW Goat anti-Mouse IgG, LI-COR) in Odyssey Blocking Buffer (LI-COR) for 2 h at room temperature. After, the membrane was washed four times with PBST, twice with PBS, and scanned with Odyssey <sup>R</sup> SA (LI-COR). For LC3 western blotting, the membranes were blocked with 5% (w/v) non-fat dry milk in TRIS-buffered saline (TBS) containing 0.1% Tween-20 (TBST) at room temperature for 1 h and incubated with the primary antibody anti-LC3B (1:1,000, Rabbit mAb #3868, Cell Signaling Technology) overnight at 4◦C. After the incubation, the membrane was washed four times with TBST and twice with TBS, then incubated with anti-rabbit HRP-conjugated (1:3,000, Amersham NA934, GE Healthcare) secondary antibody in TBST with 5% (w/v) non-fat milk for 2 h at room temperature. Next, the membrane was washed four times with TBST and twice with TBS and incubated with PIERCE ELC Plus Western blotting Substrate (Cat# 32134, Thermo Scientific) according to manufacturer's instructions. Finally, the membrane was scanned with cDigit Blot Scanner (LI-COR). We used ImageJ software to determine protein bands density. Relative densitometry normalized to actin is expressed as the mean ± SD of three different experiments.

#### Fluorescence Microscopy

To determine autophagy, cells were growing on glass slides into 24-well-plates. They were seeded at 5 × 10<sup>4</sup> cells per well with 1 ml growth medium. Twenty-four hours later, cells were cotransfected with a red fluorescent protein fused to LC3 (RFP-LC3) expression vector and the indicated plasmid, and then treated with gemcitabine. Next, cells were fixed with 4% pformaldehyde in PBS for 15 min, and immediately washed several times with PBS. Samples were mounted in DABCO (Sigma– Aldrich) and observed using a fluorescence microscope Nikon Eclipse 200 (Plan100), or an inverted LSM Olympus FV1000 using an UPLSAPO 60X O NA: 1.35 objective. We consider a cell positive for autophagy when RFP-LC3 has a punctate staining and not diffused protein remains. The number of fluorescent cells with punctate staining per 100 fluorescent RFP-LC3 transfected cells was determined in three independent experiments. To quantify, the number of fluorescent cells with punctate staining was counted in six random fields representing 100 fluorescent cells and expressed as the mean ± SD of combined results.

## In silico Analysis

Genomic details and characteristics from human VMP1 gene were collected from the Ensembl Genome database. VMP1 promoter prediction was done using the Gene2Promoter utility; transcription factors binding sites and additional information was obtained using RegionMiner, MatBase, and MatInspector tools (Genomatix Software). Supporting evidence was found using the Neural Network Promoter Prediction program (BDGP version 2.2) and FPROM (Softberry). Alibaba 2.1 and PROMO 3.0 were also used for transcription factors' consensus sequences search.

## Cloning of VMP1 Promoter Into Reporter Vector

Genomic DNA from HeLa cells was extracted using TRIzol reagent (Invitrogen). PCR was performed to amplify a fragment of 3,005 bp and then three progressively shorter ones of 1,977, 1,469, and 883 bp from the 5′ upstream region of VMP1 using specific primers and HeLa DNA as template. PCR was done in a final volume of 50 ml as follows: 1× Pfu DNA Polymerase Buffer (with MgSO4), 300 mM dNTPs (deoxynucleosyde triphosphates), 0.4 mM Primers sense/antisense, 50 ng genomic DNA, 1.5U Pfu DNA Polymerase (Promega). Amplification was performed according to an initial denaturation step of 1.5 min at 94◦C, followed by 30 cycles at 94◦C for 30 s, 59◦C for 30 s, 73◦C for 8 min, and a final step at 73◦C for 5 min. PCR products were run in agarose gels and amplicons were recovered with a Gel Band Purification Kit (GE Healthcare). Primers had restriction sites for SalI (Forward) and HindIII (Reverse) enzymes for subsequent cloning of the fragments in pGemT easy vector (Promega). Amplicons integrity was confirmed by sequence analysis (MacroGen). All constructs were then subcloned in the luciferase reporter pGL3 Basic Vector (Promega), with SalI and HindIII enzymes. As a result, the constructs pGL3.vmp1-883, pGL3.vmp1-1469, pGL3.vmp1- 1977, and pGL3.vmp1-3005 were obtained.

## Luciferase Reporter Assays

For luciferase assays cells were plated 24 h before transfection in 12-well-plates at 1.4 × 10<sup>5</sup> cells per well. Cells were used at 60% confluence for pGL3.vmp1 promoter constructs transfection with FuGENE6 Transfection Reagent (Promega). Ratio used in each case was 1.5 ml FuGENE6 per 1 mg DNA. When two plasmids were transfected, we used 0.4 mg of pGL3 reporter and 0.6 mg of expression vector. In shRNA assays we used 1.5 mg of shRNA and 0.5 mg pGL3 reporter vectors. Treatments were done 24 h after transfection. In co-transfection experiments with

to all gemcitabine concentrations tested, and PANC-1 and MIAPaCa-2 cells are sensitive from 200 µM gemcitabine. (B) Cells were treated with 20 µM gemcitabine and cell viability was determined at 0, 12, 24, 48, and 72 h. There were no significant differences within 72 h of treatment in PANC-1 cells, and within 48 h of treatment in MIAPaCa-2 cells. However, a significant reduction in cell viability was observed in BxPC-3 cells from 12 h of treatment. Results are presented as percentage of viable cells compared to the untreated control (mean ± SE; *n* = 3). \**p* < 0.05 vs. untreated cells, \*\**p* < 0.01 vs. untreated cells.

expression vectors, cells were processed 48 h after transfection. Each condition was tested in triplicate. Cells were washed with cold PBS and lysed with 100 ml of 1× Cell Culture Lysis Buffer (Promega). 96-well-plates were used for activity assays. In each well 40 µl of luciferase substrate was added to 20 µl of lysate and a 5-s reading was done for luminescence measurement in a Victor3 1420 multilabel counter (PerkinElmer). Results were normalized to protein concentration. Relative light units (RLU) value was calculated as luciferase activity/protein concentration.

## Chromatin Immunoprecipitation (ChIP) Assay

Chromatin immunoprecipitation was conducted following the Pierce Agarose Chip kit (Thermo Scientific). Briefly, PANC-1 cells were culture into 100 mm culture dishes at 3 × 10<sup>6</sup> cells per dish with 13 ml growth medium. The next day, cells were treated with 20µM gemcitabine and 24 h were cross-linked with 1% formaldehyde directly into the media for 10 min at room temperature. The cells were then washed and scraped with PBS and collected by centrifugation at 800 × g for 5 min at 4◦C, resuspended in cell lysis buffer and incubated on ice for 15 min. The pellet was then resuspended in nuclear lysis buffer and sheared to fragment DNA to about 700 bp. Samples were then immunoprecipitated using a E2F1 antibody or normal rabbit IgG (Millipore) overnight at 4◦C on a rotating wheel. Following immunoprecipitation, samples were washed and eluted using the chromatin immunoprecipitation kit in accordance with the manufacturer's instructions. Cross-links were removed at 62◦C for 2 h followed by 10 min at 95◦C and immunoprecipitated DNA was purified and subsequently amplified by PCR. PCR was performed using seven primer sets for the seven areas containing potential E2F1 binding sites in the VMP1 promoter sequence: forward, 5′ -GCAT CTCACTTTGTCACCCAG-3′ and reverse, 5′ -ACTTGAGG TCAGGAGTTCGAGAC-3′ ; (2) forward, 5′ -CAGGCTGTTCTC AAACTCCTGG-3′ and reverse, 5′ -GCACCATACTAGACTC TGGGA; (3) forward, 5′ -TCCCAGAGTCTAGTATGGTGC-3 ′ and reverse, 5′ -CGATATCGCTCCATTGCTCTCCA-3′ ; (4) forward, 5 ′ -GAGTAGCTGGGATTACAGGC-3 and reverse, 5′ - ACCTGAGGTCAGAAGTTCGAGAC-3′ ; (5) forward, 5′ -GTCT CGAACTTCTGACCTCAGGT-3′ and reverse, 5′ -CAGCTGGG CACTTATGAATATCCC-3′ ; (6) forward, 5′ -GATATTGGTCTC CTTCGCCCTGT-3′ and reverse, 5′ -GCAAGAGGAAGAATGA CTGCTC-3′ ; and (7) forward, 5′ -GAGCCTAACTGAAATCCCG CGA-3′ and reverse, 5′ -CAAGCTCTGAGGACAGCCTCA-3. PCR products were visualized on a 2% agarose gel.

#### Statistical Analysis

Data are expressed as mean ± SD. We performed a minimum of three independent experiments, where individual data points were based on at least technical duplicates each. Student's t-test was used for comparisons between two groups and ANOVA test to assess more than two groups. P < 0.05 were considered statistically significant. Statistical analysis of data was performed using GraphPad Prism 6.

## RESULTS

## Gemcitabine Requires VMP1 Expression to Induce Autophagy in Pancreatic Cancer Cells Carrying Oncogenic KRAS

In order to analyze the time course effect of gemcitabine treatment on VMP1 expression we used PANC-1 and MIAPaCa-2 pancreatic tumor cells harboring a KRAS activating mutation,

that are highly resistant to chemotherapy, and BxPC-3 pancreatic tumor cells that do not carry KRAS mutation. The relative sensitivity of pancreatic cancer cells to gemcitabine treatment was analyzed. PANC-1, MIAPaCa-2, and BxPC-3 cells were treated with 20, 200, and 2,000µM gemcitabine. The relative number of viable cells was determined by the trypan blue dye exclusion test 24 h later. **Figure 1A** shows cell viability as a percentage relative to control according to the gemcitabine dose. BxPC-3 cells were more sensitive in all doses of gemcitabine analyzed (**Figure 1A**). However, cell viability was significantly reduced from 200 µM gemcitabine in PANC-1 and MIAPaCa-2 cells (**Figure 1A**). Thus, gemcitabine 20 µM is the lowest tested dose in which BxPC-3 cells are sensitive, and PANC-1 and MIAPaCa-2 are resistant, at least 24 h after treatment. In view of this, we analyzed cell viability of PANC-1, MIAPaCa-2, and BxPC-3 cells under 20 µM gemcitabine treatment during 72 h. There were no significant differences during the 72 h of treatment for PANC-1 cells and up to 48 h of treatment for MIAPaCa-2 cells. On the other hand, BxPC-3 cells showed a significant reduction in cell viability from 12 h of treatment (**Figure 1B**). These results indicate that at 12 and 24 h with 20 µM gemcitabine, PANC-1 and MIAPaCa-2 cells are resistant, and by comparison, BxPC-3 cells are sensitive to treatment. Consequently, from now on we use this dose of gemcitabine for the following experiments.

Next, PANC-1 and MIAPaCa-2 cells were incubated with 20µM gemcitabine for 24 h and we evaluated VMP1 mRNA expression by qPCR assays. A significant induction of VMP1 mRNA was found after 12 and 24 h of gemcitabine treatment in PANC-1 and MIAPaCa-2 cells (**Figure 2A**). LC3 is currently used as a specific marker of autophagy (46). During the autophagic process, the cytosolic form of LC3 (LC3-I) undergoes C-terminal proteolytic and lipid modifications (LC3-II) and translocates from the cytosol to the autophagosomal membrane (47, 48). Then, we analyzed VMP1 expression and LC3 lipidation to determinate autophagy by western blot. **Figure 2B** shows that gemcitabine induced VMP1 protein expression and LC3-II formation, and therefore autophagy, in PANC-1 and MIAPaCa-2 cells. The quantification of western blots is shown in **Figure 2C**.

Afterward, PANC-1 and MIAPaCa-2 cells were co-transfected with RFP-LC3 and shVMP1 expression vectors and treated with 20µM gemcitabine for 24 h. A diffuse pattern of LC3- RPF expression in the cytoplasm indicates that autophagy is not occurring, and when autophagy is induced RFP-LC3 is relocated as dots indicating the formation of autophagosomes. We counted cells with punctate RFP-LC3 staining to determinate autophagy (46). **Figure 2D** shows that gemcitabine induced RFP-LC3 redistribution to autophagosomes was significantly reduced when VMP1 expression was down-regulated in both cell lines. **Figure 2E** shows downregulation of VMP1 by shVMP1 expression in PANC-1 and MIAPaCa-2 cell lines. Therefore, gemcitabine requires VMP1 expression to induce autophagy in pancreatic tumor cells harboring a KRAS activating mutation.

#### Gemcitabine Induces E2F1 Activation of VMP1-Mediated Autophagy Only in Pancreatic Tumor Cells That Carry Oncogenic KRAS

Following, in order to analyze VMP1 expression and autophagy in cells that carry wild type KRAS, BxPC-3 cells were incubated with 20µM gemcitabine for 24 h. Interestingly, gemcitabine treatment did not increase VMP1 mRNA and protein levels compared to basal conditions in BxPC-3 cells (**Figures 3A,B**). Then, we analyzed LC3 lipidation to determinate autophagy, and western blot analyses shows that gemcitabine treatment did not induce autophagy evidenced by LC3-II formation in BxPC-3 cells (**Figure 3B**). In addition, choloroquine treatment was used to evaluate autophagic process by inhibiting autophagy flux (46). The use of chloroquine alone or in combination with gemcitabine treatment induced LC3-II accumulation compared to control cells or cells only treated with gemcitabine in the same proportion, respectively, indicating that gemcitabine did not interrupt autophagy flux in these cells. The quantification of western blots is shown in **Figure 3C**. These results suggest BxPC-3 cells have a basal VMP1 expression and autophagy that are not up-regulated by gemcitabine.

E2F transcription factors are involved in cell proliferation and DNA repair (49). As gemcitabine incorporation into DNA is critical for its toxicity (50), we evaluated E2F1 expression in response to gemcitabine treatment. To characterize this response of E2F1 to gemcitabine, we chose the sensitive cell line BxPC-3 and the most resistant to treatment by comparison, PANC-1 cells. Western blot assay shows E2F1 protein levels were significantly increased after 12 and 24 h of gemcitabine treatment in PANC-1 cells, and they were similar with respect to control in treated BxPC-3 cells (**Figure 3D**). In consequence, E2F1 is a candidate to mediate increased VMP1 expression and autophagy in response to gemcitabine in PANC-1 but not in BxPC-3 cells.

Considering that VMP1 expression induces autophagosome formation (31), we tested whether E2F1 expression is capable of inducing VMP1 expression and autophagy in PANC-1 cells. We performed qPCR on samples from cells transfected with an expression vector for E2F1. As seen in **Figure 3E**, E2F1 expression induced VMP1 mRNA expression in PANC-1 tumor cells. Then, PANC-1 cells were concomitantly transfected with an expression plasmid encoding for the RFP-LC3 fusion protein and E2F1 expression vector. **Figure 3F** shows the recruitment of LC3 fusion protein in punctate structures in E2F1-transfected cells in contrast to the diffuse RFP-LC3 fusion protein signal observed in control cells. Quantitation showed that recruitment of LC3 was significantly increased in cells expressing E2F1 compared to control cells (**Figure 3F**). These results demonstrate that E2F1 is capable of inducing VMP1 expression and autophagy in pancreatic tumor cells resistant to gemcitabine treatment.

#### VMP1 Promoter Is Activated by Gemcitabine

We have previously demonstrated that starving conditions and rapamycin treatment induce VMP1 expression in HeLa cells (31). On the other hand, VMP1 expression is activated in PANC-1 human tumor cells carrying mutated (G12D) KRAS (28). In order to study the molecular mechanism that regulates VMP1 expression in the context of gemcitabine induced-autophagy in pancreatic tumor cells, a 3,005 bp sequence of the 5′ upstream region of the human gene VMP1 was amplified and cloned in the pGL3 reporter vector (pGL3.vmp1-3005) (**Figure 4A**). Following, we analyzed if this sequence cloned has a promoter activity. In these experiments, we used HeLa cells under starving conditions and rapamycin treatment as a positive control of autophagy induction, and the pGL3.vmp1-3005 construct was used to perform luciferase reporter assays. As a result, we found increased VMP1 mRNA expression and VMP1 promoter activity in response to starving conditions and to rapamycin treatment in PANC-1 and HeLa cells (**Figures 4B,C**). Next, we analyzed if gemcitabine was able to increase VMP1 promoter activity. **Figure 4D** shows that the activity of the 3,005 bp sequence of VMP1 promoter was significantly increased when PANC-1 and HeLa cells were treated with 20µM gemcitabine for 24 h.

In order to localize the essential promoter sequence in the 3,005 bp sequence of the 5′ upstream region of the human gene VMP1, we amplified and subcloned consecutive shorter fragments of this region. Three more constructs were created and named as follows: pGL3.vmp1-1977, pGL3.vmp1-1469, and pGL3.vmp1-883 (**Figure 4E**). Luciferase activity assays were performed in HeLa cells transfected with each construct and submitted to starvation, rapamycin, or gemcitabine treatment. Relative promoter activity was analyzed for each sequence (**Figure 4F**). Results showed a decreased activation for the pGL3.vmp1-1977 and pGL3.vmp1-1469 constructs comparing to the initial 3,005 bp sequence. On the other hand, the activity was increased when the shorter pGL3.vmp1-883bp construct was used. The same results were observed in all the conditions analyzed. These data suggest that essential regulation motifs involved in VMP1 expression are contained into this VMP1 shorter promoter region of 883 bp.

#### E2F1 Directly Activates the VMP1 Promoter and Regulates VMP1 Expression Under Gemcitabine Treatment

Considering that maximal promoter activity in luciferase reporter assays was observed for the pGL3.vmp1-883 construct, we analyzed this sequence using bioinformatics tools. This in silico analysis data showed a putative promoter in this sequence containing a TATA. In addition, we identified several putative binding sites for relevant transcription factors related to the cellular stress response including E2F1 (**Figure 5A**).

In order to know if VMP1 gene is a direct target of E2F1, we used a combination of transcriptional and chromatin assays to determine a possible involvement of E2F1 transcription factor in VMP1 promoter activation. First, we performed luciferase reporter assays using the shorter sequence of VMP1 promoter, pGL3.vmp1-883bp construct. Reporter studies demonstrate that expression of E2F1 led to an increase in VMP1 promoter activity (**Figure 5B**). Moreover, endogenous E2F1 can bind to VMP1 promoter in PANC-1 cells treated with gemcitabine as demonstrated by ChIP assay (**Figure 5C**). Also, gemcitabine increased VMP1 promoter activity (see shControl transfected PANC-1 cells) (**Figure 5D**). Moreover, down-regulation of E2F1 expression in PANC-1 cells treated with gemcitabine significantly reduced VMP1 promoter activity (**Figure 5D**) and VMP1 expression (**Figure 5E**). These results demonstrate that VMP1 is a novel direct target of the E2F1 transcription factor under gemcitabine treatment.

## E2F1 and EP300 Cooperate in VMP1 Promoter Activation

Transcription factors regulate gene expression through their inherent activation or repression properties, and through functional interactions with co-regulatory molecules. Here, we tested whether activation by E2F1 involves the histone acetyltransferases EP300. First, we analyzed if EP300 could induce VMP1 mRNA expression in PANC-1 cells. **Figure 6A**, shows that VMP1 mRNA expression was up-regulated in PANC-1 cells expressing EP300. Additionally, EP300 expression significantly activated the VMP1 promoter compared to control cells (**Figure 6B**). Then, we evaluated if EP300 participates

cloned into the pGL3 reporter vector (pGL3.vmp1-3005) driving Luciferase (Luc) expression. (B) qPCR assay showed an increase of VMP1 mRNA expression in HeLa and PANC-1 cells submitted to starving conditions and rapamycin treatment for 4 h (*n* = 3). Next, HeLa and PANC-1 cells were transfected with pGL3.vmp1-3005. (C) Relative Luciferase activity shows an increase of *VMP1* promoter activation in HeLa and PANC-1 cells submitted to starving (Stv) conditions and rapamycin (Rapa) treatment. (*n* = 3). (D) Relative Luciferase activity shows an increase of *VMP1* promoter activation in HeLa and PANC-1 cells submitted to gemcitabine treatment for 24 h (*n* = 3). (E) Consecutive shorter fragments of 3005 bp sequence of *VMP1* promoter were amplified and subcloned in the pGL3 reporter vector (pGL3.vmp1-1977, pGL3.vmp1-1469, pGL3.vmp1-883). (F) HeLa cells were submitted to starvation conditions, rapamycin and gemcitabine treatment for 24 h. Relative Luciferase activity showed higher levels of *VMP1* promoter activation in the shorter construct (pGL3.vmp1-883) compared to 3,005 bp sequence of *VMP1* promoter in all treatments. \**p* < 0.05.

with E2F1 in VMP1 promoter activation. Interestingly, downregulation of EP300 impaired E2F1-mediated activation of the VMP1 promoter in PANC-1 cells (**Figure 6C**). Furthermore, co-expression of E2F1 and EP300 led to a synergistic activation of the VMP1 promoter (**Figure 6D**). These findings demonstrate that E2F1 and EP300 cooperates in VMP1 promoter regulation.

Altogether, these data show that the E2F1-EP300 activator/coactivator complex is part of the novel signaling pathway controlling the promoter activity and, consequently, the expression of the autophagy-related gene VMP1 in pancreatic tumor cells carrying oncogenic KRAS.

#### DISCUSSION

Tumor cells with a high prevalence of KRAS activating mutations, like pancreatic cancer, have the distinction of a poor prognosis (4). Previously, it has been demonstrated that many human cancer cell lines with KRAS activating mutations have basal levels of autophagy (51, 52). Yang et al. (6), have showed that pancreatic cancer cells exhibit constitutive autophagy under basal conditions, and it is increased in the advanced stages of PDAC being required for malignant transformation. In a previous work, we have identified VMP1 as a transcriptional target of oncogenic KRAS signaling

pathway and demonstrated that KRAS requires VMP1 to induce and maintain basal autophagy in pancreatic tumor cells (28). On the other hand, VMP1-mediated autophagy is early induced above basal conditions by gemcitabine treatment in MIAPaCa-2 cells (36). Moreover, VMP1 is highly expressed in poorly differentiated human pancreatic cancer (41). In this study, we identified a regulatory pathway, which is activated by gemcitabine treatment, in pancreatic tumor cells carrying a KRAS mutation at amino acid position 12. This molecular mechanism involves a novel E2F1-EP300- VMP1 pathway controlling VMP1 expression triggered by gemcitabine.

According to the Ensembl Genome database, VMP1 gene is localized in Chromosome 17 of the human genome. Using informatics tools, we found out four putative promoter regions given for VMP1 gene; among them we particularly focus on one of 821 bp localized in the positive strand at the following position 57784363–57785183 of chromosome 17. In the analysis, it was also found the proposed transcriptionstarting site (TSS) (**Figure 5A**) at position 501 of the promoter area. This prediction would regulate the expression of a 12 exons transcript and we consider this transcript to be the one coding for the 406 aa protein VMP1. The TSS is located at the start of the first exon sequence, while the starting codon (ATG) is in the second exon. Considering these data, we amplified and cloned a 3,005 bp sequence of the 5′ upstream regions of the VMP1 gene sequence which includes the promoter area analyzed above. Using luciferase reporter assays we demonstrated that this region is activated by previously reported VMP1 stimuli, such as rapamycin, starvation, and gemcitabine (**Figure 4**). We found that regulatory motifs involved in VMP1 expression are contained into an 883 bp fragment in the 5′ upstream region of the VMP1 gene. This essential promoter sequence showed the highest activation

by rapamycin, starvation, and gemcitabine treatment using luciferase reporter assays.

The in silico analysis of VMP1 essential promoter sequence revealed the presence of binding sites for several transcription factors related to cellular stress. We focus our attention on E2F1 because it has been correlated with high-grade tumors and unfavorable patient survival in PDAC (53, 54). E2F1 was the first identified member of the E2F family of transcription factors. E2F activity is linked to retinoblastoma tumor suppressor (RB)–dependent cell-cycle control. E2F transcription factors are found downstream of growth factor signaling cascades, acting as transcriptional activators or repressors of genes necessary for cell cycle progression (55). Most human tumors harbor the functionally inactivated retinoblastoma protein, resulting in deregulated E2F1 and its target genes are highly up-regulated in these transformed cells (56). This up-regulation leads to the activation of cytoplasmic (PIK3CA/AKT and RAS/MAPK/ERK) and nuclear signaling cascades related to invasion and metastasis (54). Also, activation of E2F1 transcription factor has been shown to induce autophagy (57) by up-regulating the expression of the autophagy genes LC3, ULK1 (unc-51 like autophagy activating kinase 1), ATG5 and DRAM1 (DNA damage regulated autophagy modulator 1). The E2F1-mediated induction of LC3, ULK1, and DRAM1 is direct (through interaction with the promoter), whereas the up-regulation of the expression of ATG5 is indirect (58). In this work, we provide evidence of another gene related to autophagy that is up-regulated by E2F1. We demonstrated that E2F1 is able to induce autophagy in pancreatic tumor cells and regulates VMP1-mediated autophagy by a direct binding to VMP1 promoter.

Gemcitabine inhibits DNA synthesis via a process called masked chain termination where gemcitabine is incorporated into DNA via DNA polymerase α, leading to the inhibition of DNA repair and synthesis (59). It has been shown that E2F1 induces genes involved in DNA repair in normal cells and in tumor cells undergoing chemotherapy through complex formation on the promoters of these genes (54). In this sense, BxPC-3 cells that are sensitive to the dose of 20 µM gemcitabine did not increase the expression of E2F1 or VMP1 during treatment. In contrast, PANC-1 cells resistant to treatment increased the expression of E2F1, result consistent with data of previous works (60, 61). Lai et al. (60) have demonstrated that PANC-1 cells respond to gemcitabine by increasing the expression of ribonucleotide reductase M2 catalytic subunit (RRM2) through E2F1-mediated transcriptional activation, as a DNA damage response to enhance DNA repair capacity in these cells. Here, we demonstrated E2F1 and VMP1 expressions are both increased in PANC-1 cells treated with gemcitabine. Besides, pancreatic tumor cells transfected with an expression vector for E2F1 induced VMP1 expression and activated autophagy. Therefore, in this study we demonstrated another mechanism activated by E2F1 in response to chemotherapy, in which E2F1 activates VMP1 expression and autophagy as a resistance response to gemcitabine.

Autophagy is constitutively activated in oncogenic KRASdriven tumors and is necessary for the development of these tumors (6). Previously, we identified the PI3KCA (phosphatidylinositol-4,5-biphosphate 3-kinase catalytic subunit alpha)-AKT1 (AKT serine/threonine kinase 1) pathway as the signaling pathway mediating the expression and promoter activity of VMP1 in KRAS driven tumors (28). PANC-1 cells harbor a KRAS mutated allele (KRAS G12D) and these cells present basal expression of VMP1 and autophagy. In this work, we show that gemcitabine treatment increased that basal VMP1 expression and autophagy in PANC-1 cells. As well as GLI3 (GLI family zinc finger 3) is the transcription factor implicated in VMP1 promoter activation in basal conditions in PANC-1 cells (28), E2F1 plays that function under gemcitabine treatment. Here, we show that down-regulation of E2F1 reduced VMP1 expression and VMP1 promoter activity induced by gemcitabine. Besides, we demonstrated a direct binding of E2F1 on VMP1 promoter under gemcitabine treatment in pancreatic tumor cells. Therefore, these data suggest VMP1 expression is regulated by different transcription factors depending on the cellular context that autophagy is induced. Further research will be necessary to clarify if function or mechanisms involved in basal VMP1 induced autophagy differs from VMP1-induced autophagy under gemcitabine treatment.

The regulation of gene expression depends on the characteristic activation/repression properties of each transcription factor, but also on functional interactions with co-regulatory molecules. EP300 belongs to the type 3 family of lysine acetyltransferases (KAT3) (62), and this enzyme is involved in the regulation of important physiological processes such as proliferation, differentiation, and apoptosis, due to its ability to function as transcriptional coactivator interacting and regulating more than 400 transcription factors (63, 64). However, the role of EP300 in gene regulation is not only restricted to its property of allowing the binding of transcription factors to large protein complexes in the transcription machinery, but also implies the required KAT activity for the acetylation of transcription factors and histones that allow access to chromatin (65). Thus, EP300 contributes to DNA repair through histone acetylation, facilitating the recruitment of DNA repair factors to the site of damage (66). We demonstrated that EP300 potentiates the VMP1 promoter activation by E2F1. Even, down-regulation of EP300 impaired E2F1-mediated activation of the VMP1 promoter in pancreatic tumor cells. These findings demonstrate that E2F1 and p300 cooperate in VMP1 promoter regulation.

Our results agree with Hashimoto et al. (67), who have shown that autophagy has a cytoprotective effect against 5-fluorouracil and gemcitabine in pancreatic cancer cells. They demonstrated that inhibition of autophagy potentiates the inhibition of PANC-1 cell proliferation by 5-fluorouracil and gemcitabine. Here, we showed an induction of VMP1 expression and autophagy in PANC-1 and MIAPaCa-2 cells under gemcitabine treatment and down-regulation of VMP1 expression significantly reduced autophagy induced by gemcitabine. These data strongly suggest that VMP1 expression is involved in PDAC chemoresistance to gemcitabine.

In conclusion, we have identified the E2F1-EP300- VMP1 pathway that mediates gemcitabine-induced autophagy in pancreatic cancer cells (**Figure 6E**). This pathway would be activated by gemcitabine like a resistance mechanism. Our results point at E2F1 as a regulatory factor modulating gemcitabine induced VMP1 mediated autophagy in human pancreatic cancer cells and mechanistically integrate the autophagic degradative process into the complex network of events involved in PDAC chemoresistance.

#### DATA AVAILABILITY STATEMENT

The datasets generated for this study are available on request to the corresponding author.

## AUTHOR CONTRIBUTIONS

AR designed and performed experiments, analyzed and interpreted data, prepared figures, and wrote the manuscript. CC performed experiments, contributed to interpret data, and prepared figures. FR performed experiments. VB developed analytical tools. TO developed expression vectors. CG contributed to interpret data. MV developed the hypothesis, designed experiments, analyzed, and interpreted data, and wrote the manuscript. All authors reviewed the manuscript.

#### FUNDING

This work was supported by grants from: Consejo Nacional de Investigaciones Científicas y Técnicas (CONICET) [PIP 2015-2017 GI−11220150100160CO]; Agencia Nacional de Promoción Científica y Tecnológica (ANPCyT) [PICT2013- 2048; PICT2016-1032] and Universidad de Buenos

#### REFERENCES


Aires [UBACYT 2014-2017 GC−20020130100764BA; UBACyT 2018–2020−20020170100082BA].

## ACKNOWLEDGMENTS

We thank all IBIMOL staff for the support received. We are grateful to Roxana Resnik for her assistance in English. We thank to Sociedad Argentina de Gastroenterología (SAGE) the permission to use material presented at the Pre AGA 2015 conferences (68).


**Conflict of Interest:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2020 Ropolo, Catrinacio, Renna, Boggio, Orquera, Gonzalez and Vaccaro. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Increased Expression of Autophagy Protein LC3 in Two Patients With Progressing Chronic Lymphocytic Leukemia

Daniela S. Arroyo1,2†, Cecilia M. Rodriguez 1,2†, Claudio Bussi <sup>3</sup> , Clarisa Manzone-Rodriguez 2,4, Darío Sastre<sup>1</sup> , Viviana Heller <sup>1</sup> , Carmen Stanganelli <sup>5</sup> , Irma Slavutsky <sup>6</sup> and Pablo Iribarren2,4 \*

<sup>1</sup> Facultad de Ciencias Médicas, Hospital Nacional de Clínicas, Universidad Nacional de Córdoba, Córdoba, Argentina, <sup>2</sup> Departamento de Bioquímica Clínica, Facultad de Ciencias Químicas, Universidad Nacional de Córdoba, Córdoba, Argentina, <sup>3</sup> The Francis Crick Institute, London, United Kingdom, <sup>4</sup> Centro de Investigaciones en Bioquímica Clínica e Inmunología (CIBICI-CONICET), Córdoba, Argentina, <sup>5</sup> Patología Molecular, Instituto de Investigaciones Hematológicas, Academia Nacional de Medicina, Buenos Aires, Argentina, <sup>6</sup> Laboratorio de Genética de Neoplasias Linfoides, Instituto de Medicina Experimental, CONICET-Academia Nacional de Medicina, Buenos Aires, Argentina

#### Edited by:

Maria Ines Vaccaro, University of Buenos Aires, Argentina

#### Reviewed by:

Mario Rossi, Austral University, Argentina Prathibha Ranganathan, Centre for Human Genetics, India

> \*Correspondence: Pablo Iribarren piribarren@unc.edu.ar

†These authors have contributed equally to this work

#### Specialty section:

This article was submitted to Cancer Endocrinology, a section of the journal Frontiers in Endocrinology

Received: 29 May 2019 Accepted: 27 April 2020 Published: 22 July 2020

#### Citation:

Arroyo DS, Rodriguez CM, Bussi C, Manzone-Rodriguez C, Sastre D, Heller V, Stanganelli C, Slavutsky I and Iribarren P (2020) Increased Expression of Autophagy Protein LC3 in Two Patients With Progressing Chronic Lymphocytic Leukemia. Front. Endocrinol. 11:321. doi: 10.3389/fendo.2020.00321 Chronic lymphocytic leukemia (CLL) is the most common type of adult leukemia in the western hemisphere. It is characterized by a clonal proliferation of a population of CD5+ B lymphocytes that accumulate in the secondary lymphoid tissues, bone marrow, and blood. Some CLL patients remain free of symptoms for decades, whereas others rapidly become symptomatic or develop high-risk disease. Studying autophagy, which may modulate key protein expression and cell survival, may be important to the search for novel prognostic factors and molecules. Here, we applied flow cytometry technology to simultaneously detect autophagy protein LC3B with classical phenotypical markers used for the identification of tumoral CLL B cell clones. We found that two patients with progressing CLL showed increased expression of the autophagy protein LC3B, in addition to positive expression of CD38 and ZAP70 and unmutated status of IGHV. Our data suggest that activation of autophagy flux may correlate with CLL progression even before Ibrutinib treatment.

#### Keywords: chronic lymphocytic leukemia, autophagy, LC3, progressing, cancer

## INTRODUCTION

Chronic lymphocytic leukemia (CLL) represents 25% of all leukemias and 1.3% of all cancers in the western hemisphere, but it has lower prevalence in Asia (1, 2). It is characterized by the expansion of a population of monoclonal CD5+ B lymphocytes that accumulate in blood, bone marrow, and secondary lymphoid tissues (3, 4). Chronic lymphocytic leukemia frequently presents with adverse prognostic features in older patients, significantly affecting their survival (1).

The average age at the time of diagnosis is around 70 years old, though, rarely, it is seen in people under age 40. Patients with CLL may be asymptomatic and may initially be diagnosed by the detection of lymphocytosis on a routine complete blood cell count (1). Sometimes the patients develop anemia, thrombocytopenia, lymphadenopathy, and/or hepatosplenomegaly (5). Other cases may present fever, fatigue, night sweats, and weight loss (1). As we describe in a previous report (4), chronic lymphocytic leukemia can present two clinical forms, aggressive and indolent. The worst prognosis is associated with CD38 expression, high expression levels of ZAP-70, and the absence of mutations in the immunoglobulin heavy chain variable (IGHV) genes (4, 6, 7). Discrimination of patients with different outcomes can be done by finding chromosomal alterations, which are normally detected in >80% of cases (4, 8). The most frequent genomic aberration (deletions at chromosome 13q14) can be detected by FISH analysis. Most of the CLL patients show constitutively elevated expression of Bcl-2, indicating a role for resistance to apoptosis in the disease pathogenesis.

The particular ability of autophagy to promote cell survival during metabolic stress or cell death as a result of an imbalance in cell metabolism, where autophagic cellular consumption exceeds the cellular capacity for synthesis, is a promising avenue for cancer therapy (9). As described by Bologna et al. (3), autophagy is activated in leukemia cells upon treatment with different chemotherapeutic agents, inducing cell death. In particular, many currently used drugs for CLL, including fludarabine, dexamethasone, idelalisib, and Bcl-2 antagonists, have been suggested to have an autophagy-mediated effect (3).

It was proposed that conventional CLL prognostic markers like genetic mutations, the mutation status of the IGHV, and expression of ZAP-70 and CD38, have predictive value for the responses to first-line therapy in CLL (4, 10). Many initiatives have aimed at integrating all of the prognostic factors defined previously into a single prognostic score (11, 12). As stated by Strati et al. (1), it has recently been suggested that "published evidence is sufficient to recommend that FISH and IGHV analysis be performed as standard clinical tests for all patients with newly diagnosed CLL in those countries with the resources to do so" (1, 13). Recently, there was a report of differential expression of at least 20 miRs in B cells from progressive CLL patients compared to non-progressive CLL controls (14). Several of these miRs promote resistance to apoptosis and/or progression of neoplastic B cell clones. Therefore, studying cell responses able to modulate miR expression and cell survival, such as autophagy, may be important for the search for novel prognostic factors and molecules (14).

## BACKGROUND

We present two cases of progressing CLL; the characteristics of the patients are summarized in **Table 1**. Two additional nonprogressing CLL patients are included for comparison purposes. These patients were included in our previous study (4). Here, we describe the patients accordingly to onset, diagnosis, prognostic markers, and evolution.

#### Patient #1

A 58-year-old man was diagnosed with CLL in February 2008. At that time, routine analysis showed a total white blood cell count of 34 × 10<sup>9</sup> /L (normal range: 4.5–10 × 10<sup>9</sup> /L), 82% of lymphocytes (normal range: 20–45%), with splenomegaly and no anemia or thrombocytopenia. Lactate dehydrogenase (LDH) and B2-microglobulin (B2M) values were 312 UI/L and 4.85 mg/L, respectively (reference values: LDH: 180–450 UI/L; B2M: 0.8– 2.20 mg/L). Flow cytometry analysis revealed a clonal B cell population with a typical CLL phenotype. Prognostic marker analysis showed 58% of ZAP-70- and 80% of CD38-positive cells (15). Thus, a diagnostic of CLL Rai stage II was done, and the patient was treated with bendamustine from 2009 to 2016 because progressive systemic symptoms were present. At that moment, genetic markers were: 17p deleted 4.7 %, 11q deleted 62.5%, 13q14 14.4%, and the IGHV gene was unmutated. A stringent follow-up was adopted until July 2016, when he arrived at our Center because of progression with lymphadenopathy, hepatosplenomegaly, and systemic symptoms. At this time, he received targeted drug treatment with Ibrutinib for 16 months, achieving partial response. Finally, he progressed to Richter Syndrome, and the patient died 1 month later.

#### Patient #2

A 51-year-old man was diagnosed with CLL in February 2010. At that time, routine analysis showed a total white blood cell count of 51.7 × 10<sup>9</sup> /L (normal range: 4.5–10 × 10<sup>9</sup> /L), 91% of lymphocytes (normal range: 20–45%), with hepatosplenomegaly and no anemia or thrombocytopenia. Lactate dehydrogenase (LDH) and B2-microglobulin (B2M) values were 257 UI/L and 3.6 mg/L, respectively (reference values: LDH: 180–450 UI/L; B2M: 0.8–2.20 mg/L). Flow cytometry analysis revealed a clonal B cell population with a typical CLL phenotype, expressing prognostic markers ZAP-70 and CD38 (15). Thus, a diagnostic of CLL Rai stage II was done, and progression was detected in October 2010. In March 2011, first-line treatment of six cycles with Fludarabine, Cyclophosphamide, and Rituximab was administered, followed by two additional cycles in May 2015. Genetic markers were: 17p deleted 20%, 13q 53.9%, 11q normal, and the IGHV gene was unmutated. Relapse was observed in 2014 with profuse sweating and progression to stage IIB. In 2016, further progression to stage IV was observed with weight loss, hepatosplenomegaly, lymphadenopathy, asthenia, sweating, and low platelet count. Ibrutinib treatment was initiated in May 2016, and the patient died in March 2018.

## Patients #3 and #4

Patients 3 and 4 presented non-progressing CLL, RAI stages I and 0, respectively, mutated IGHV gene, and negative for CD38 and ZAP70. They were diagnosed in 2006 and 2009, respectively. They subsequently maintained the typical nonprogressive phenotype, and both are still alive (**Table 1**).

## LC3 Expression in Cells From CLL Patients

At the moments of progression of patients 1 and 2, we evaluated autophagy protein LC3B expression in CLL cells by flow cytometry (FlowCellectTM, MilliporeSigma, Darmstadt, Germany). Peripheral blood mononuclear cells (PBMC) from patients with CLL were cultured in the presence or the absence of bafilomycin and then stained using antibodies specific for CD5, CD19, and LC3B. Chronic lymphocytic leukemia B cells were identified and gated using the combination of CD5 plus CD19 staining, for the evaluation of LC3B expression. Lipidated membrane-located LC3B fraction (LC3BII) was accumulated in

#### TABLE 1 | Characterization of the CLL patients.


Chronic lymphocytic leukemia was diagnosed according to standard clinical and laboratory criteria. At the time of analysis, all patients were free from clinically relevant infectious complications. For all in vitro studies, written and informed consent was obtained from patients in accordance with the Declaration of Helsinki. CLL#1 and CLL#2 represent a subgroup of patients with adverse prognostic factors, and CLL#3 and CLL#4 represent patients with more favorable prognostic factors. M, male; F, female; UM, unmutated; MM, mutated.

FIGURE 1 | LC3B expression in CLL cells. (A) PBMC cells from patients with CLL were treated with 25 nM bafilomycin (Baf) or medium (untreated control) for 24 h. Flow cytometric analysis of LC3B expression in CLL cells is shown. (i) Dot plot of CD19 vs. side scatter (SSC) of B lymphocytes from CLL patients. (ii) Dot plot of CD19 vs. CD5 shows clonal CLL cells and residual T cells. (iii) Histogram showing LC3B expression (MFI) in clonal B cells treated with Baf and untreated cells. (B) LC3B index B/M was calculated by dividing MFI of LC3B of CD19+CD5+ (clonal B cells) treated with bafilomycin by MFI of LC3B of untreated clonal B cells (\*p < 0.05, unpaired t-test). (C) Automatic population separator (APS) shows two populations: LC3B+ CLL cells from two patients with unmutated IGHV (Purple) and LC3B+ CLL cells from two patients with mutated IGHV (Green). (D) PBMC cells from all four patients (CLL#1, Patient #1; CLL#2, Patient #2; CLL#3, Patient #3; CLL#4, Patient #4) were treated with bafilomycin 25 nM for 2 h. After that, MFI of LC3B was determined by flow cytometry, as described in (A).

the presence of bafilomycin (**Figure 1A**). Samples from nonprogressing (mutated IGHV) patients 3 and 4 were also included as controls.

We observed higher expression of LC3BII in cells from unmutated progressing CLL patients compared to the cells from mutated non-progressing CLL patients (p < 0.05) (**Figure 1B**). In addition, using InfinicytTM software, we merged files of all the patients into one file and then analyzed immunophenotype-based automatic separation of cell clusters (automatic population separation, APS) (InfinicytTM software, Cytognos S.L., Salamanca, Spain) based on LC3BII expression (**Figure 1C**). The APS algorithm clearly discriminated two groups of data corresponding to samples from progressing and non-progressing CLL patients, respectively (**Figure 1C**). This suggests that LC3BII expression may contribute to the discrimination between progressing and non-progressing patients. In addition, similar results were obtained when LC3BII expression was studied in cells from CLL patients by flow cytometry (**Figure 1D**) and the classical Western immunoblotting detection of LC3B (**Figure 2A**). In the presence of bafilomycin, increased LC3BII levels were observed in samples from unmutated progressing CLL patients (**Figure 1D**). Minimal

LC3BII detection was observed in samples from mutated patients (CLL#3 and CLL#4) (**Figure 1D**). In additional experiments, we analyzed p62/SQSTM1 and Beclin expression in new samples from survivor (mutated IGHV) patients 3 and 4 by Western immunoblotting. We did not find significant differences in the expression levels of autophagy proteins between untreated and bafilomycin-treated samples (**Figure 2B**).

It is important to notice that increased LC3BII expression was detected by Western immunoblotting (**Figure 2A**) when the patients progressed but before Ibrutinib treatment. Similar results were obtained by flow cytometry (**Figure 1**) after Ibrutinib treatment.

These results, taken together, suggest that autophagy activation may correlate with CLL progression beyond Ibrutinib treatment.

#### DISCUSSION

Here we found that two patients with progressing CLL showed increased expression of the autophagy protein LC3B in addition to positive CD38 and ZAP70 expression and unmutated status of IGHV.

Autophagy, and autophagy-related proteins (ATG), play a central role in integrating many stress signals to determine the fate of cells (16). There are multiple reports in the literature of autophagy providing resistance to anticancer treatments in vitro (17–20), but the resistance mechanisms have yet to be completely determined (16). Autophagy is considered a fundamental survival mechanism that allows cells to adapt to a hostile microenvironment through the recycling of cytosolic molecules in double-membrane vesicles named autophagosomes (21). This mechanism can be induced by several stressors blocking both extrinsic and intrinsic apoptotic pathways (21). It has been described that neoplastic cells can exploit autophagy to survive under hypoxia and low-nutrient conditions (22, 23). Recently, it has become evident that combinatory drug therapy can benefit from the cross-sensitization induced in tumoral cells by cross-modulation of the molecular pathways targeted by each drug. For instance, we recently observed that rapamycin, a mTOR inhibitor, enhanced Fludarabine-induced cytotoxicity in CLL B cells (4). It was reported that pre-treatment of CLL cells with Bruton's tyrosine kinase inhibitor Ibrutinib, whether ex vivo or in vivo in patients, enhances mitochondrial Bcl-2 dependence, increasing the killing of CLL cells by Venetoclax (24). Similarly, we observed that cells from patients with progressing CLL treated in vivo with Ibrutinib were more sensitive to in vitro treatment with Venetoclax than cells from patients with non-progressing CLL (data not shown).

Kipps et al. highlighted that the "clinical course of newly diagnosed CLL is very variable; some patients remain free of symptoms and are fully active for decades, whereas others rapidly become symptomatic or develop high-risk disease, which requires treatment soon after diagnosis and might result in death due to therapy-related and/or disease-related complications (2). However, most patients have a clinical course that is in between these two extremes.

More robust prognostic markers are provided by newer techniques, such as flow cytometry, cytogenetics, and molecular biology" (2). Here, we applied flow cytometry technology to simultaneously detect autophagy protein LC3B together with classical phenotypical markers that identify tumoral CLL B cell clones. In addition, we exploited immunophenotype-based (including LC3B expression) automatic separation of cell clusters (APS) to discriminate different groups of data that correlated with the disease progression and IGHV mutational status of the patients. Our results suggest that activation of autophagy flux may correlate with CLL progression even before Ibrutinib treatment. Kong et al. (25) found increased levels of ATG5 and Beclin mRNA in a Chinese cohort of CLL patients compared to healthy controls. However, these results were inconsistent with the findings by Kristensen et al. (26) showing high expression of Beclin being associated with more aggressive disease. The detailed study of the dynamics of autophagosome formation and disappearance during the autophagic flux may solve discrepancies in the interpretation of the role of autophagy in pathogenesis, progression, and therapheutic outcome in CLL.

Autophagy is a very complex response that involves the expression, modification, association, and degradation of ATG proteins. Alterations in the expression of autophagy genes contribute to the tumorigenesis process in numerous types of cancer during tumor initiation, progression, and development and to the maintenance of the malignant state (27). The activation of autophagy flux plays key roles in controlling the tumor microenvironment and in the therapeutic response (28). However, in different subtypes of hematopoietic malignancies, the role of autophagy in cell transformation and the impact of autophagy on the response to different treatment strategies remains ambiguous. In this regard, it has been reported that decreased Bcl-2 levels and an increase in Beclin-1 expression correlate with a favorable clinical outcome in high-grade Bcell lymphomas (29, 30). In line with these observations, it was reported that high expression of LC3 or BECN1 is associated with a favorable outcome in multiple myeloma (31). Nevertheless, other studies have demonstrated an opposite role for the autophagy machinery in disease malignancy development, showing that autophagy activation is required to induce multiple myeloma cell survival (32, 33). Moreover, autophagy machinery activation has been shown to be involved in the progression of other hematopoietic cancers. Indeed, Giatromanolaki et al. reported that patients with follicular and diffuse large Bcell lymphomas express high LC3 levels compared to normal

#### REFERENCES


B-cells (34). Considering all of this information, the role of autophagy in lymphoid malignancies is still debated and might be subtype-specific. While several studies have addressed the impact of autophagy in treatment response in several subtypes of hematopoietic malignancies, further studies are needed to better understand the effect of changes in the autophagy flux during disease progression and therapy responses.

## DATA AVAILABILITY STATEMENT

All datasets generated for this study are included in the article/supplementary material.

#### ETHICS STATEMENT

The studies involving human participants were reviewed and approved by Hospital Nacional de Clínicas, Facultad de Ciencias Médicas, Universidad Nacional de Córdoba, Argentina. The patients/participants provided their written informed consent to participate in this study.

## AUTHOR CONTRIBUTIONS

DA and CR designed and carried out all the experiments and wrote the manuscript. CB helped with experiments and project discussion. CM-R revised the manuscript. DS helped with sample preparation. VH helped in the organization of the clinical data. CS and IS carried out and helped discuss the results of molecular biology studies. PI conceived and designed the study, supervised all of the experiments, and wrote the manuscript. All authors contributed to the article and approved the submitted version.

#### FUNDING

This work was supported in part by the Secretaría de Ciencia y Tecnología from Universidad Nacional de Córdoba (SECyT), Agencia Nacional de Promoción Científica y Tecnológica (ANPCyT), Fondo para la Investigación Científica y Tecnológica (FONCyT), Consejo Nacional de Investigaciones Científicas y Técnicas (CONICET), and Instituto Nacional del Cancer, Argentina.

## ACKNOWLEDGMENTS

The authors thank Alejandra Romero, Paula Icely for cell culture assistance, and Pilar Crespo and Paula Abadie for cell-sorting technical support.

<sup>3.</sup> Bologna C, Buonincontri R, Serra S, Vaisitti T, Audrito V, Brusa D, et al. SLAMF1 regulation of chemotaxis and autophagy determines CLL patient response. J Clin Invest. (2016) 126:181–94. doi: 10.1172/ JCI83013

<sup>4.</sup> Rodriguez CM, Bussi C, Arroyo DS, Sastre D, Heller V, Stanganelli C, et al. Effects of rapamycin in combination with fludarabine on primary chronic lymphocytic leukemia cells. Leuk Lymphoma. (2019) 60:1299– 303. doi: 10.1080/10428194.2018.1529309


**Conflict of Interest:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2020 Arroyo, Rodriguez, Bussi, Manzone-Rodriguez, Sastre, Heller, Stanganelli, Slavutsky and Iribarren. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.