Differentiation of cellular lineages is facilitated by asymmetric segregation of fate determinants between dividing cells. In budding yeast, various aging factors segregate to the aging (mother)-lineage, with poorly understood consequences. In this study, we show that yeast mother cells form a protein aggregate during early replicative aging that is maintained as a single, asymmetrically inherited deposit over the remaining lifespan. Surprisingly, deposit formation was not associated with stress or general decline in proteostasis. Rather, the deposit-containing cells displayed enhanced degradation of cytosolic proteasome substrates and unimpaired clearance of stress-induced protein aggregates. Deposit formation was dependent on Hsp42, which collected non-random client proteins of the Hsp104/Hsp70-refolding machinery, including the prion Sup35. Importantly, loss of Hsp42 resulted in symmetric inheritance of its constituents and prolonged the lifespan of the mother cell. Together, these data suggest that protein aggregation is an early aging-associated differentiation event in yeast, having a two-faceted role in organismal fitness.https://doi.org/10.7554/eLife.06197.001
Aging is a complex process. Studies involving a single-celled organism called budding yeast are commonly used to investigate the factors that contribute to aging. When these yeast cells divide, a small daughter cell buds out from a large mother cell. A mother cell has a limited lifespan and produces a finite number of daughter cells and then dies (a phenomenon referred to ‘replicative aging’). However, when a daughter cell forms, the daughter's age is reset to zero, giving it the full potential to produce new offspring.
Previous research on budding yeast has shown that damaged or aggregated proteins accumulate in the mother cells but not the daughter cells, and that this accumulation of proteins contributes to shortening the lifespan of the mother cell. Furthermore, protein aggregation has also been associated with a number of age-related diseases in humans, including neurodegenerative disorders such as Alzheimer's and Parkinson's disease. However, it remains unclear how cells respond to protein aggregation that occurs during aging.
Many studies that have previously investigated this question have relied on exposing cells to stressful conditions, such as high temperatures, in order to trigger proteins to aggregate. But now, Saarikangas and Barral have studied how proteins aggregate under normal, unstressed conditions in budding yeast as they age. The experiments revealed that most unstressed yeast cells develop a single deposit of aggregated proteins already during early aging. This age-associated structure proved to be a different type of response than the protein aggregation that occurs during stress.
Furthermore, the deposit did not form as a consequence of the cell having obvious problems with folding its proteins, nor did the deposit hinder cells from coping with stressful conditions that trigger protein misfolding. Rather, this deposit supported the ability of the cell to degrade defective proteins. This suggests that the deposit represents an early adaptive response to aging, which might consequently provide aged cells some advantage over their younger counterparts.
Saarikangas and Barral also found that this protein deposit was always retained in the mother cell and not passed onto the daughters at cell division. Further experiments showed that an enzyme called heat shock protein 42 was responsible for collecting target proteins and bring them together to form the single deposit. Reducing the levels of this enzyme prevented the deposit from forming and extended the lifespan of the mother cells. Thus, these findings suggest that mother cells collect harmful protein aggregates into a single deposit and prevent them from entering the daughter cells. Further work is needed to understand how the deposit is preferentially retained within the mother cell.https://doi.org/10.7554/eLife.06197.002
Aging results in an increasing decline of the organism's fitness over time (Lopez-Otin et al., 2013). Remarkably, this process segregates asymmetrically during budding yeast division: the mother cell forms an aging lineage, whereas the daughters generated by these mothers rejuvenate to form eternal lineages, similar to the segregation of soma and germ lineages in metazoans. Such lineage separation requires that the inheritance of factors that promote aging, such as defective/deleterious organelles, proteins, and DNA, is asymmetric during cell division (Sinclair and Guarente, 1997; Aguilaniu et al., 2003; Erjavec et al., 2007; Henderson and Gottschling, 2008; Shcheprova et al., 2008; Liu et al., 2010; Zhou et al., 2011; Hughes and Gottschling, 2012; Higuchi et al., 2013; Clay et al., 2014; Denoth Lippuner et al., 2014; Henderson et al., 2014; Higuchi-Sanabria et al., 2014; Thayer et al., 2014; Katajisto et al., 2015). Therefore, how cells are able to recognize, sort, and coordinate the asymmetric segregation of aging factors and other fate determinants is an outstanding question in biology (Neumuller and Knoblich, 2009).
Protein aggregates and/or damaged proteins are a hallmark in the etiology of many human disorders associated with aging (Hartl et al., 2011; Wolff et al., 2014), and their presence correlates with aging of mitotically active yeast and drosophila stem cells (Aguilaniu et al., 2003; Erjavec et al., 2007; Bufalino et al., 2013; Coelho et al., 2013). Studies on budding yeast have shown a correlation between the accumulation of protein aggregates and replicative aging by demonstrating that Hsp104-mediated protein disaggregation is required for full replicative life span (Erjavec et al., 2007), and that over-expression of Mca1, which counteracts the formation of stress- and age-associated protein aggregates (Lee et al., 2010; Hill et al., 2014), extends the life span of yeast mother cells (Hill et al., 2014).
How cells respond to protein aggregation that occurs specifically during aging has remained elusive since most studies investigating the cellular responses to protein aggregation have relied on over-expression of non-native, aggregation prone proteins, proteostasis inhibitors, or other stressors, such as heat (Kaganovich et al., 2008; Liu et al., 2010; Specht et al., 2011; Zhou et al., 2011; Malinovska et al., 2012; Spokoini et al., 2012; Winkler et al., 2012; Escusa-Toret et al., 2013; Zhou et al., 2014). These studies have uncovered specific modes of cytosolic compartmentalization that take place when cells encounter proteotoxic stress. For example, cells stressed with heat respond by forming multiple protein aggregates (referred to as peripheral aggregates, stress foci, Q-bodies, or CytoQ) at the surface of the ER (Specht et al., 2011; Spokoini et al., 2012; Escusa-Toret et al., 2013; Miller et al., 2015; Zhou et al., 2014; Wallace et al., 2015). These structures, hereafter referred as Q-bodies, contain acutely misfolded proteins that are sorted between the nuclear and cytoplasmic degradation/deposit sites by the Hook family proteins Btn2 and Cur1 (Malinovska et al., 2012), and coalesce together by the aid of small heat shock proteins; Hsp42 in budding yeast (Specht et al., 2011; Escusa-Toret et al., 2013), and Hsp16 in fission yeast (Coelho et al., 2014). Simultaneously, Q-bodies are being rapidly resolved by the protein disaggregase Hsp104 (Parsell et al., 1994; Specht et al., 2011; Zhou et al., 2011; Spokoini et al., 2012; Escusa-Toret et al., 2013), together with other heat shock responsive chaperones such as Hsp70 and Hsp82 (Escusa-Toret et al., 2013). The formation of Q-bodies seems to aid stress survival, as the deletion of HSP42 resulted in defective tolerance of prolonged heat stress (Escusa-Toret et al., 2013). The asymmetric inheritance of Q-bodies by the mother cells is promoted by the geometry of the bud neck (Zhou et al., 2011), tethering to mitochondria (Zhou et al., 2014), and by actin cable-mediated retrograde transport, which is dependent of Hsp104 and Sir2 (Liu et al., 2010; Song et al., 2014). Notably, Sir2 is also a key player in processes that underlie the asymmetric segregation of damaged mitochondria (Higuchi et al., 2013) and the accumulation of extrachromosomal DNA circles (Sinclair and Guarente, 1997; Kaeberlein et al., 1999) to the aging mother cell.
Prolonged Q-body-inducing stress (heat or over-expression of thermolabile proteins) combined with proteasome inhibition can lead to the formation of a dynamically exchanging deposit of ubiquitylated proteins named the juxtanuclear quality compartment, JUNQ (Kaganovich et al., 2008; Escusa-Toret et al., 2013). This structure is regulated by the Upb3 deubiqutinase (Oling et al., 2014), by proteosomal activity (Andersson et al., 2013) and by lipid droplets (Moldavski et al., 2015), and it was also shown to appear during replicative aging (Oling et al., 2014). The faithful inheritance of this structure by the mother cell is dependent on its association with the nucleus (Spokoini et al., 2012). More recently, it was shown that the ‘JUNQ’ might actually reside inside the nucleus, and it was thus renamed as intranuclear quality control compartment, INQ (Miller et al., 2015). The JUNQ/INQ assembly is dependent on Btn2-aggregase (Miller et al., 2015), a protein also found to be involved in prion curing (Kryndushkin et al., 2008, 2012). Apart from the JUNQ/INQ structure, terminally aggregating proteins, such as the amyloidogenic prions Rnq1 and Ure2, were shown to partition to an non-dynamic, vacuole-associated deposit called the insoluble protein deposit IPOD (Kaganovich et al., 2008; Tyedmers et al., 2010b), which has remained less well characterized.
Despite this wealth of data, it remains unclear how these exogenous/stress-induced aggregation models relate to protein aggregation that takes place during physiological ‘healthy’ aging. Particularly, it is unclear why/how protein aggregates arise during aging, how are they segregated during cell division and, importantly, what is their consequence to the protein quality control of the aging cell, as well as to the aging process itself. To illuminate these aspects, we probed the role of protein aggregation during unperturbed replicative aging. Our findings indicate that protein aggregation is a prevalent and highly coordinated event of early aging and is not solely associated with proteostasis deterioration. Instead, we provide evidence that age-associated protein aggregation may initially benefit the cytosolic protein quality control, but eventually becomes involved with age-associated loss of fitness.
To address the role of protein aggregation in unperturbed, physiological aging, we analyzed microscopically the replicative age-associated protein aggregation landscape in budding yeast by visualizing different chaperone proteins that mark aberrantly folded and aggregated proteins. By employing the Mother Enrichment Program (MEP) (Lindstrom and Gottschling, 2009) (Figure 1—figure supplement 1A), we harvested cells of different age and first analyzed the localization of endogenous GFP-tagged protein-disaggregase Hsp104 (Parsell et al., 1994; Glover and Lindquist, 1998), a broad sensor for protein aggregates (Figure 1A, Haslberger et al., 2010). Interestingly, we found many cells displaying an aggregate (typically a single bright Hsp104-labeled focus) and this portion increased in a progressive, age-dependent manner such that >80% of cells that had undergone more than 6 divisions displayed such a structure (Figure 1A,B), as previously reported (Aguilaniu et al., 2003; Erjavec et al., 2007). Co-localization analysis with Hsp104 demonstrated that the Hsp70 proteins Ssa1 and Ssa2, the small heat shock protein Hsp42, and the Hsp40 protein Ydj1 readily localized to these aggregates, the Hsp26 was found to be enriched in only 15% of Hsp104-labeled foci, while no accumulation of Hsp40 protein Sis1 or the Hsp90 protein Hsp82 was detected (Figure 1C, Figure 1—figure supplement 2A–C). Importantly, age-dependent appearance of these structures was also detected in diploid cells, in other strain backgrounds (W303), independently of the MEP procedure, and when different fluorophores where used for tagging Hsp104 (Figure 1—figure supplement 2D–H), indicating that their formation represents a general, age-dependent phenomenon of budding yeast cells.
To characterize the nature of these protein deposits further, we performed live-cell imaging by acquiring images every 15 min over several hours and covering the entire depth of the cell. This showed that most mother cells started to form this aggregate already after budding three to four times (Figure 1D, 390 min). The fastest growth phase of the aggregate took place during the first half an hour following its detection, after which the Hsp104 signal intensity increased only marginally (Figure 1D,E). While the aggregates initially underwent occasional dissolution, they became stable within a few hours from nucleation (Figure 1E). Pedigree analysis of aggregate inheritance in cells of different age showed that they faithfully segregated to the aging mother cell (98% of divisions), irrespective of its age (Figure 1D,F, Video 1). To further analyze the persistence and behavior of the deposits over the entire replicative life span of cells, we used a microscope-coupled microfluidic dissection platform (Lee et al., 2012) (Figure 1—figure supplement 1B). This showed that the aggregate was efficiently maintained as one compartment, and whenever new foci emerged, they typically merged soon after with the pre-existing deposit. We quantified the number of Hsp104 foci (1 or >1) and its post-nucleation stability in cells that were tracked for at least 10 divisions after the aggregate had appeared. These cells preferentially (>80% of time) displayed only a single Hsp104 focus (Figure 1G–H), which was very stable, typically persisting until the last divisions of the cell (Figure 1G,I). Moreover, this structure was largely non-dynamic, displaying limited exchange of Hsp104 with the cytoplasm, as determined by fluorescent recovery after photobleaching (FRAP) analysis in diploid cells that expressed one GFP- and one mCherry-tagged copy of HSP104 (Figure 1J). We bleached and measured the recovery of the mCherry signal (Figure 1J) and by fitting nine recovery curves, found that on average a large fraction (59%) of Hsp104 was immobile, while the mobile fraction displayed a half-time recovery rate of 8.9 s (Figure 1K). Together, these data show that a large majority of unstressed yeast cells develop a protein deposit early during replicative aging. This deposit displays limited exchange with the cytoplasm (assessed by Hsp104), is efficiently maintained as a single compartment, and is faithfully inherited by the aging mother cell.
To better understand how the age-associated protein deposit forms and what are its physiological constituents, we selected two proteins containing glutamine- and/or asparagine-rich domains, Sup35 and Dcp2, which are known to undergo conformational switches and to aggregate, and we monitored their localization in respect to the age-associated protein deposit. Sup35 (eRF3) is a translation termination factor that can undergo stable amyloid-like prion conversion from non-prion [psi–] to prion state [PSI+] (Chernoff et al., 1993; Ter-Avanesyan et al., 1994; Wickner, 1994; Patino et al., 1996; Paushkin et al., 1996). Thus, we tested whether Sup35-GFP is targeted to the age-dependent aggregate and if the prion status plays a role in the targeting to and/or in the formation of the age-associated protein deposit. Importantly, Sup35-GFP clearly accumulated into Hsp104-mCherry-labeled age-associated protein deposit in 33% of [PSI+] cells that contained such a deposit, whereas no accumulation was detected in the non-prion [psi−] cells (Figure 2A–C). As expected, the formation of this bright Sup35 focus was age-dependent (Figure 2D), in line with earlier observations (Derdowski et al., 2010), but interestingly the [PSI+] state did not have an overall effect on Hsp104-labeled deposit formation (Figure 2E). Similarly, the elimination of the [RNQ+] prion by deleting RNQ1 did not negatively influence the formation of the age-associated deposit (data not shown). Notably, time-lapse imaging showed that Sup35 was recruited into the pre-existing age-associated protein deposit (marked by Hsp104-mCherry). These Sup35-enriched age-associated protein deposits segregated asymmetrically towards the aging mother cell in 98% of mitoses (Figure 2F,G). These data provide evidence that the Hsp104-labeled focus is a bona fide deposit site for aggregating proteins, and that prion conversion of Sup35 promotes its gradual storage into the age-associated protein deposit, but does not potentiate its formation per se.
In contrast, Dcp2, a Q/N-rich component of the reversible P-body mRNP aggregates (Reijns et al., 2008), did not accumulate into the age-associated deposit (Figure 2H,I), even after induction of P-body formation by reducing the glucose level in the medium to 0.1% (Decker et al., 2007) (0/57 Hsp104 foci with Dcp2-GFP, Figure 2J). Furthermore, we never observed age-associated protein deposits and P-bodies fusing with each other (Figure 2J). This suggests that P-bodies and age-associated protein deposits have different physicochemical properties, representing two distinct modes of cytosolic sub-compartmentalization (Hyman et al., 2014; Kroschwald et al., 2015). Altogether, these data show that the substrates of the age-associated deposit are non-random and possibly amyloid-like, which might explain their irreversible nature.
During recent years, numerous distinct protein deposits, including Q-bodies, JUNQ/INQ, and the IPOD, have been discovered and thus we wanted to explore whether the age-associated protein deposit matches the identity of any of these structures. We first looked at the behavior of Q-bodies, which assemble in response to heat-stress. In stark contrast to the age-associated aggregates (Figure 1G,I), Q-bodies induced upon heat stress (42°C, 30 min) were transient and readily solubilized after the removal of the stress factor (Figure 3A; Liu et al., 2010; Escusa-Toret et al., 2013; Wallace et al., 2015). Comparative analysis of the localization of different chaperone proteins between Q-bodies and the age-associated protein deposit showed that all markers of the age-associated protein deposit were also found to accumulate in Q-bodies (Figure 3B). Conversely, we found chaperones such as Hsp82 and Btn2 localizing to Q-bodies (Figure 3B) albeit being excluded from the age-dependent protein deposit (Figure 1C). Together, the reversible nature and the difference in associated chaperones demonstrate that the Q-bodies and age-associated protein deposits can be distinguished.
We then examined whether the age-associated protein deposit would bear the characteristics of the JUNQ/INQ compartment, which is formed during prolonged stress and/or in response to proteasome inhibition. To this end, we induced the GAL-promoter-driven expression of the JUNQ/INQ marker, the unstable proteasome substrate human von Hippel-Lindau tumor suppressor protein (VHL) (Kaganovich et al., 2008; Miller et al., 2015), and monitored the accumulation of Hsp104-mCherry into the newly forming VHL-foci. VHL typically formed a single focus, which in 87% of the cases did not accumulate Hsp104-mCherry within 30 min after its appearance (Figure 3C,D, N = 29). However, the VHL expression was often associated with overall increased Hsp104-mCherry expression over time, suggesting that its expression activates a stress response. Hence, we visualized the JUNQ/INQ without over-expression of exogenous substrates. To accomplish this, we used an Hsp82-GFP, Hsp104-mCherry expressing strain deleted of the RPN4 gene, leading to increased burden of proteasomal substrates, due to decreased amount of functional proteasomes (Xie and Varshavsky, 2001). Indeed, many of rpn4Δ cells displayed numerous Hsp82-Hsp104 double-positive Q-body-like puncta (data not shown). In addition, we found cells that displayed two puncta: a Hsp104-Hsp42 double-positive JUNQ-like deposit, and a Hsp104-positive, Hsp82-negative deposit that fills the criteria of the age-associated deposit (Figure 3E). To consolidate this further, we investigated the co-localization of the endogenous JUNQ/INQ marker Btn2 (Miller et al., 2015) together with the age-associated protein deposit (Hsp104). This analysis showed that in 98.6% of the cases, Btn2 did not localize to the age-associated Hsp104-foci (Figure 1C and Figure 3F). Moreover, from all thirty identified cells displaying endogenous Btn2 foci, we found only one case in which Hsp104 was enriched at this site (Figure 3G, N = 1268 cells). Together, these data suggest that the age-associated protein deposit and the JUNQ/INQ are discrete structures that may exist in parallel.
Finally, we monitored the resemblance between the age-associated protein deposit and IPOD. To visualize cells during IPOD appearance, we imaged Hsp104-mCherry-expressing cells together with the canonical IPOD marker, the galactose-inducible Rnq1-GFP (Kaganovich et al., 2008), immediately after placing cells to galactose-containing media. This showed that the IPOD typically appeared as a single focus to which Hsp104 rapidly accumulated (Figure 3H,I) (>98% of Rnq1-foci displayed accumulation of Hsp104 within 30 min after their appearance, N = 53). However, by dissecting the Rnq1-GFP appearance dynamics in cell with a pre-existing age-associated protein deposit (marked by Hsp104-mCherry), we found that the aggregating Rnq1-GFP did not accumulate to the age-associated deposit (see white arrowhead in Figure 3H), but rather it formed a new aggregate to which Hsp104 then strongly accumulated, while its intensity at the age-associated protein deposit declined (Figure 3H). To rule out possible artifacts induced by Rnq1-GFP over-expression, we also looked at the co-localization between the age-associated protein deposit and GFP-tagged endogenous Rnq1 in its prion [RNQ+]-state. We analyzed altogether 117 Hsp104 foci-containing [PIN+, PSI+] cells and found that in 98.3% of the cases, Rnq1-GFP did not accumulate to this deposit (Figure 3J). Within this cohort, we found two cells with an Rnq1-GFP-aggregate, which was in both cases enriched with Hsp104-mCherry (Figure 3K). However, both of the Rnq1-aggregate containing cells also displayed an additional Hsp104 focus to which Rnq1 had not accumulated (Figure 3K). Together with the results from the Rnq1 over-expression, these data suggest that Rnq1 aggregate is likely to represent a structure that is different from the majority of the age-associated protein deposits.
In sum, these data point out marked differences between the age-dependent protein deposit and the previously characterized Q-bodies, IPOD, and the JUNQ/INQ that could derive from context dependent (stress vs aging) differences in cellular responses to protein aggregation.
Appearance of protein aggregates has commonly been associated with defects in protein quality control (Tyedmers et al., 2010a). Therefore, we wanted to elucidate if the formation of the age-associated protein deposit is a sign of defective protein quality control. First, we examined whether aged cells display a decline in dealing with proteotoxic stress. As shown before, young cells clear heat-induced protein aggregates (Q-bodies) rapidly after stress removal (Figure 3A). Hence, we first wanted to test if this recovery period is affected by the age of the cell (Figure 4A). We measured the clearance time of Q-bodies (defined as two or more Hsp104 foci) between young (average age 0.4 generations, between 0 and 1 generations, N = 70) and aged cells (average age 9.2 generations, between 6 and 19 generations, N = 50) following acute heat stress (Figure 4B). Surprisingly, no difference in the mean time of protein aggregate clearance was detected between these populations: young: 71 ±5 min old: 72 ±4 min, n.s.) (Figure 4C), suggesting that aged cells, despite having a protein deposit, are fit to cope with proteotoxic stress.
We then used a temperature-controlled microfluidic device that enabled us to categorize cells depending on their pre-stress age-associated protein deposit-status, and monitor them under the microscope prior to, during and after undergoing acute proteotoxic stress conditions (Figure 4D). The duration of the Q-body response (state in which cells display two or more Hsp104-foci) was plotted over time and demonstrated that cells with and without a pre-existing age-associated deposit showed a comparable Q-body response when heat was applied, while cells that were not exposed to heat did not show similar Q-body response during this time period (Figure 4E, N = 53–97 cells). Interestingly, the deposit-containing cells responded slightly faster and to a lesser extend when compared to their clonal counterparts without a pre-existing protein deposit (Figure 4E).
We then asked if exposure to proteotoxic stress (Q-body state) would promote the formation of age-associated protein deposits. To this end, we compared the aggregate status of single cells (Hsp104-mCherry: no/one/several puncta) at time points 0 and 380 min, between stress-experienced (between 50 and 80 min) and non-stressed (constantly at 30 °C) cells (Figure 4F, N = 32–77 per group). This analysis showed that the majority of non-stressed cells (62.5%) without a deposit at the beginning of the experiment displayed a single aggregate at the end of the experiment (Figure 4G), in accordance with the appearance kinetics of age-associated deposit (Figure 1B,D). Surprisingly, when the cells without a prior aggregate encountered heat stress (i.e. conditions that induce proteotoxic stress), a substantially smaller portion of them (36.4%) displayed an aggregate at the end of the experiment (Figure 4G). Similarly, there was a slight increase of cells without an aggregate in the stress-encountered cohort (17.7% vs 6.8%) among cells that displayed an age-associated deposit at the beginning of the experiment (Figure 4G). However, irrespectively of stress, cells with an age-associated deposit at the beginning of the experiment preferentially displayed a single deposit at the end of the experiment (non-stressed 82%, stress-experienced 79%) (Figure 4G). These results indicate that mild exposure to proteotoxic stress conditions counteract the formation of age-associated protein deposits and demonstrate that cells with an age-associated deposit prior to encountering the stress predominantly reverted back to this state. Altogether, these data indicate that despite forming a protein deposit, aged cells are still fit to coping with acute proteotoxic stress and suggest that the formation of the age-dependent protein deposit is not due to the overload of the general quality control machinery.
The notion that aged cells with a protein deposit handled proteotoxic stress comparably to their young counterparts prompted us to test the effect of the age-associated deposit on the function of the ubiquitin–proteasome system (UPS), which has an important role in yeast replicative aging (Kruegel et al., 2011). Pathological polyQ protein fragments have been shown to impair the UPS function both in yeast and mammalian cells (Bennett et al., 2005; Park et al., 2013), and UPS substrates have been shown to accumulate in aged cells (Andersson et al., 2013). On the other hand, stress-induced protein aggregates (Q-bodies) do not result from failed protein degradation, but may actually benefit the proteostasis of cells undergoing stress (Escusa-Toret et al., 2013). We therefore wanted to analyze the contribution of the age-associated deposit on the functionality of the UPS. In order to selectively measure the effect of the age-associated protein deposit on UPS function, we made use of the auxin-inducible degron (AID) system (Nishimura et al., 2009), which enabled us to measure the degradation rate of cytosolic (AID-GFP) and nuclear (AID-GFP-NLS) UPS substrates in vivo (Figure 5A,B). Cells co-expressing Hsp104-mCherry and AID-GFP were switched to auxin-containing media at the onset of imaging, and the GFP intensities were measured over time from neighboring cells containing or not an Hsp104-puncta. The normalized values were pooled and fitted with a non-linear one-phase decay function. Importantly, this showed that the decay rate of AID-GFP was significantly accelerated in cells with an age-associated deposit when compared to neighboring cells that did not contain an Hsp104-focus (Figure 5C,D) (rate constant (K): 0.155 ± 0.007 (with) vs 0.129 ± 0.006 (without), p < 0.01, N = 91 cells/category). However, no difference was detected in the decay rate of nuclear GFP between cells with or without an age-associated protein deposit (Figure 5E,F) (K: 0.075 ± 0.004 (with) vs 0.079 ± 0.005 (without), n.s.), demonstrating that the effect of the age-associated protein deposit on the UPS function is specific to the cytosolic compartment. Collectively, these results suggest that the age-associated deposit promotes cytosolic quality control by enabling more efficient clearance of degradation substrates.
Next, we wanted to understand what are the assembly principles of the age-associated deposit. Molecular chaperones generally recognize, refold, or sort aberrantly folded proteins, making them prime candidates to regulate the age-associated protein deposit assembly. Thus, we addressed the contribution of chaperones localizing to the age-associated protein deposit to its formation. Time-lapse microscopy showed that in 24 out of 25 cells, Hsp42 was at the age-associated protein deposit before Hsp104 and was the first protein we find to mark this structure (Figure 6A). Ssa1 and Hsp104 were recruited subsequently, appearing concurrently up to 120 min after appearance of the Hsp42-focus (Figure 6A and data not shown). To identify the role of these chaperones in the assembly of the age-associated protein deposit, we investigated the consequences of their deletion. Interestingly, loss of the Hsp70 proteins Ssa1 and Ssa2, which act synergistically with the disaggregase Hsp104 in protein refolding (Glover and Lindquist, 1998; Kim et al., 2013), resulted in rapid formation of the Hsp104 focus (Figure 6B,C) and a similar, but milder effect was detected upon deletion of HSP26. In stark contrast, deletion of HSP42 abolished age-associated protein deposit formation (Figure 6B,C), which parallels to its function in Q-body assembly (Specht et al., 2011; Escusa-Toret et al., 2013). To probe the role of Hsp104 and Hsp42 further, we analyzed the consequences of their over-expression on age-associated protein deposit formation (monitored by endogenous Hsp104-GFP). Over-expression of Hsp42 caused precocious nucleation of age-associated protein deposit and promoted its growth (Figure 6D–F). On the contrary, a significantly smaller fraction of cells over-expressing Hsp104 contained an age-associated protein deposit and while age-associated deposits still occasionally formed, they were rapidly destabilized (Figure 6D–F). Collectively, these data suggest that these chaperones functionally oppose each other (Figure 6G): the Hsp104/Hsp70 proteins prevent age-associated protein deposit formation possibly by refolding cytoplasmic substrates, whereas Hsp42 coordinates the collection of these substrates from the cytoplasm into the age-associated protein deposit (Figure 6G).
To consolidate this model, we further investigated the interplay between the refolding machinery (Hsp70/Hsp104) and the nucleation/growth promoting Hsp42. We used cells expressing Ssa1-GFP to visualize the age-associated protein deposit in the absence of Hsp104 (Figure 7A). Similarly to what we found for the deletion of SSA1 and SSA2 (Figure 6B,C), the hsp104Δ mutant cells formed the age-associated protein deposit more rapidly during aging. Again, the deletion of HSP42 had an opposite effect, hampering formation of the Ssa1-GFP focus (Figure 7A,B). Remarkably, age-associated protein deposit formation was severely impaired in the hsp104∆ hsp42∆ double-mutant cells. These cells were crowded with punctate Ssa1-GFP (speckles), which failed to coalesce into a single regular age-associated protein deposit (Figure 7A,B). The portion of cells that contained multiple Ssa1-GFP foci was significantly higher in hsp104Δ hsp42Δ double-mutant cells than in hsp104Δ single-mutant cells. Interestingly, whereas >80% of wild-type and hsp104∆ mutant cells that contained an aggregate displayed a single Ssa1-GFP focus, the majority of hsp42∆ and hsp104∆ hsp42∆ mutant cells harbored several foci (Figure 7C).
To decisively dissect the individual contributions of Hsp104 and Hsp42 on age-associated protein deposit formation, we conceived an in vivo reconstitution assay where we could rapidly re-introduce either Hsp104 or Hsp42 to the aggregate-enriched hsp104Δ hsp42Δ double-mutant cells. To accomplish this, we mated hsp104Δ hsp42Δ double-mutant cells with wild-type (re-introducing both Hsp104 and Hsp42), hsp104∆ single-mutant (to reintroduce only Hsp42), hsp42Δ single-mutant (to reintroduce only Hsp104), or with the hsp104 hsp42 double-mutant cells (reintroducing none of the chaperones) (Figure 7D). We monitored the deposit by imaging the fusing cells at 1-min intervals, using Ssa1-GFP as a reporter (Figure 7D). Importantly, simultaneous reintroduction of Hsp104 and Hsp42 was sufficient to clear the cytoplasm of Ssa1-GFP speckles already within 5 min (Figure 7E, no foci). In contrast, introducing Hsp42 alone resulted in rapid disappearance of cytoplasmic speckles with concurrent formation of a single Ssa1-GFP focus (Figure 7E, one focus). In contrary, reintroduction of only Hsp104 led to the slow and incomplete clearance of the Ssa-1 speckles, which was a reminiscent, but a milder phenotype as that observed upon conjugating the cell with hsp104Δ hsp42Δ double-mutant partner (Figure 6E, multiple foci). Collectively, this analysis supports the conclusion that Hsp42 functions to collect aggregates into the age-associated protein deposit structure, whereas Hsp104 functions to disaggregate/refold age-associated protein deposit destined substrates. In the context of the age-associated protein deposit cargo Sup35 (Figure 2), it is interesting to note that Hsp42 over-expression promotes [PSI+] curing, while its deletion results in enhanced [PSI+] induction, further supporting its role as the depositor for the age-associated protein deposit (Duennwald et al., 2012).
We then asked whether age-associated protein deposit formation ensures the asymmetric inheritance of protein aggregates by the aging lineage. Examining its mitotic segregation (monitored with Ssa1-GFP) in 96 wild-type and 186 hsp104Δ mutant cell divisions demonstrated that the age-associated protein deposit was in both cases inherited by the aging mother cell with high fidelity (in 98% of divisions; data not shown). In strong contrast, Ssa1-GFP foci were frequently inherited by the daughter cells when both HSP104 and HSP42 were deleted (Figure 8A, Video 2). By examining 61 mitotic events by time-lapse microscopy in the double-deleted cells, we observed Ssa1-GFP-foci relocating from the mother to the bud in 53 cases, but did not find any biased retrograde movement from the bud to the mother (Fig. 8A, Video 2 and data not shown). Quantification of the percentage of buds with protein deposits demonstrated that more than 65% of all hsp104Δ hsp42Δ double-mutant buds displayed at least one aggregate, compared to less than 4% in hsp104Δ and hsp42Δ single-mutant and in wild-type cells (Figure 8B). Altogether, these data provide evidence that collection of protein aggregates into a single deposit by Hsp42 promotes their asymmetric retention in the mother cell during cell division.
Finally, we tested the significance of the age-associated protein deposit pathway and asymmetric protein aggregate segregation on replicative aging (Figure 8C). In accordance to previous work (Erjavec et al., 2007), mutant hsp104∆ cells were short-lived compared to wild-type cells (21.5 vs 27 generations, p < 0.001), supporting the idea that accelerated aggregate accumulation promotes aging. Further supporting this notion, deletion of HSP42 extended the lifespan of the cells by more than 40% (39 generations, p < 0.001). The lifespan of the hsp104Δ hsp42Δ double-mutant cells was similar to that of the hsp104Δ single-mutant cells (23 vs 21.5 generations, respectively, ns.) and significantly shorter than that of the wild-type (p < 0.01) and hsp42Δ mutant cells (p < 0.001). The over-expression of HSP42 had no effect on the lifespan (data not shown), suggesting that the wild-type levels of Hsp42 are sufficient to provide its maximal effects on life span, which is consistent with the appearance of the deposit early in life span in the wild-type cells. Together, these data imply that the Hsp104/Hsp70 system, which counteracts protein aggregation, is essential for longevity, while the Hsp42-dependent circuit promotes aging in the mother lineage, presumably by building the age-associated protein deposit and thereby establishing the asymmetric inheritance of age-associated protein aggregates.
Protein aggregation has been frequently associated with aging and aging-associated diseases. Here, by using the budding yeast as a model for replicative aging, we identify an Hsp42-promoted and Hsp104/Hsp70-counteracted pathway that deposits age-associated protein aggregates and thereby ensures their biased segregation to the aging mother-lineage upon cell division (see model in Figure 7D). This might represent a common mechanism to drive asymmetric segregation of aberrant proteins in dividing cells, as stressed fission yeast cells utilize a similar Hsp16-dependent mechanism for asymmetric partitioning of misfolded proteins (Coelho et al., 2014). Intriguingly, our data suggest that depositing age-associated protein aggregates to the mother cells might be a mixed blessing. On one hand, the age-associated deposit can promote spatial protein quality control (Figures 4, 5) and promote cell diversity as an asymmetrically dividing structure that may harbor protein conformation encoded epigenetic information, such as the prion fold of Sup35 (Figure 2). At the same time, the long-term consequences of this pathway appear negative for the aging lineage (Figure 8C). Why the deposit-forming pathway ultimately becomes associated with age-dependent loss of fitness remains to be determined. In this regard, it will be of great importance to elucidate how age-associated deposit relates to other aging pathways, such as those influencing the function of mitochondria (McMurray and Gottschling, 2003; McFaline-Figueroa et al., 2011), vacuoles (Hughes and Gottschling, 2012) and nuclear pores (Shcheprova et al., 2008; Denoth-Lippuner et al., 2014; Webster et al., 2014), and how the deposit localizes relative to these organelles and their components (Chong et al., 2015).
Protein deposit-containing cells efficiently coped with acute proteotoxic stress (Figure 4) and displayed improved degradation of cytosolic UPS substrate during early- to mid-life span (Figure 5). It is important to note, however, that our assays were conducted with relatively young cells that had recently acquired the protein deposit, and hence it is possible that UPS function starts to decline later during the aging process, as shown previously (Andersson et al., 2013). Curiously, the decline in proteostasis has been associated with protein aggregation pathologies such as Alzheimer's, Huntington's, and Parkinson's disease (Vilchez et al., 2014). The decreased degradation of cytosolic UPS substrates in a polyQ disease model was found to be due to the sequestration of the Hsp40 protein Sis1 (Park et al., 2013)—a protein that was not associated with the age-associated deposit (Figure 1C). It is thus plausible that the presence of the deposit might, for example, lead to increased levels of available (substrate unbound) Sis1, thereby allowing more efficient nuclear import and degradation of cytosolic UPS substrates. Altogether, these data suggest that there is a need to better differentiate between pathophysiological protein aggregation and ‘homeostatic’ protein aggregation that takes place during physiological aging, of which the latter might initially help to coordinate chaperones and UPS factors involved in protein quality control. Related to this, the age-associated protein deposit did not clearly fill out the set criterion for any of the previously characterized cytosolic deposits, including Q-bodies, JUNQ/INQ, or IPOD, but rather seems to exist in parallel with the JUNQ and IPOD structures (Figure 1, Figure 3)—although, for example, the chaperones regulating these deposits were often shared between these structures. This favors the notion that pathological aggregation processes (mimicked e.g., by VHL and Rnq1 over-expression) can hijack endogenous proteostasis regulatory mechanisms, possibly underlying their harmful effects in cells (Park et al., 2013). Altogether, these apparent differences between the stress-induced, pathological, and aging-associated protein aggregates (Figure 2, Figure 3) demonstrates that protein aggregates are not all equal in their composition or in the way they are being recognized and dealt by the cell and reinforces the importance to discriminate between different aggregation models.
In the future, it will be important to identify the cargo deposited to the age-associated protein deposits. In this aspect, it is interesting to note that many of the long-lived asymmetrically retained proteins that accumulate to mother cells as they age (Thayer et al., 2014) were among those localizing to the age-associated protein deposit (including Hsp104, Ssa1, Ssa2, and Hsp26). One example of an age-associated deposit resident protein was the prion-like Sup35. This storage rendered Sup35 to be inherited by the mother cells (unlike a prion, analogous to a mnemon (Caudron and Barral, 2013)) (Figure 2), which is consistent with the size-dependent transmission model of Sup35-aggregates (Derdowski et al., 2010). However, aging does not enhance Sup35-dependent [PSI+] prion generation (Shewmaker and Wickner, 2006). Together, these notions suggest that Sup35 cannot escape the age-associated deposit under normal growth conditions, perhaps owing to the rigid amyloid-like packing (Saibil et al., 2012), being consistent with the durable and non-dynamic nature of these deposits. However, this notion may come with a caveat: we found that the age-associated deposit could be, at least temporarily, disassembled by the heat shock response (Figure 4). This suggests the captivating possibility that changes in the environment might trigger the spread from the age-associated deposit. For cells facing fluctuating environments in the wild, the strategy to stochastically store prion-like proteins in aged cells might prove beneficial, as it could allow stress-dependent spread of the prion conformers from a subpopulation of aged-cells to their daughters, promoting faster adaptation to the changing environment (Newby and Lindquist, 2013).
Since asymmetric inheritance of damaged proteins and protein aggregates seems to be conserved in metazoan stem cells (Rujano et al., 2006; Bufalino et al., 2013) that are also subjected to both replicative decline and segregation of lineages, we hypothesize that similar pathways that selectively segregate protein-aggregate based fate determinants asymmetrically during cell division are likely to be conserved and may contribute to cellular lineage specification across species.
Yeast strains used in this study are listed in Supplementary file 1. Strains were generated either manually (Janke et al., 2004) or derived from collections: (http://web.uni-frankfurt.de/fb15/mikro/euroscarf/col_index.html or http://clones.lifetechnologies.com/cloneinfo.php?clone=yeastgfp). Plasmids (pAG415-GDP) over-expressing Hsp104 (O-1360) and Hsp42 (O-2252), and the [PIN+][PSI+] (yJW508) and [PIN+][psi–] (yJW584) parental strains (Osherovich and Weissman, 2001) were a generous gift from Simon Alberti (Max Planck Institute of Molecular Cell Biology and Genetics Dresden, Germany). The Mother Enrichment parental strains (Lindstrom and Gottschling, 2009) were a kind gift from Dan Gottschling (Fred Hutchinson Cancer Research Center, Seattle USA) and the plasmids encoding VHL-GFP and Rnq1-GFP were provided by Judith Frydman (via Addgene). The integrative plasmids for the auxin-mediated degradation assays (pADH1-OsTIR1-9myc, pADH1-eGFP-IAA17-NLS, pADH1-eGFP-IAA17) have been described in (Nishimura et al.) and were kindly provided by Matthias Peter (ETH Zurich, Switzerland).
Cells subjected to imaging were re-inoculated from overnight cultures to O.D. 600 nm 0.05 in YPD and grown to O.D. 0.5–0.8 at 30°C, centrifuged 500 g for 5 min, resuspended in synthetic complete (SC) media (-his), and mounted between a coverslip and an agar pad (SC-his). The over-expression of Rnq1-GFP and VHL-GFP was induced by switching cells to 2% galactose at the onset of imaging. The auxin-mediated degradation assays were performed similarly as above by placing cells to 0.5 mM 3-Indoleacetic acid (Sigma–Aldrich, #12886) at the onset of imaging. The MEP was performed as described in (Lindstrom and Gottschling, 2009). Briefly, cells were grown in log phase for 15 hr before coupling the cell wall with Sulfo-NHS-LC-Biotin (Pierce). Following 2 hour recovery period, the MEP was engaged by addition of estradiol to a 1 μM final concentration. After desired incubation period, cells were coupled to uMACS Streptavidin MicroBeads (Miltenyi Biotec) and affinity purified with MACS separation columns (Miltenyi Biotec). Calcofluor staining was done by incubating cells with 5 μg/ml Fluorescent Brightener 28 (Sigma–Aldrich) for 1 min prior to centrifugation and resuspension to SC media. The Q-bodies were induced by incubating cells in a water bath at +42 °C for 30 min, or by using a temperature-controlled microfluidic chamber (see microfluidics). For the mating experiments, cells (1.85×107) of the opposite mating type were centrifuged 500 g for 5 min and resuspended to 40 μl SC media, mixed and immediately imaged with 1-min intervals (see microscopy for details).
Wide-field microscopy was performed at 30 °C with a DeltaVision microscope coupled to a coolSNAP CCD HQ2 camera (Roper), 250W Xenon lamps and 100X/1.40 and 60x/1.42 NA Olympus oil immersion objectives. Z-sectioning was performed with 500-nm spacing (unless otherwise indicated), obtaining 9 or 11 (live-cell imaging) or 15 (calcofluor stained cells) stacks. Images were deconvolved with Softworx software (Applied Precision). Microfluidics on the aging chip were imaged with a DeltaVision microscope every 45 min for total of 66 hr by obtaining seven z-sections with 0.6-μm spacing. FRAP was done with temperature-controlled Zeiss LSM 510 microscope controlled with AIM LSM4.0 software, 63x 1.4NA Oil DIC Plan-Apochromat objective at 30°C, using DPSS and Argon lasers. Diploid cells expressing HSp104-GFP/Hsp104mCherry were imaged every 5 s at three z-planes with 800-nm spacing five times before bleaching the mCherry signal at the age-associated protein deposit with 100% DPSS laser power, after which the recovery of mCherry at the age-associated protein deposit was monitored over the period of two minutes by orienting with the GFP-signal.
The temperature ramp experiments (Figure 4) were carried out with ONIX microfluidic perfusion system equipped with a micro-incubator temperature controller CellASIC), using Y04C microfluidic plates (CellASIC). The temperature setup was the following: 30°C (50 min), 42°C (30 min) 30°C (300 min) and the cells were imaged every 10 min as described in (microscopy). The experiments were carried out in synthetic full medium with even flow rate of 2 psi. Microfluidics on the aging chip was performed with 1.5 μl/min flow rate as described in (Denoth-Lippuner et al., 2014).
All image analyses were performed with Image J software (http://imagej.nih.gov/ij/). The aggregation foci were scored by eye from maximum intensity projected images (spanning the entire volume of the cell) and were defined as puncta that display high-intensity over the surrounding cytoplasmic background signal. For age-associated protein deposit intensity measurements, the integrated density was measured at the site over time at the age-associated protein deposit from stacked, non-processed sum-projected videos and was then normalized to the cytoplasmic Hsp104-GFP intensity.
For the auxin-mediated degradation experiments, the 11-plane z-stack covering 5.5 μm was sum projected and an integrated density of a defined area (18 μm2 (GFP) or 16 μm2 (GFP-NLS) was measured over time from the region of interest (ROI) and the neighboring background region (BG). To distinguish between cells that contained an aggregate from those that did not contain an aggregate, cells were categorized based on the first frame Hsp104-mCherry signal into the two respective groups. The ROI values were background subtracted [ROI(t)-BG(t)] and normalized to the first value [ROI(tx)/ROI(t1)]. The average of the pooled values was fitted with Prism5 software using non-linear one-phase decay.
For the FRAP analysis, raw data were background subtracted, corrected for acquisition-induced bleaching and normalized, and the curve was fitted from pooled values with Prism5 software using one-phase association non-linear fitting. The FRAP data where the aggregate structure was lost from the focal plane during recovery period image acquisition were discarded.
Lifespan of virgin daughters was analyzed on YPD plates using Zeiss Axioscope 40 microdissection microscope as previously described in (Denoth-Lippuner et al., 2014).
All statistical analyses and graph preparations were done with Prism5 software. The error bars represent ±SEM from experimental triplicates with independent clones and statistical analyses were conducted with one-way ANOVA using Newman–Keuls post test, t-test, or Gehan-Breslow-Wilcoxon test.
Enhancing protein disaggregation restores proteasome activity in aged cellsAging 5:802–812.
The asymmetric segregation of damaged proteins is stem cell-type dependentThe Journal of Cell Biology 201:523–530.https://doi.org/10.1083/jcb.201207052
Edc3p and a glutamine/asparagine-rich domain of Lsm4p function in processing body assembly in Saccharomyces cerevisiaeThe Journal of Cell Biology 179:437–449.https://doi.org/10.1083/jcb.200704147
Budding yeast as a model organism to study the effects of ageFEMS Microbiology Reviews 38:300–325.https://doi.org/10.1111/1574-6976.12060
Towards a unifying mechanism for ClpB/Hsp104-mediated protein disaggregation and prion propagationBiochemistry and Cell Biology 88:63–75.https://doi.org/10.1139/o09-118
A mother's sacrifice: what is she keeping for herself?Current Opinion in Cell Biology 20:723–728.https://doi.org/10.1016/j.ceb.2008.09.004
Mother-daughter asymmetry of pH underlies aging and rejuvenation in yeasteLife, 3, 10.7554/eLife.03504.
Role of asymmetric cell division in lifespan control in Saccharomyces cerevisiaeFEMS Yeast Research 14:1133–1146.https://doi.org/10.1111/1567-1364.12216
Liquid-liquid phase separation in biologyAnnual Review of Cell and Developmental Biology 30:39–58.https://doi.org/10.1146/annurev-cellbio-100913-013325
Molecular chaperone functions in protein folding and proteostasisAnnual Review of Biochemistry 82:323–355.https://doi.org/10.1146/annurev-biochem-060208-092442
Curing of the [URE3] prion by Btn2p, a Batten disease-related proteinThe EMBO Journal 27:2725–2735.https://doi.org/10.1038/emboj.2008.198
Metacaspase Yca1 is required for clearance of insoluble protein aggregatesProceedings of the National Academy of Sciences of USA 107:13348–13353.https://doi.org/10.1073/pnas.1006610107
Whole lifespan microscopic observation of budding yeast aging through a microfluidic dissection platformProceedings of the National Academy of Sciences of USA 109:4916–4920.https://doi.org/10.1073/pnas.1113505109
The mother enrichment program: a genetic system for facile replicative life span analysis in Saccharomyces cerevisiae413–422, Genetics, 183, 411SI–413SI, 10.1534/genetics.109.106229.
Molecular chaperones and stress-inducible protein-sorting factors coordinate the spatiotemporal distribution of protein aggregatesMolecular Biology of the Cell 23:3041–3056.https://doi.org/10.1091/mbc.E12-03-0194
Blessings in disguise: biological benefits of prion-like mechanismsTrends in Cell Biology 23:251–259.https://doi.org/10.1016/j.tcb.2013.01.007
Propagation of the yeast prion-like [psi+] determinant is mediated by oligomerization of the SUP35-encoded polypeptide chain release factorThe EMBO Journal 15:3127–3134.
A role for Q/N-rich aggregation-prone regions in P-body localizationJournal of Cell Science 121:2463–2472.https://doi.org/10.1242/jcs.024976
Heritable yeast prions have a highly organized three-dimensional architecture with interfiber structuresProceedings of the National Academy of Sciences of USA 109:14906–14911.https://doi.org/10.1073/pnas.1211976109
Hsp42 is required for sequestration of protein aggregates into deposition sites in Saccharomyces cerevisiaeThe Journal of Cell Biology 195:617–629.https://doi.org/10.1083/jcb.201106037
The SUP35 omnipotent suppressor gene is involved in the maintenance of the non-Mendelian determinant [psi+] in the yeast Saccharomyces cerevisiaeGenetics 137:671–676.
Identification of long-lived proteins retained in cells undergoing repeated asymmetric divisionsProceedings of the National Academy of Sciences of USA 111:14019–14026.https://doi.org/10.1073/pnas.1416079111
Cellular strategies for controlling protein aggregationNature Reviews Molecular Cell Biology 11:777–788.https://doi.org/10.1038/nrm2993
Prion induction involves an ancient system for the sequestration of aggregated proteins and heritable changes in prion fragmentationProceedings of the National Academy of Sciences of USA 107:8633–8638.https://doi.org/10.1073/pnas.1003895107
Hsp70 targets Hsp100 chaperones to substrates for protein disaggregation and prion fragmentationThe Journal of Cell Biology 198:387–404.https://doi.org/10.1083/jcb.201201074
RPN4 is a ligand, substrate, and transcriptional regulator of the 26S proteasome: a negative feedback circuitProceedings of the National Academy of Sciences of USA 98:3056–3061.https://doi.org/10.1073/pnas.071022298
Randy SchekmanReviewing Editor; Howard Hughes Medical Institute, University of California, Berkeley, United States
eLife posts the editorial decision letter and author response on a selection of the published articles (subject to the approval of the authors). An edited version of the letter sent to the authors after peer review is shown, indicating the substantive concerns or comments; minor concerns are not usually shown. Reviewers have the opportunity to discuss the decision before the letter is sent (see review process). Similarly, the author response typically shows only responses to the major concerns raised by the reviewers.
[Editors’ note: this article was originally rejected after discussions between the reviewers, but the authors were invited to resubmit after an appeal against the decision.]
Thank you for choosing to send your work entitled “Early differentiation of the aging lineage in yeast by a protein deposit” for consideration at eLife. Your full submission has been evaluated by Randy Schekman (Senior Editor) and three peer reviewers, and the decision was reached after discussions between the reviewers. Based on our discussions and the individual reviews below, we regret to inform you that your work will not be considered further for publication in eLife.
We consider the work to be technically well done, but largely redundant with a lot of published literature – which you have not referenced properly. In addition there are specific concerns raised by the reviewers that we feel would require substantial experimental work, beyond the normal we suggest for a manuscript that could be returned with revisions. Consequently, in its current form, we believe this work does not advance the fields of aging or protein aggregation as far as other manuscripts that have been published recently in eLife. However, we do like the direction you have taken and think there are “good bones” in this work. Please take the advice of the reviewers into account in designing a new manuscript that you may wish to have us consider at a future date.
The authors observe a single dot of Hsp104-GFP that forms in mother cells after several buddings, and is retained in the mother cell indefinitely. This focus is labeled as well by Ssa1 and Ssa2 (the major cytoplasmic Hsp70s), Hsp42 (a “small” heat shock protein) and Ydj1p (an Hsp40). These foci are shown to be distinct from “Q-bodies” (multiple aggregates formed after heat-stress) or “P-bodies”.
1) There has been a proliferation of “foci” of protein aggregates. What is the relation of the AIDOPOD to the JUNQ (Kaganovich, Kopito and Frydman, 2008) locus, both of which are juxtanuclear? Are these the same aggregates that Nyquist has described (without this name) as retained in the mother cell in an Hsp104-dependent process? Are these the Hsp42 foci of aggregated proteins described by Bukau's group (Specht et al., 2011)? Btn2p forms foci of prion (Kryndushkin and Wickner, 2008) and non-prion aggregates (Kryndushkin and Shewmaker, 2012) and binds well to Hsp42 (Malinovska and Alberti, 2012): are the AIDOPOD foci the same as the Btn2 aggregation foci?
2) In the subsection “AIDOPOD formation is required for asymmetric inheritance of protein aggregates and may promote aging”, the authors state that “Altogether, these data provide evidence that collection of protein aggregates into a single deposit (AIDOPOD) by Hsp42 promotes their asymmetric retention in the mother cell during cell division.” This is not established by the data presented, which examines only the presence of various chaperones in these foci: Hsp104, Hsp70s (Ssa1p, Ssa2p), and Hsp42. Are there protein aggregates in these foci? Or do the authors mean that these are foci of aggregated Hsp104, Hsp70 and Hsp42? I doubt that. It is not established that there are any protein aggregates in the AIDOPOD foci, and they might even be membrane-bound collections of soluble proteins.
3) Figure 3: In [PSI+] cells, Sup35-GFP forms multiple dots (shown by many groups, but first by Patino and Lindquist in 1996). In Figure 3B, there is just one big dot, and otherwise, the GFP seems to been even across the cell, just like the [psi-] cells in Figure 6A. And Sup35-GFP dots in [PSI+] cells are seen in over 50% of the cells, not just in older cells. Something is wrong here.
4) Subheading “Formation of a protein deposit during early replicative aging”: “…Hsp104, a broad sensor for protein aggregates (Figure 1A, (Jackrel et al, 2014)).” Jackrel et al do not show that a dot of Hsp104 means an aggregate of other proteins.
5) The Introduction is not very informative. How about Nyström's asymmetric protein segregation studies and aging? References are cited, but the results in the papers are not mentioned. How about Guarente's work on role of Sir2 and aging?
In summary, the work here is largely well done, but it seems likely that the authors' new name applies to foci described by Bukau, Nyström, Alberti, Kryndushkin, Kaganovich and others. And it is not clear that they are dealing with aggregates.
This manuscript from Saarikangas and Barral characterizes age-induced protein aggregates that are asymmetrically retained in mother cells during replicative aging in S. cerevisiae. Their findings join a fairly extensive literature about protein aggregates, but the authors propose that the aggregates they define are distinct from previously described protein deposits and furthermore, that they contribute to replicative lifespan.
Protein aggregates have been observed and studied in cells for 50 years. But in the last nearly 20 years, there has been a steady appreciation that canonical heat-shock induced proteins are associated with aggregates (Lindquist lab and others). In yeast, Hsp104 was found to promote “disaggregation” of stress-induced protein aggregates, along with the aid of the Hsp70 family of chaperones and the Hsp26 and Hsp42 families of small heat-shock proteins.
More recently it has become clear that not all aggregates are equal. For instance, stress induced “Q bodies” initially form as numerous small protein aggregates that eventually coalesce into one or two larger, terminal aggregates (Kaganovich et al., 2008; Escusa-Toret et al., 2013). Q-body formation is promoted by Hsp42 and antagonized by the disaggregase Hsp104. The best-characterized terminal protein aggregates are the juxtanuclear quality control (JUNQ) compartment and the vacuole-associated insoluble protein deposit (IPOD) compartment. JUNQ is composed of ubiquitinated misfolded proteins near proteasomes on the ER, whereas IPOD consists of vacuole-associated aggregates of terminally misfolded proteins that appear unable to be degraded.
Protein aggregates were first appreciated in yeast replicative aging when carbonylated proteins (a marker of oxidative damage – Aguilaniu et al., 2003) were found to be retained in the mother cell. These damaged proteins reside within Hsp104 foci (aggregates – Erjavec et al., 2007) and their retention in the mother is dependent upon Sir2 and actin.
More recently, the kinetics of Hsp104 foci formation after stress induction has been characterized (Zhou et al., 2014). The foci initiate during translation at the ER and ultimately form aggregates at the ER and mitochondria. The aggregates on mitochondria are retained in the mother in Fis1-dependent process.
In the current manuscript, Saarikangas and Barral characterize an age-induced protein aggregate, which they call the AIDOPOD. It forms under normal, non-stressed conditions and is retained in mother cells. The experiments in this manuscript are technically sound and presented clearly. However, some of the data overlap substantially with previously published work, which is discussed below along with a list of the major findings.
1) AIDOPOD properties: forms naturally in old mother cells, usually within the first three generations; composed of heat-shock proteins Hsp104, Ssa1, Ssa2, Hsp42, and Ydj1, but not Hsp26, Sis1, or Hsp82; inherited asymmetrically to the mother.
Most of the AIDOPOD properties (age-induced foci, asymmetric retention) have been previously described for Hsp104-GFP foci. The current manuscript adds the observation that these foci co-localize with several other heat-shock proteins, although most of these markers have been characterized previously in stress-induced aggregates in young cells.
2) The AIDOPOD fusion-after-mating assays are largely novel and may represent a generally useful tool for studying the effects of different chaperones on protein aggregate formation/disassembly. However, as noted above, previous work has demonstrated by other means that stress-induced foci coalesce into larger aggregates, which presumably aids in asymmetric retention (Escusa-Toret et al., 2013, Coelho et al., 2014).
3) AIDOPOD formation is Hsp42-dependent and antagonized by Hsp104.
A paradigm in which Hsp42 promotes aggregate formation while Hsp104 dissolves aggregates is already well-established and extensively tested in stress-induced aggregates (Specht et al., 2011, Escusa-Toret et al., 2013). The current manuscript extends this paradigm to age-induced protein aggregates.
4) hsp42Δ cells lacking an AIDOPOD are long-lived, whereas hsp104Δ cells, which form an AIDOPOD at an earlier age, are short-lived. hsp104Δ double mutants have many protein aggregate foci that distribute more symmetrically between daughter and mother, and these cells are as short-lived as the hsp104Δ single mutant.
The short lifespan of hsp104Δmutants has been reported, as well as the inability of Hsp104 overexpression to extend lifespan (not reported here, but reported in Erjavec et al., 2007; Andersson et al., 2013). The current authors appear to be the first to demonstrate and interpret the long lifespan of hsp42Δ mutants. The authors argue that Hsp104 is essential for a normal lifespan, whereas Hsp42, by aggregating misfolded or damaged proteins into the AIDOPOD in the mother, initiates mother/daughter aging asymmetry. Both aspects of this model have been proposed separately, but this work is the first to unify the models with regards to aging. A likely prediction from their model is that Hsp42 overexpression might shorten lifespan, as these cells accumulated an AIDOPOD at an earlier age than wild-type cells. The current manuscript would benefit substantially from this experiment.
Taken together, the data presented here will provide a valuable resource for future studies on age-induced protein aggregates. However, much of the manuscript has merely extended existing models from protein aggregates that are stress-induced, particularly Q-bodies, to protein aggregates that are age-induced. Furthermore, previous studies on age-induced Hsp104 aggregates demonstrated that their asymmetric retention is actin and Sir2-dependent (Erjavec et al., 2007), and that their presence is intimately linked to proteasome function (Andersson et al., 2013). The current manuscript ignores this existing framework with respect to characterizing the AIDOPOD and provides minimal mechanistic insight. Therefore, although valuable, the current manuscript does not substantially advance the field or lead to new paradigms, and would be better suited to a journal that accepts more modest advances.
Yeast cells have been proposed to model diverse aspects of aging related to aging in humans. But this argument has failed to capture the imagination of colleagues studying other organisms. This paper could make an important difference.
Even more importantly, the asymmetric segregation of cell fate determinants is of great importance in biology and there are no systems to study this that are as tractable as yeast.
This paper provides an elegant demonstration of the organism's potential for rigorous thoughtful dissection of these problems. For example, the distinction between the age-related deposits and general stress deposits is surprising and important. The same is to be said for the distinction with P-bodies. The results of mating were fascinating. As were the results of the Hsp42 deletion and the interplay between Hsp42 deletion/OE and the deletion/OE of Hsp104, and of mating to further dissect their roles relative to that of Hsp70.
The paper is very well written (except see below) and the diversity and thoughtfulness of the experiments is very impressive.
That said, there are several points that need to be addressed.
Background information, writing, and referencing:
The paper has the abbreviated feeling of having been contributed to another journal with extreme pressure for publication space.
It should be re-written for eLife. For example, no information – or referencing – is provided about the biochemical function of Hsp42, which emerges as such a key player here. Nor is there any reference to phenotypes of Hsp42 mutants. In fact there are very few phenotypes reported, which makes the role of this protein in the aggregation foci particularly interesting.) There is virtually no referencing about Hsp104. Saibil's important work on Hsp104 and Ssa1 localization to specific unique foci in prion-containing cells isn't referenced. Nor are any of the foundational papers on Hsp104 function with Hsp70 in protein disaggregation.
Too many uses of “remarkably”.
Since so much of the work depends on the localization of fusion proteins, the ability of these proteins to rescue WT activity (that is, their functionality) should be referenced or established.
It was very convincing that Hsp104 “aggregation foci” remain in the mother cells. But it was very difficult to see the aggregation foci that are newly arising after several cell divisions in daughters that are becoming aging mothers – Figure 1D, green arrowheads, and movies. This is such a crucial initial point, the evidence has to be convincing. The Materials and methods section here is very brief. Since these foci are difficult to see, was there an objective measure of the formation of these foci? For example, through quantitative high-content imaging? That is clearly implied by Figure 1E, but not established. I have no doubt that the authors are seeing what they claim, but it isn't demonstrated adequately in the figures for readers to appreciate.
A very substantial fraction of yeast cells contain a prion known as [RNQ+]. This prion is known to require Hsp104, to interact with Hsp104 and Hsp40, and to localize to specific cellular sub-compartments. Moreover, it is the one prion that most profoundly influences the aggregation of other proteins. It must be determined whether the cells the authors are studying contain this prion or not. And for at least the beginning phenomena, it should be looked at in [RNQ+] and [rnq-] cells.
I assume the cells they are analyzing for Sup35 foci are negative for the prion it forms? This should be stated. (It would be great to know if this mechanism of Sup35 concentration influenced prion formation. What is the effect of an Hsp42 KO? BUT this is outside the scope of this paper. I am just suggesting it to the authors as an interesting subject. I didn't find it at all surprising that Sup335 didn't influence Hsp104 organization – it shouldn't if the protein aggregating foci are a result of general aging. Perhaps the authors should explain the reasoning behind these observations.)
Is it accurate to compare the foci to an “organelle”? Or to call it a “compartment”? I think their own experiments suggest they are in much more dynamic association with cytoplasmic aggregates than is suggested by that terminology. I am asking the authors to think about whether this phrasing is necessary.
[Editors’ note: what now follows is the decision letter after the authors submitted for further consideration.]
Thank you for resubmitting your work entitled “Protein aggregates are associated with replicative aging without compromising protein quality control” for further consideration at eLife. Your revised article has been favorably evaluated by Randy Schekman (Senior Editor) and three reviewers. The manuscript has been improved but there are some remaining issues that need to be addressed before acceptance, as outlined below:
Reviewer #1 (Major Comments):
I believe that the authors have largely responded to my comments. The comparison of the current work's age-related aggregate site to other aggregate-collecting sites described by others is an important addition to this paper.
Reviewer #1 (Minor Comments):
The work of Nyström is referenced more extensively, but I am not sure there is a frank statement in the paper like that in the response to reviews: “It is likely that the Hsp104-labeled aggregates observed by Erjavec et al. (2007) in aged cells refer to the same structure as those described in our work.”
Moreover, the statement at the end of the Introduction – “Our findings unveil a prevalent and highly coordinated early aging-associated aggregation response that opposes the view of aging and age-associated protein aggregation as a purely stochastic deterioration event” – I never thought it was thought to be “purely stochastic”. Rather, Nyström showed quite clearly that the retention of aggregates in the mother cell was very specific and directed. An opinion to the contrary from another group seems to have been refuted. And “unveils” is a little too strong.
The authors have satisfactorily addressed the concerns I had with the earlier version of this manuscript. While there are similarities between the characterization of Hsp104 foci as described by Nyström and this work, the new characterization of the AIDOPOD and its differences between all the previously described stress induced foci (IPOD, JUNQ, Q-bodies) has never been explored before. Hence I feel it provides a bit of clarity to an otherwise confusing field – especially with respect to the aging field. Note that in the revised manuscript, the authors do a much better job of explaining differences between aging and external stress derived aggregates. Hence, I believe the manuscript is now suitable for publication in eLife.
The paper is much improved. Its importance is nicely summarized in the last paragraph of the Introduction. I would also add that the pace with which protein foci are being reported in human diseases of aging, in yeast models of these proteinopathies, in the prion field, for diverse RNP granules, purinosomes, liquid-to-solid phase transitions of intrinsically disordered proteins, etc. has been accelerating at an amazing rate. Hence there is all the more need to have some distinctions and order brought to the field, for these distinct highly organized age-related deposits.
Wading into the realm of this literature is formidable for the naive reader. So I would suggest that the author note and comment on two recent papers that came out since they submitted their work.
Drummond's group has reported a fascinating formation of profuse, functional protein “aggregates” by particular groups of proteins in response to stress. It should be just mentioned that the current age-related deposits in this manuscript are distinct from those. (Also, see paragraph three, Introduction, the Drummond aggregates don't depend on translation).
Brenda Andrew’s group has reported on the dynamic behavior of many GFP protein foci and overlaps or distinctions should be mentioned.https://doi.org/10.7554/eLife.06197.016
- Juha Saarikangas
- Juha Saarikangas
- Juha Saarikangas
- Yves Barral
- Juha Saarikangas
- Yves Barral
The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.
This study was financially supported by the ETH Zürich, FEBS Long-Term Postdoctoral Fellowship and a Finnish Cultural Foundation (Post Doc Pool) Fellowship (to JS) and through an advanced grant of the European Research Council to YB. We would like to thank Simon Alberti, Judith Frydman, Dan Gottschling, Björn Hegemann, and Matthias Peter for reagents, ScopeM, especially Tobias Schwarz for microscopic support, Marek Krzyzanowski, Sung-Sik Lee, and Bernhard Sebastian for kindly sharing their microfluidic expertize and Fabrice Caudron, Annina Denoth-Lippuner, and Ana-Maria Farcas for helpful comments on the manuscript.
- Randy Schekman, Reviewing Editor, Howard Hughes Medical Institute, University of California, Berkeley, United States
- Received: December 20, 2014
- Accepted: October 19, 2015
- Version of Record published: November 6, 2015 (version 1)
© 2015, Saarikangas and Barral
This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.