Diverse mechanisms of autophagy dysregulation and their therapeutic implications: does the shoe fit?

ABSTRACT

In its third edition, the Vancouver Autophagy Symposium presented a platform for vibrant discussion on the differential roles of macroautophagy/autophagy in disease. This one-day symposium was held at the BC Cancer Research Centre in Vancouver, BC, bringing together experts in cell biology, protein biochemistry and medicinal chemistry across several different disease models and model organisms. The Vancouver Autophagy Symposium featured 2 keynote speakers that are well known for their seminal contributions to autophagy research, Dr. David Rubinsztein (Cambridge Institute for Medical Research) and Dr. Kay F. Macleod (University of Chicago). Key discussions included the context-dependent roles and mechanisms of dysregulation of autophagy in diseases and the corresponding need to consider context-dependent autophagy modulation strategies. Additional highlights included the differential roles of bulk autophagy versus selective autophagy, novel autophagy regulators, and emerging chemical tools to study autophagy inhibition. Interdisciplinary discussions focused on addressing questions such as which stage of disease to target, which type of autophagy to target and which component to target for autophagy modulation.

Abbreviations: AD: Alzheimer disease; AMFR/Gp78: autocrine motility factor receptor; CCCP: carbonyl cyanide m-chlorophenylhydrazone; CML: chronic myeloid leukemia; CVB3: coxsackievirus B3; DRPLA: dentatorubral-pallidoluysian atrophy; ER: endoplasmic reticulum; ERAD: ER-associated degradation; FA: focal adhesion; HCQ: hydroxychloroquine; HD: Huntingtin disease; HIF1A/Hif1α: hypoxia inducible factor 1 subunit alpha; HTT: huntingtin; IM: imatinib mesylate; MAP1LC3B: microtubule associated protein 1 light chain 3 beta; NBR1: neighbour of BRCA1; OGA: O-GlcNAcase; PDAC: pancreatic ductal adenocarcinoma; PLEKHM1: pleckstrin homology and RUN domain containing M1; polyQ: poly-glutamine; ROS: reactive oxygen species; RP: retinitis pigmentosa; SNAP29: synaptosome associated protein 29; SPCA3: spinocerebellar ataxia type 3; TNBC: triple-negative breast cancer.

KEYWORDS:

• Autophagy
• beclin1
• BNIP3
• cancer
• hydroxychloroquine
• LV320
• mitophagy
• neurodegeneration
• polyQ
• proteinopathies

Autophagy eats toxic proteins

Aggregating proteinaceous inclusion bodies are hallmarks of neuropathies such as Alzheimer, Parkinson and Huntington diseases (HD) [1]. The notorious trinucleotide repeats, CAG, that translate into polyglutamine (polyQ) tracts in disease-associated proteins, such as mutant HTT (huntingtin) in HD and ATXN3 (ataxin 3) in spinocerebellar ataxia type 3 (SPCA3), promote the assembly of these proteins into disease-causing oligomers and aggregates [2]. Dr. David Rubinsztein discussed the normal function of polyQ and the pathogenic consequences of expanded polyQ tracts beyond aggregation; specifically, he described the wild-type and expanded function of polyQ in the deubiquitinase enzyme ATXN3. His group had recently discovered the interplay between ATXN3 and BECN1 a regulator of autophagy induction [3]. ATXN3 deubiquitinates BECN1, which prevents it from being degraded. The interaction between ATXN3 and BECN1 is mediated by the polyQ stretch in wild-type ATXN3. Dr. Rubinsztein showed that disease-causing expansions of polyQ tracts increase the binding of ATXN3 to BECN1. However, expanded polyQ tracts reduce the deubiquitinase activity of ATXN3, thus conferring a ‘dominant negative-like effect’ over BECN1. Other proteins with expanded polyQ, such as mutant HTT and ATN1 (atrophin 1) – a causal protein for dentatorubral-pallidoluysian atrophy (DRPLA) – also had the ability to displace BECN1 from wild-type ATXN3, which impaired starvation-induced autophagy in HD and DRPLA cells. These effects occurred when the disease-causing proteins were soluble. The subtle effects of polyQ tracts on basal autophagy, which were magnified during starvation, may contribute to the late onset of polyQ diseases, at which point aggregating mutant proteins become autophagy substrates. The inability to induce autophagy leads to the toxic effects of accumulating aggregates and the severity of polyQ diseases; autophagy upregulation is hence viewed as a sensible therapeutic option for polyQ disease. As we continue to search for drugs to enhance autophagy, Dr. Rubinsztein emphasized the importance of understanding how autophagy is impaired in any given disease, as a drug that induces autophagy can bear a different consequence when autophagy induction is impaired, as opposed to when autophagosome clearance is compromised.

To kill or not to kill? – Autophagy versus mitophagy

Dr. Kay Macleod explores autophagy in tumor survival and progression, and in her keynote presentation, she highlighted the context-dependent differential roles of general autophagy and selective autophagy in the metastatic cascade during progression of breast carcinoma. Her group discovered that autophagy acts through targeted degradation of PXN (paxillin) to foster the disassembly of focal adhesions (FAs) to promote cell motility and invasion of tumor cells [4,5]. Autophagy inhibition, achieved through stable knockdown of either ATG5 or ATG7, or with the autophagy inhibitory drug, chloroquine, resulted in reduced lung metastasis but no changes in primary tumor burden in the mouse 4T1 orthotopic mammary tumor model. Conversely, loss of the mitophagy protein BNIP3 resulted in rapid progression to malignant disease and lung metastasis in the mouse MMTV-PyMT mammary carcinoma model[6]. BNIP3 is a hypoxia-inducible mitochondrial protein that directly interacts with the ubiquitin-like modifier MAP1LC3B/LC3B (microtubule associated protein 1 light chain 3 beta), through a conserved LC3-interacting region motif, to promote mitophagy. Dr. Macleod’s group discovered that BNIP3 is robustly expressed in adenomas and early carcinomas but its expression is gradually lost upon progression to metastatic disease. They showed that accumulation of dysregulated mitochondria – through the inhibition of BNIP3-mediated mitophagy – induces redox stress with excessive reactive oxygen species (ROS) accumulating, and is associated with increased glycolysis and reduced oxidative metabolism. This results in the upregulation of HIF1A (hypoxia inducible factor 1 subunit alpha) and its target genes, including key glycolytic enzymes, promoting a reliance on glycolysis in BNIP3-null tumors. Quenching ROS in BNIP3-null tumors attenuated HIF1A induction and significantly reduced tumor growth and progression to metastasis. BNIP3-null tumors were more sensitive to acute autophagy inhibition compared to wild-type cells in a glycolysis-independent manner, suggesting that loss of BNIP3 creates a dependency on autophagy for cell survival and a dependency on glycolysis for cell growth. Additionally, BNIP3 deletion in human triple-negative breast cancer (TNBC) was associated with increased proliferative index, lymph node metastases and poor prognosis; hence BNIP3 is currently being investigated as a potential biomarker for progression in TNBC. Dr. Macleod’s work underlines the context-dependent roles of autophagy versus mitophagy in primary and metastatic breast cancer: autophagy sustains survival of the primary tumor and promotes metastasis, whereas functional BNIP3-mediated mitophagy delays the progression to metastatic disease. These discoveries emphasize the need to understand the roles of different forms of autophagy to identify effective therapies.

To target or not to target?

Defective or downregulated autophagy is linked to proteinopathies

Dr. Yanping Zhu (Dr. David Vocadlo’s lab) discovered that autophagy is upregulated in response to the pharmacological inhibition of the deglycosylation enzyme OGA (O-GlcNAcase), in an MTOR-independent manner in Alzheimer disease (AD) mouse brain. OGA inhibition also led to significant reduction in the levels of pathological MAPT/tau species in 2 AD mouse models, which was suggested to be a result of enhanced autophagy. Dr. Fanny Lemarie (Dr. Michael Hayden’s lab) reported a loss-of-function mutation in HTT protein that could explain dysfunctional autophagy in Huntington disease (HD) [7]. Wild-type HTT protein was discovered to possess a previously unknown autophagy-inducing domain (HTT[553–585]), which is released through caspase cleavage at Asp552 and Asp586, and myristoylated at Gly553. The myristoylated HTT fragment localizes to the endoplasmic reticulum (ER) and promotes ER-mediated autophagosome formation. The HD-linked mutation in the HTT protein diminishes myristoylation, leading to impaired autophagy flux. Restoring myristoylation at Gly553 – by blocking caspase cleavage at D586 – completely ameliorates the disease phenotype in HD mice.

Autophagy supports cancer survival in a tissue-dependent manner

Drs. Katharina Rothe (Dr. Xiaoyan Jiang’s lab) and Svetlana Bortnik (Dr. Sharon Gorski’s lab) investigated ATG4B-mediated autophagy in chronic myeloid leukemia (CML) and ERBB2/HER2⁺ breast cancer, respectively. ATG4B regulates the initiation and sustenance of autophagy flux by proteolytically activating LC3B. Activated LC3B is required for the dynamic expansion and curvature of phagophores. Recycling of LC3B by ATG4B from mature autophagosomes ensures the availability of cleaved LC3B for continuous autophagy flux to occur. Dr. Rothe observed highly dysregulated autophagy and ATG4B expression in CML stem/progenitor cells from the ABL1 tyrosine kinase inhibitor (TKI) IM (imatinib mesylate) non-responders [8]. Genetic inhibition of ATG4B or its pharmacological targeting by a small molecule compound called LV320 resulted in autophagy inhibition as well as sensitization of drug-resistant stem/progenitor cells to TKIs. LV320 worked synergistically with TKIs, particularly under serum deprivation, suggesting that ATG4B-mediated inhibition of autophagy in combination with TKIs may be able to circumvent resistance in CML. Dr. Svetlana Bortnik reported that ERBB2/HER2⁺ breast cancer cells are dependent on ATG4B-mediated autophagy during stress, such as starvation [9]. Genetic targeting of ATG4B in combination with trastuzumab (ERBB2/HER2-targeting antibody) significantly reduced viability compared to trastuzumab treatment alone in both treatment-sensitive and resistant ERBB2/HER2⁺ breast cancer cell lines under fed conditions. A study performed in pancreatic ductal adenocarcinoma (PDAC) cell lines by Paalini Sathiyaseelan (Dr. Sharon Gorski’s lab) showed that ATG4B knockout was sufficient to inhibit LC3B-mediated autophagy but it was not sufficient to alter cell viability under fed or starved conditions. Together, these studies emphasize the context and tissue-dependent responses to ATG4B inhibition in cancer.

Host autophagy is required for pathogen survival

Yasir Mohamud (Dr. Honglin Luo’s lab) explained that enteroviruses such as the coxsackievirus B3 (CVB3) can hijack autophagic flux for viral growth in cells, resulting in the inflammation of the heart or myocarditis [10]. CVB3-encoded proteinases cleaved selective autophagy receptors such as SQSTM1/p62 and NBR1 (NBR1, autophagy cargo receptor) into fragments, resulting in reduced autophagic degradation of aggregates, or aggrephagy. Fragmented autophagy receptors displayed a dominant-negative effect, which further crippled the clearance of ubiquitinated protein aggregates. Two vital proteins for autophagosome-lysosome fusion, the SNARE protein SNAP29 (synaptosome associated protein 29) and tethering protein PLEKHM1 (pleckstrin homology and RUN domain containing M1) were also cleaved by CVB3 proteinases. The resultant inhibition of autophagy flux supported increased viral growth, probably due to the accumulation of autophagosomes providing a site for viral RNA replication and assembly.

What to target? Autophagy interactors, regulators and tools

Runxia Wen (Dr. Orson Moritz’s lab) described a relationship between autophagy and the inherited degenerative disease, retinitis pigmentosa (RP), which is often caused by misfolding of the photosensitive protein RHO (rhodopsin). Accumulation of autophagic structures was observed in degenerating rod photoreceptors in a Xenopus laevis model of RP. Runxia Wen reported that while upregulation of autophagy is associated with misfolding of RHO, autophagy induction was dependent on phototransduction itself – determined using the genetic deletion of a gene encoding a chromophore-synthesizing enzyme, essential for phototransduction, in both dark-reared and light-exposed rods. This mechanism of light-mediated autophagy is under investigation.

Yahya Mohammadzadeh (Dr. Ivan R. Nabi’s lab) presented a novel function for the Cue domain in AMFR/Gp78 (autocrine motility factor receptor), an E3 ubiquitin ligase involved in the ER-associated degradation (ERAD) pathway that is essential in differentiating healthy versus damaged mitochondria. AMFR was previously shown to mediate PRKN/parkin-independent mitophagy. Point mutations within the Cue domain resulted in mitophagy induction even in the absence of the mitochondria depolarizing agent, carbonyl cyanide m-chlorophenylhydrazone (CCCP), which suggests a regulatory function of the Cue domain in AMFR.

Tamiza Nanji (Dr. Calvin K. Yip’s lab) explained that the key difference between the Atg1 complex in budding and fission yeast is that the budding yeast Atg1 complex consists of the subunits Atg29 and Atg31 in place of Atg101, which is found in the ULK complex (homolog of the Atg1 complex) in mammalian cells [11]. Although both the budding yeast and fission yeast Atg1 complexes share a common scaffolding subunit, Atg17, fission yeast Atg17 adopts a rod-shaped overall structure distinct from the inherently curved-shaped budding yeast Atg17. Furthermore, while fission yeast Atg17 can interact with budding yeast Atg29-Atg31, this protein cannot complement the functional loss of Atg17 in budding yeast. Finally, fission yeast Atg101 does not bind fission yeast Atg17, but instead stabilizes Atg13 through interaction via its HORMA (Hop1, Rev1, and Mad2) domain, showcasing the unique biochemical features of an Atg101-containing Atg1/ULK complex.

Dcp-1, an effector caspase that was previously shown to positively regulate starvation-induced autophagy in Drosophila [12], was discovered by Dr. Courtney Choutka (Dr. Sharon Gorski’s lab) to activate compensatory autophagy when proteasomal activity is impaired [13]. The effector caspase itself is affected by proteasomal activity that is mediated by a heat shock protein called Hsp83, the functional homolog of human HSP90. Loss of Hsp83 resulted in elevated Dcp-1 and pro-Dcp-1 that was not due to transcriptional upregulation but rather a reduction in proteasomal-mediated turnover. Loss-of-function Hsp83 mutants activated autophagy and cell death during Drosophila mid-oogenesis as well as displayed decreased levels of proteasomal activity. Flies with a combination of Dcp-1 and Hsp83 mutations no longer showed an autophagic response but interestingly still experienced cell death, effectively separating these 2 disparate pathways. RNAi experiments that reduced Dcp-1 in combination with Hsp83 or proteasomal subunit Rpn11 failed to induce autophagy – validating the Dcp-1-compensatory autophagy relationship in Drosophila. Evidence to support an evolutionarily conserved role for autophagy regulation by a human caspase was also presented.

Wieslawa Dragowska (Dr. Marcel Bally’s lab) and Dr. Peter Clark (Dr. Robert Young’s lab) introduced chemical tools that can be used as a proof of concept to test therapeutic benefits of autophagy inhibition. Wieslawa Dragowska described a liposomal copper sulfate-loaded hydroxychloroquine (HCQ) formulation that increased the total drug exposure over time ^~ 850-fold compared to free HCQ, as shown by pharmacokinetics analysis measuring the area under the curve from 0 to infinity AUC_0-∞ in plasma samples derived from CD1 mice [14]. Liposomal HCQ in combination with rapamycin or gefitinib resulted in a greater formation of lipidated LC3B (LC3B-II) in various normal mouse tissues and in JIMT-1 tumors (gefitinib-resistant breast cancer model), indicative of repressed autophagy, compared to free HCQ combinations. Although treatment (4 weeks) of immunocompromised 129S6/SvEvTac-Rag2^tm1Fwa (Rag2M) mice bearing JIMT-1 tumor xenografts with the liposomal HCQ and gefitinib combination engendered a decent 41% reduction in tumor volume, no therapeutic gain was observed over the free HCQ and gefitinib combination. Dr. Peter Clark described a new ATG4B targeting chemical probe called DB2-082 that results in autophagy suppression in cellular assays and in vivo. Yet, cellular cytoxocity limits its use as a robust chemical tool to study ATG4B-mediated autophagy inhibition in vitro and in vivo, and so further structural optimization is in progress.

Conclusion

The ability to modulate autophagy effectively in clinical settings will depend on identifying the specific disease contexts where autophagy induction or inhibition will be beneficial. Importantly, there is also a need to understand the molecular mechanisms of autophagy dysregulation in these contexts. For example, in the context of neurodegenerative disease, autophagy induction strategies are not a ‘one size fits all’ if a component of the core machinery is nonfunctional. Similarly, autophagy inhibition strategies in cancer need to consider the context-dependent potential for compensatory mechanisms and/or autophagy-independent effects. Further, an improved understanding of regulators of bulk autophagy, mitophagy and other forms of selective autophagy in normal physiology and disease may offer additional new avenues and targets to implement more refined or even personalized autophagy modulation strategies.

Acknowledgments

We thank Dr. Stephanie McInnis for organizing many aspects of the symposium. We would also like to express our appreciation for all the speakers, poster presenters and other attendees who participated in this exciting symposium.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

We gratefully acknowledge our funders and sponsors: BC Cancer Agency, BC Cancer Foundation, Simon Fraser University through support from the Vice-President Academic Conference Fund, Canada’s Michael Smith Genome Sciences Centre, Canadian Institutes of Health Research, The Centre for Drug Research and Development (CDRD), New England BioLabs, and ThermoFisher Scientific.