Abstract
Spatial Protein Quality Control (sPQC) sequesters misfolded proteins into specific, organelle-associated inclusions within the cell to control their toxicity. To approach the role of sPQC in cellular fitness, neurodegenerative diseases and aging, we report on the construction of Hsp100-based systems in budding yeast cells, which can artificially target protein aggregates to non-canonical locations. We demonstrate that aggregates of mutant huntingtin (mHtt), the disease-causing agent of Huntington’s disease can be artificially targeted to daughter cells as well as to eisosomes and endosomes with this approach. We find that the artificial removal of mHtt inclusions from mother cells protects them from cell death suggesting that even large mHtt inclusions may be cytotoxic, a trait that has been widely debated. In contrast, removing inclusions of endogenous age-associated misfolded proteins does not significantly affect the lifespan of mother cells. We demonstrate also that this approach is able to manipulate mHtt inclusion formation in human cells and has the potential to be useful as an alternative, complementary approach to study the role of sPQC, for example in aging and neurodegenerative disease.
Introduction
Aging is a ubiquitous process in life affecting most organisms, in particular animals but also many unicellular organisms. It is widely assumed that aging is caused by an accumulation of cellular damage over time1,2,3. In the budding yeast Saccharomyces cerevisiae, the cellular damages considered to contribute to aging include protein aggregates, oxidatively damaged proteins, defective mitochondria and vacuoles, and extra-chromosomal ribosomal DNA circles (ERCs)4,5. Yeast cells have developed strategies to asymmetrically segregate such damages and toxic molecules during cell division, so that daughter cells (also known as buds) produced during cytokinesis are mostly free from age-related damages. For example, asymmetric inheritance has been shown for damaged and aggregated proteins6,7, defective mitochondria8,9, vacuoles10, and ERCs11. While these studies demonstrate an association between the segregation of specific cellular damages and aging/rejuvenation, causal links between such damages and aging have not been firmly established. For example, while genetic manipulations causing an increased accumulation of protein aggregates shortens yeast replicative lifespan and manipulation causing a decreased accumulation of aggregates prolongs lifespan12,13,14,15,16,17, it is not clear if protein aggregates themselves are the cause of aging as such manipulations are expected to alter a multitude of processes in the cell.
Protein aggregates are formed by a variety of different routes when misfolded proteins overwhelm the protein quality control (PQC) systems of the cell. Such aggregation is harmful to the cell due to the loss of function of the proteins aggregating and also due to gain of toxic, non-canonical, functions of the aggregates formed. A failure of PQC leads for instance to neurodegenerative diseases in humans, where aggregated proteins impair cellular functions and eventually cause cell death18. In some organisms, protein aggregates can be cleared from the cell by ‘disaggregases’ of the Hsp100 family, such as Hsp104 in yeast19. If such clearance fails, aggregates are sequestered into large inclusions at certain cellular sites in the cell. It has been suggested that this sequestration of misfolded and aggregated proteins into large inclusions at specific sites might decrease the toxicity of such aggregates. This may be due to, for example, a reduction of the exposed surface area restricting interactions with other functional proteins and limiting titration of PQC components, including chaperones20,21. In addition, it has been shown that inclusions associated with the mitochondria are cleared out faster than inclusions that are not22, indicating that specific locations in the cell may be more efficient in certain PQC processes.
In yeast, misfolded and aggregated proteins, when formed, first accumulate at multiple sites, known as stress foci, CytoQs or Q-bodies throughout the cytosol and at the surface of various organelles, such as the endoplasmic reticulum (ER), vacuole and mitochondria. Upon prolonged proteostatic stress, aggregates in stress foci coalesce into larger inclusions by an energy-dependent process23. These large inclusions are categorized by their proximity to specific organelles (juxtanuclear quality control site (JUNQ) on the cytosolic side of the nuclear membrane24; intranuclear quality control site (INQ) in close proximity to the JUNQ, but within the nucleus, next to the nucleolus25; peripheral vacuole-associated insoluble protein deposit (IPOD) proximal to the vacuole23,24; and a site adjacent to mitochondria26). Different sorting mechanisms and factors appear to play a role in the sorting of misfolded proteins to each specific site25,27,28 but it appears that most misfolded proteins studied are sequestered to all these sites23,24,29. However, the vacuole-associated IPOD30 seems to be the deposition site for amyloid proteins including the protein mutant huntingtin, which is the causative agent of Huntington’s disease (HD)24. In contrast, mammalian cells sequester aggregated proteins, including mutant huntingtin, in a deposition site close to the centrosome, known as the aggresome31,32.
That spatial control of aggregates is important for cellular fitness, rejuvenation, and longevity is inferred from results using mutations causing defects in spatial PQC (sPQC)7,15,16,33. This study was aimed at generating a different approach to studying the importance of sPQC by artificially transporting protein aggregates to non-canonical, but targeted, locations in the cell by using protein fusions to the yeast aggregate-binding protein Hsp104. We demonstrate that this approach is successful in forming inclusions at non-canonical sites in both yeast and human cells and has the potential to be useful as an alternative, complementary strategy to study the role of sPQC in aging and neurodegenerative disease. Using this approach, we show that mother cells freed of large inclusions of mutant huntingtin display a reduced probability of dying suggesting that such large inclusions are not entirely inert.
Results and discussion
Generation of an artificial aggregate targeting system (ATS) to transport aggregates into daughter cells
Hsp104 has been shown to recognize stress-induced and age-related protein aggregates and to contribute to their retention in the mother cell of Saccharomyces cerevisiae7,33,34. However, the role of this mother cell retention of aggregates is not clear but has been suggested to contribute to the aging of mother cells and the rejuvenation of daughters6,7,35,36. In order to further elucidate the role of such retention of aggregates, and their sequestration to certain organelle-associated sites in the cell (i.e., IPOD, JUNQ, INQ, and mitochondria)23,24,25,26, we aimed to remove aggregates from mother cells during cytokinesis by targeting them to daughter cells. In order to do so, we first fused Hsp104-GFP to the myosin motor protein Myo2 arguing that Hsp104 fused to Myo2 could bring aggregates to daughter cells by anterograde transport along actin cables. While this system was successfully targeting inclusions to the bud, it led in many cases to a lethal two-budded phenotype (Supplementary Fig. 1a). In a second attempt, we instead fused Hsp104-GFP to Pea2, a protein component that is transported with Myo2 along actin cables to the polarisome protein complex at the tip of the bud37. The HSP104-GFP-PEA2 fusion gene was integrated into the genome at the MET15 locus, leaving the endogenous HSP104 and PEA2 genes unaltered (Fig. 1a). This approach showed successful targeting to the bud (Fig. 1b) and we refer to this Hsp104-Pea2 chimera system as ‘Aggregate Targeting System 1’ (ATS1; Fig. 1a, c). Interestingly, Hsp104-GFP-Pea2 formed a single inclusion, indicating that most Hsp104-GFP-Pea2 is sequestered into the same site, suggesting that the specific targeting of Hsp104 may cause molecular crowding/seeding that induces Hsp104 inclusion formation. This is different from the normal, mainly cytosolic, localization of Hsp104 (Fig. 1b). To test the hypothesis of a seed/crowding-triggered inclusion formation, we induced artificial crowding of Hsp104 by fusing it to the p53 tetramerization domain. In support of our hypothesis, this triggered the formation of several Hsp104 inclusions per cell in the absence of proteostatic stress (Fig. 1d). To analyze whether the ATS1 and transport of aggregates were dependent on the myosin/actin system, we first checked for co-localization of the ATS1 with myosin. As expected, we found extensive co-localization of ATS1 with the Myo2 motor protein (Fig. 1e) as well as with actin filaments and patches (Supplementary Fig. 1b). Further, ATS1 inclusion formation and transport into the bud failed if the mutant Myo2R1419D was used instead of wild-type Myo2 (Fig. 1f, g): Mutant Myo2R1419D has been demonstrated to be unable to interact with Pea237. Co-localization experiments also showed that Spa2, another component of the polarisome complex, localized to the ATS1 whereas a lesser association was observed for Bud6, which is also localized to the bud neck (Supplementary Fig. 1c, d). To investigate if ATS1 can recognize and transport protein aggregates into the bud, we expressed the misfolding proteins huntingtin 103Q and 103QP (mHtt103Q and mHtt103QP, respectively). mHtt103QP is the Huntington’s disease associated N-terminal fragment of human huntingtin and exhibits an expansion from a 25 to a 97 poly-glutamine stretch leading to misfolding and aggregation38. mHtt103Q lacks the adjacent proline-rich region38. We found that ATS1 could recognize mHtt103QP and mHtt103Q aggregates, both when the gene was under control of an inducible GAL1-promoter (Fig. 1h) or a constitutive promoter (Supplementary Fig. 1e), with up to 71% of daughters containing mHtt inclusions and an up to 65% reduction of mother cells containing mHtt inclusions (Fig. 1i). A time-lapse example of ATS1-dependent binding and transport of mHtt103QP-GFP aggregates into the bud is shown in Fig. 1j and Supplementary Movie 1. We term this general principle of targeting proteins to an artificial inclusion via chimeras ‘INTACs’ (inclusion-targeting chimeras), in analogy to proteolysis-targeting chimeras (PROTACs)39,40 or lysosome-targeting chimeras (LYTACs)41. In addition to mHtt, we investigated if ATS1 can bind and transport other misfolding model proteins using the temperature-sensitive alleles guk1-7, pro3-1, and gus1-3 (ref. 42). All these misfolding model proteins showed inclusions co-localizing with ATS1 and were transported into the bud, indicating a general ability of ATS1 to bind misfolded proteins forming aggregates (Supplementary Fig. 1f–h). To check if ATS1 caused cellular defects, we measured growth using a drop test and the activity of a reporter of the Hsf1-dependent stress response43. We could not detect any growth defects for the ATS1 strain (Supplementary Fig. 1i) and only a slight increase in cellular Hsf1 activity in comparison to controls (Supplementary Fig. 1j). To investigate if Hsp104 in the artificial ATS1 inclusions is functional, we tested whether aggregates in the inclusions were cleared with similar dynamics as aggregates in wild-type cells using the misfolding protein pro3-1. We found that pro3-1 aggregates were cleared with a similar rate as aggregates in cells without ATS1, demonstrating successful disaggregation as well as functionality of Hsp104 in the ATS1 inclusions (Supplementary Fig. 1k, l).
