. Author manuscript; available in PMC: 2020 Apr 18.

Published in final edited form as: Cell. 2019 Apr 11;177(3):711–721.e8. doi: 10.1016/j.cell.2019.02.044

Redox State Controls Phase Separation of the Yeast Ataxin-2 Protein via Reversible Oxidation of its Methionine-Rich Low Complexity Domain

Masato Kato ^1,^*, Yu-San Yang ^1,^*, Benjamin M Sutter ¹, Yun Wang ¹, Steven L McKnight ^1,^#, Benjamin P Tu ^1,^#

PMCID: PMC6752730 NIHMSID: NIHMS1522968 PMID: 30982603

Summary:

Yeast ataxin-2, also known as Pbp1, senses the activity state of mitochondria in order to regulate TORC1. A domain of Pbp1 required to adapt cells to mitochondrial activity is of low sequence complexity. The low complexity (LC) domain of Pbp1 forms labile, cross-β polymers that facilitate phase transition of the protein into liquid-like or gel-like states. Phase transition for other LC domains is reliant upon widely distributed aromatic amino acids. In place of tyrosine or phenylalanine residues prototypically used for phase separation, Pbp1 contains 24 similarly disposed methionine residues. Here we show that the Pbp1 methionine residues are sensitive to H₂O₂-mediated oxidation in vitro and in living cells. Methionine oxidation melts Pbp1 liquid-like droplets in a manner reversed by methionine sulfoxide reductase enzymes. These observations explain how reversible formation of labile polymers by the Pbp1 LC domain enables the protein to function as a sensor of cellular redox state.

Graphical Abstract

graphic file with name nihms-1522968-f0008.jpg

eTOC Blurb

Many proteins that are able to undergo phase transitions contain repeats rich in hydrophobic residues. By contrast, yeast ataxin-2 condensates rely on an abundance of methionines, providing a means for redox-sensitive regulation of its material properties.

Introduction

In response to variable environmental conditions, yeast cells utilize mitochondria in one of two metabolic states. In the presence of fermentable carbon sources, yeast cells use mitochondria for the de novo production of purines, pyrimidines, glutamate, glutamine, nicotinamide adenine dinucleotide (NAD+) and other nitrogenous metabolites (Chen et al., 2017; Laxman et al., 2014). Under conditions of limiting carbon source, yeast mitochondria instead adopt a state of intense respiration wherein molecular oxygen is burned for the purpose of ATP production. Under intense respiration, the mitochondrial pathways normally used for de novo biosynthetic reactions are compromised. Under such respiratory conditions, the target of rapamycin (TOR) pathway must be carefully regulated between growth and autophagic states (Chen et al., 2017). Autophagy in cells undergoing intense respiration, by allowing for the hydrolysis and re-cycling of pre-existing macromolecules, facilitates production of metabolites that would otherwise be synthesized de novo in the mitochondria of cells growing in a nutrient-rich environment.

Experimental observations presented in the accompanying manuscript led to the discovery of the yeast ortholog of ataxin-2, hereafter designated Pbp1, as being essential for cells to control TORC1 activity in response to the activity state of mitochondria (Yang et al., 2019). Previous studies have implicated Pbp1 (poly(A)-binding protein binding protein 1) to be associated with the poly(A)-tails of mRNAs (Mangus et al., 1998) and also to be important to stress granule assembly (Buchan et al., 2008). Related studies of mammalian and fly versions of ataxin-2 have led to the conclusion that it is a component of various forms of neuronal RNA granules (Nonhoff et al., 2007; Ralser et al., 2005), and that it may be involved in the formation of long-term memory and neurodegeneration (Bakthavachalu et al., 2018). Surprisingly, we have found that a key region of Pbp1 necessary for synchronizing mitochondrial and cellular metabolic states corresponds to a low complexity (LC) domain. Upon purification and incubation under physiological conditions, the LC domain of Pbp1 becomes phase separated sequentially into liquid-like droplets and hydrogels (Yang et al., 2019). In this regard, the Pbp1 LC domain behaves similarly to the LC domain of the fused in sarcoma (FUS) protein (Kato et al., 2012), as well as the LC domains of a variety of other RNA-binding proteins (Chong and Forman-Kay, 2016; Lin et al., 2015; Molliex et al., 2015; Xiang et al., 2015).

Phase separation of the FUS LC domain is impeded upon mutational change of tyrosine residues to serine (Han et al., 2012; Kato et al., 2012). Moreover, many subsequent studies have confirmed the importance of tyrosine and/or phenylalanine residues widely distributed among the many LC domains that are subject to phase separation into either liquid-like droplets or hydrogels (Lin et al., 2017; Wang et al., 2018; Xiang et al., 2015). Unlike the LC domains heretofore characterized in many different RNA-binding proteins, the LC domain of Pbp1 is not enriched in aromatic amino acids. The sequence of the Pbp1 LC domain instead contains an unusually high concentration of methionine residues. Within the 150 C-terminal amino acid residues of Pbp1, methionine is represented 24 times, yielding one of the most methionine-rich regions of any gene product of the yeast proteome (Yang et al., 2019). Systematic mutagenesis experiments have given evidence of a correlative relationship between the in vivo function of these methionine residues and the ability of the Pbp1 LC domain to become phase separated into liquid-like droplets having slow exchange properties, and to assemble into labile, cross-β polymers (Yang et al., 2019).

Here we report evidence that the methionine residues conspicuous of the LC domain of Pbp1 may represent redox sensors that facilitate a molecular connection between mitochondria and TORC1. These residues are oxidized in both test tube reactions and living cells upon exposure to low concentrations of hydrogen peroxide (H₂O₂). Methionine oxidation is shown to reverse phase separation of the Pbp1 LC domain. Oxidation is, moreover, relatively protected in a conformationally ordered subregion of the Pbp1 LC domain that facilitates assembly of the protein into labile, cross-β polymers. Finally, H₂O₂-mediated melting of Pbp1 liquid-like droplets is reversed by enzymatic reduction of methionine oxidation via a coupled, electron reduction circuit dependent upon two methionine sulfoxide reductase enzymes, thioredoxin, thioredoxin reductase and NADPH. We propose that the LC domain of Pbp1 is a redox sensor that properly synchronizes cell physiology to the metabolic state of mitochondria.

Results

Effects of methionine oxidation on the yeast ataxin-2 low complexity domain

Liquid-like droplets formed from the LC domains of FUS or Pbp1 were exposed to varying concentrations of excess H₂O₂ (0.01% to 1%). Visual inspection by light microscopy gave evidence of H₂O₂-mediated melting of Pbp1 droplets at intermediate concentrations over the time period examined (3 h) (Figure 1A and Movie S1). H₂O₂ administration failed to affect liquid-like droplets formed from the FUS LC domain at even the highest level of the oxidant tested (Figure 1A and Movie S2). Subsequent to exposure to H₂O₂, protein samples were recovered, solubilized in sodium dodecyl sulfate (SDS) and subjected to polyacrylamide gel electrophoresis (PAGE). Migration of the Pbp1 protein was observed to be altered as a function of H₂O₂ concentration (Figure 1B). Protein samples exposed to the oxidant migrated more slowly than the starting protein, reaching a limit impediment to migration at 1% H₂O₂. Incubation of liquid-like droplets composed of the FUS LC domain with H₂O₂ caused only a modest alteration of the protein as monitored by SDS-PAGE (Figure 1B). An obvious difference in the sequences of the LC domains of Pbp1 and FUS is methionine content. The 214 amino acid residues specifying the FUS LC domain contain but a single methionine residue (M184), whereas the LC domain of Pbp1 contains 24 methionine residues (Figure 1C). From such experiments we tentatively conclude that H₂O₂ effects oxidation of methionine residues in the Pbp1 LC domain, and that such oxidation is incompatible with otherwise attractive interactions necessary for the protein to phase separate into liquid-like droplets.

Figure 1. — A. Liquid-like droplets of His₆-tagged Pbp1 LC (2 mg/mL, ~100 μM) and His₆-tagged FUS LC (2 mg/mL, ~80 μM) domains were treated with the indicated concentrations of H₂O₂ for 3 h. Scale bar = 25 μm. See also Movies S1 and S2.

B. Liquid-like droplet samples of Pbp1 LC and FUS LC domains were oxidized with different concentrations of H₂O₂ for 30 min, followed by SDS-PAGE analyses. Concentrations of H₂O₂ were, from left to right: 0.01%, 0.025%, 0.05%, 0.075%, 0.1%, 0.2%, 0.3%, 0.4%, 0.5%, 0.6%, 0.7%, 0.8%, 0.9% and 1.0%.

C. Amino acid sequences of Pbp1 LC and FUS LC domains. Methionine residues are highlighted in blue.

Site-directed mutational studies of the Pbp1 LC domain have pointed to the correlative importance of certain methionine residues to both phase transition and the in vivo function of the protein in living yeast cells (Yang et al., 2019). Concomitant mutation of eight methionine residues to serine, including residues M591, M595, M605, M606, M614, M616, M618 and M625 (M_7-14 within the LC domain), yielded a protein variant designated “MS8a” (Figure 2A). This mutated variant of Pbp1 was observed to be incapable of polymerizing into labile cross-β polymers (Figure 2B) and was functionally inactive in its ability to couple mitochondrial activity state to autophagy in living yeast cells (Yang et al., 2019). The MS8a variant did become phase separated into liquid-like droplets, but these droplets differed from those formed from the native LC domain of Pbp1 in showing aberrantly rapid protein exchange (Figure 2C). A second variant of Pbp1 also bearing eight methionine-to-serine mutations, including residues M635, M638, M659, M667, M681, M682, M689, and M696 (M_17-24 within the LC domain), yielded a protein variant designated “MS8b”, which was fully capable of polymerizing into labile cross-β polymers (Figure 2B). As deduced from fluorescence recovery after photo-bleaching (FRAP) experiments, droplets formed from the MS8b variant are endowed with a slow exchange rate indistinguishable from droplets derived from the native LC domain (Figure 2C). Finally, the MS8b variant effectively recapitulated the function of Pbp1 in living yeast cells (Figure 2D).

Figure 2. — A. Diagrams showing positions of methionine residues in the LC domain of Pbp1 wild type protein (WT) and those replaced to serine residues in MS8a or MS8b variants.

B. Electron micrographs of dialyzed samples of GFP-fusion LC domains of Pbp1 MS8a or MS8b variants. Scale bar = 500 nm.

C. FRAP analyses of liquid-like droplets formed from LC domains of wild-type Pbp1, MS8a or MS8b variants. Plots are generated from 9 droplets in 3 independent experiments.

D. GFP-cleavage assay that is indicative of mitophagy. Pbp1 MS8a variant is non-functional for autophagy *in vivo* as indicated by the absence of free GFP, whereas Pbp1 MS8b is fully capable of inducing autophagy. YPL: rich medium, SL: minimal medium.

The MS8a and MS8b variants were exposed to graded increases in H₂O₂ before being sized by SDS-PAGE. The pattern of effect of H₂O₂ on the MS8a variant differed from that observed for the intact Pbp1 LC domain in two ways. Following saturation of H₂O₂-mediated oxidation, the extent of impediment to SDS-PAGE migration was reduced for the MS8a variant relative to the native protein (Figure 3A and Figure S1). We tentatively offer that this difference reflects the fact that the MS8a variant contains only 16 methionine residues instead of the 24 methionine residues present in the native Pbp1 LC domain, and that alteration in protein migration is reflective of methionine oxidation. It was further observed that H₂O₂ treatment of the MS8a variant achieved half-maximal change in protein migration at a lower concentration of the oxidant than that required of the intact Pbp1 LC domain (Figure 3B). Half-maximal oxidation of the native LC domain of Pbp1 was observed at 0.33% H₂O₂, as compared to the 0.17% level required to half-maximally oxidize the MS8a variant. For unknown reasons, H₂O₂ more readily oxidizes the 16 residues of the MS8a variant relative to the 24 methionine residues in the native LC domain of Pbp1.

SDS-PAGE analysis of the MS8b variant yielded three differences compared with the native protein (Figure 3A and Figure S1). Like the MS8a variant, saturation oxidation of MS8b yielded protein that was affected less in its SDS-PAGE migration than the fully oxidized form of the native protein. As speculated for the MS8a mutant, we hypothesize that H₂O₂-mediated oxidation of the MS8b variant is restricted to its 16 methionine residues, thus explaining why oxidation causes a less dramatic change in its migration on SDS gels as compared with the native protein that contains 24 methionine residues. SDS gel analysis of the MS8b variant revealed half-maximal oxidation at a higher H₂O₂ concentration than that required for half-maximal oxidation of the native LC domain of Pbp1 (~0.45% versus 0.33%). From the latter observation we offer that the 16 methionine residues retained by the MS8b variant are, for some reason, more resistant to H₂O₂-mediated oxidation than the 16 methionine residues retained by the MS8a variant.

Two distinguishable oxidation products resulting from treatment of the MS8b variant with intermediate concentrations of H₂O₂ were revealed (Figure 3A). These two species migrate slightly differently on SDS-PAGE gels. The methionine residues mutated in the MS8a variant – relative to those of its MS8b counterpart - are more important to Pbp1 for formation of labile cross-β polymers, more important for the formation of “slow FRAP” liquid-like droplets, and more important for the in vivo function of Pbp1 (Figure 2) (Yang et al., 2019). As such, we tentatively conclude that some of the 16 methionine residues remaining in the MS8b variant may be comparatively resistant to H₂O₂-mediated oxidation. Oxidation resistance of the eight methionine residues mutated to serine in the MS8a variant, yet retained by the MS8b variant, may reflect the fact that they exist in an ordered structural state forming the cross-β core of Pbp1 polymers. The eight methionine residues retained by the MS8b variant, but located on either side of the cross-β core, may retain enhanced sensitivity to H₂O₂-mediated oxidation. The more rapid oxidation of the latter residues, relative to methionine residues within the cross-β core, may account for the appearance of the two partially oxidized isoforms of Pbp1 as visualized upon exposure of the MS8b variant to intermediate concentrations of H₂O₂.

A related mutational variant of the Pbp1 LC domain changed a block of eight methionine residues to tyrosine instead of serine, yielding the MY8a variant. This variant simultaneously changed methionine residues M_7-14 to tyrosine (Figure 4A). Liquid-like droplets formed from the MY8a variant were compared with droplets formed from the native LC domain of Pbp1 for sensitivity to H₂O₂-mediated droplet melting. As shown in Figure 4B, liquid-like droplets formed from the MY8a variant of the Pbp1 LC domain were resistant to H₂O₂-mediated droplet melting over the time period examined. The MS8a variant, which simultaneously changed M_7-14 to serine, yielded a protein that failed to facilitate autophagy under conditions of intense respiration, failed to form labile cross-β polymers, and formed aberrantly “FRAP fast” liquid-like droplets (Figure 2). By contrast, the MY8a variant retained the capacity to form labile cross-β polymers (Figure 4C) and formed phase separated liquid-like droplets that were less sensitive to H₂O₂-mediated oxidation (Figure 4B), and which exhibited slower FRAP kinetics (Figure 4D). Moreover, as will be shown in the following section of this report, the MY8a variant specified increased autophagy in living yeast cells. In other words, the MS8a and MY8a mutants displayed reciprocally opposed phenotypes when assayed in vivo, as well as reciprocally opposed biochemical activities as assayed in test tubes.

Figure 4. — A. Diagrams showing positions of methionine residues in the LC domain of Pbp1 that were replaced with tyrosine residues in the MY8a variant. In the MF8a variant, the same methionines were replaced with phenylalanine.

B. Liquid-like droplets of His₆-tagged Pbp1 LC WT and MY8a variant (~100 μM) were treated with different concentrations of H₂O₂ for 3 h. Scale bar = 25 μm.

C. Electron micrographs of dialyzed samples of GFP-fusion LC domains of Pbp1 WT and MY8a variant. Scale bar = 500 nm. A MF8a mutant LC domain was cloned and expressed, but could not be purified and studied in vitro due to its high propensity to aggregate.

D. FRAP analyses of liquid-like droplets formed from LC domains of wild-type Pbp1 and MY8a variant. Plots are generated from 9 droplets in 3 independent experiments.

Oxidation of Pbp1 in living yeast cells

To test whether the LC domain of Pbp1 might be responsive to oxidation in vivo, intact yeast cells growing in culture medium requiring intense mitochondrial respiration were treated with modest concentrations of hydrogen peroxide. Addition of H₂O₂ caused full-length Pbp1 to migrate more slowly on SDS-PAGE gels, suggesting that methionine residues within the protein can be oxidized in living cells (Figure 5A). Such changes in the migration of Pbp1 are consistent with the observed H₂O₂ -induced changes in SDS-PAGE migration for the purified Pbp1 LC domain in vitro (Figure 3A).

Figure 5. — A. Pbp1 can become oxidized in cells. Yeast cells expressing Pbp1-Flag growing in respiratory medium were treated with the indicated concentrations of H₂O₂ for 30 min, quenched, and rapidly harvested for analysis of Pbp1 by SDS-PAGE. Note the retarded migration of Pbp1 in response to higher concentrations of H₂O₂, which likely reflects a maximally oxidized state. Idh1-GFP serves as a loading control.

B. Treatment of cells with H₂O₂ increases the amount of Pbp1 in soluble versus pellet fraction. Cells expressing Pbp1-Flag were grown in SL medium for 6 h with or without H₂O₂. Lysates were centrifuged for separating soluble (S) and pellet (P) fractions. G6PDH is a marker for the (S) fraction.

C. H₂O₂ addition to cells causes Pbp1 to be distributed more uniformly. Cells expressing Pbp1-yoEGFP were treated with or without 0.3% H₂O₂ for 10 min. Plot comparing standard deviation/mean of Pbp1-GFP intensity in individual cells with or without H₂O₂ treatment (p = 1 × 10⁻⁴, two-sample Kolmogorov-Smirnov test).

D. Modest concentrations of H₂O₂ inhibit autophagy. Yeast cells expressing the indicated variants of full-length Pbp1-Flag were switched from YPL to SL medium for 6 h in the presence or absence of the indicated concentrations of H₂O₂. Autophagy was assayed by the GFP-cleavage assay.

E. Full-length MY8a, MF8a variants of Pbp1 are more abundant in pellet fraction compared to WT, whereas the MS8a variant is more abundant in the soluble fraction compared to WT. Cells were grown and processed as in (B). Replicate experiments are shown in Figure S2.

F. MY8a, MF8a variants exhibit increased autophagy. Autophagy was induced and assayed in the indicated strains as in (C).

Having observed methionine oxidation to readily dissolve liquid-like droplets formed by the Pbp1 LC domain (Figure 1A), we next tested whether H₂O₂ treatment might cause dissolution of Pbp1-containing assemblies in cells. Under culture conditions demanding intense mitochondrial respiration, the majority of Pbp1 partitions to the pellet fraction following centrifugation, which reflects propensity of the protein to self-associate into a condensate. Administration of H₂O₂ to cells caused a significant portion of Pbp1 to shift to the soluble fraction, and the protein became more uniformly distributed throughout the cell, suggesting that oxidation of the LC domain may promote dissolution of Pbp1 condensates (Figure 5B–C). Moreover, modest concentrations of H₂O₂ also resulted in significant inhibition of autophagy (Figure 5D), consistent with the melting of Pbp1 condensates and corresponding activation of TORC1 signaling.

Pbp1 methionine mutants resistant to oxidization in vitro exhibit gain of function in vivo

Substitution of M_7-14 by tyrosine with the region of the Pbp1 LC domain required to form labile, cross-β polymers conferred resistance to melting by H₂O₂ in vitro (Figure 4A), suggesting a special role for this subset of methionine residues in condensate formation. We introduced the MY8a as well as the MF8a mutational variant, in the context of full-length Pbp1, into living yeast cells as expressed from its endogenous chromosomal locus. These variants in which methionines were replaced with aromatic amino acids caused increased amounts of the Pbp1 protein isolated from cells to sediment to the pellet fraction (Figure 5E and Figure S2). By contrast, the MS8a variant was observed to destabilize the assemblies as deduced by the presence of increased amounts of Pbp1 in the soluble fraction of cellular lysates. Cells expressing the full-length Pbp1 MY8a or MF8a variant also exhibited enhanced autophagy compared with WT cells (Figure 5F), consistent with the idea that these mutations cause a gain-of-function phenotype owing to the presence of a more stable condensate. By contrast, cells expressing the MS8a variant of Pbp1 exhibited an impediment to autophagy interpreted to reflect a less stable condensate and consequent reduction in the inhibition of TORC1 signaling (Figure 2D).

Pbp1 mediates the cellular response to oxidative stress and mitochondrial dysfunction

Cells expressing the MS8a variant exhibited increased sensitivity to H₂O₂ (Figure 6A), comparable to cells lacking Pbp1 altogether. In stark contrast, yeast cells expressing the Pbp1 MY8a or MF8a variant – compared with WT cells – were equally or slightly more resistant to oxidative stress (Figure 6A). These Pbp1 variants also showed a similar pattern of sensitivity to antimycin A, an inhibitor of mitochondrial electron transport chain complex III (Figure S3). Treatment of yeast cells with either H₂O₂, antimycin A, or cyanide each led to the inhibition of autophagy (Figure 5C and 6B), consistent with observations that TORC1 signaling is normally activated in response to mitochondrial dysfunction (Khan et al., 2017; Liu and Butow, 2006). However, expression of the MY8a or MF8a variants of Pbp1 reduced the ability of these agents to block autophagy, suggesting that such condensates more potently inhibit TORC1 owing to their increased resistance to oxidation-induced dissolution (Figure 5C and 6B). Taken together, these observations reveal that Pbp1 not only senses H₂O₂ through key methionine residues within its LC domain as a proxy for mitochondrial dysfunction, but is also required to properly modulate TORC1 signaling for the cellular response to mitochondrial respiratory dysfunction.

Enzymatic reduction of oxidized Pbp1 revives formation of liquid-like droplets:

Oxidation of methionine takes place on its sulfur atom, producing one of two possible stereogenic sulfoxides corresponding to two stable diasteroisomers of L-methionine sulfoxide (Achilli et al., 2015). Moskovitz, Stadtman, and colleagues described the discovery and enzymatic properties of two evolutionarily unrelated enzymes designated methionine sulfoxide reductase A (MsrA) and methionine sulfoxide reductase B (MsrB) (Moskovitz et al., 1997; Moskovitz et al., 1995). These enzymes efficiently reduce oxidized methionine so long as being coupled to a redox relay composed of thioredoxin, thioredoxin reductase and NADPH (Figure 7A). We expressed and purified MBP-fusion MsrA and MsrB (yeast Mxr1 and Mxr2), His₆-tagged thioredoxin (yeast Trx1), and His₆-tagged thioredoxin reductase (yeast Trr1) so that it might be possible to test the reversal of H₂O₂-mediated oxidation of Pbp1 in the presence of NADPH.

Liquid-like droplets were formed using the purified LC domain of Pbp1. Subsequent to H₂O₂-mediated melting, reaction mixes were quenched by addition of a two-fold molar excess of sodium sulfite (Na₂SO₃) relative to H₂O₂, then exposed to the MsrA and/or MsrB enzymes in combination with thioredoxin, thioredoxin reductase and NADPH. Partial re-formation of liquid-like droplets, as observed by droplet size and number, was observed upon addition of either MsrA or MsrB so long as the reactions were supplemented with thioredoxin, thioredoxin reductase and NADPH (Figure 7B). Addition of both methionine sulfoxide reductase enzymes, supplemented by all co-factors required for redox relay, led to complete droplet re-formation (Figure 7B and Movie S1). Withdrawal of any of the co-factor supplements prevented re-formation of Pbp1 liquid-like droplets. We likewise observed that enzymatically inactive variants of MsrA and MsrB failed to yield any display of revival of H₂O₂-melted liquid-like droplets (Figure 7B).

Pbp1 protein samples exposed to the aforementioned components were examined by SDS-PAGE (Figure 7C). Exposure of the oxidized Pbp1 protein to MsrA led to a form of the protein that migrated approximately half-way between the fully reduced and oxidized forms of the protein (so long as co-incubated with thioredoxin, thioredoxin reductase and NADPH). MsrB likewise achieved intermediate reduction of oxidized Pbp1 as deduced by SDS-PAGE (Figure 7C). Reactions supplemented by both MsrA and MsrB and all redox relay co-factors yielded Pbp1 protein that migrated on SDS gels at a position indistinguishable from the fully reduced protein (Figure 7C). These data give evidence that H₂O₂-mediated oxidation of Pbp1 yields both methionine sulfoxide diastereomers. Moreover, having observed partial revival of liquid-like droplets from oxidized Pbp1 by either of the two methionine sulfoxide reductase enzymes, we tentatively conclude that H₂O₂ treatment of Pbp1 leads to modification of the protein in the forms of both sterogenic sulfoxide isomers. Since partial droplet revival was achieved by either of the methionine sulfoxide reductase enzymes, revival does not appear to require complete reduction of all methionine residues within the LC domain of the protein. In conclusion, these data confirm four facts: (i) enzymatic reduction of oxidized Pbp1 protein by the MsrA and MsrB methionine sulfoxide reductase enzymes cannot be achieved if even a single constituent of the redox relay is dropped out of the reaction (thioredoxin, thioredoxin reductase, or NADPH); (ii) no evidence of reduction was observed when NADPH was replaced by the oxidized cofactor (NADP+); (iii) active site mutants of the MsrA and MsrB enzymes failed to reduce the oxidized form of Pbp1 and revive liquid-like droplets; and (iv) enzymatic reduction of H₂O₂-oxidized Pbp1 by the MsrA and MsrB enzymes fully restores the ability of the protein to become phase separated into liquid-like droplets. Similarly, the addition of catalytically active MsrA, MsrB, and components of the redox relay to full-length, oxidized Pbp1 isolated from yeast cells restored the protein to the reduced form as assessed by SDS-PAGE analysis (Figure 7D), confirming that Pbp1 is also subject to reversible methionine oxidation in vivo.

Discussion

Here we describe studies of the unusual properties of the low complexity (LC) domain of the yeast ortholog of ataxin-2. Like many other LC domains, a purified form of the Pbp1 LC domain becomes phase separated upon incubation under conditions of physiological salt concentration and neutral pH. Pbp1 rapidly forms liquid-like droplets upon incubation at a low mg/ml protein concentration. Prolonged incubation of the protein results in the formation of hydrogel droplets composed of labile, amyloid-like polymers (Yang et al., 2019). Unlike other LC domains studied to date, phase separated Pbp1 is solubilized upon exposure to low concentrations of hydrogen peroxide (H₂O₂) (Figure 1A). This phenomenon of liquid-like droplet dissolution was found to parallel H₂O₂-mediated oxidation of methionine residues as visualized by changes in Pbp1 protein migration on SDS polyacrylamide gels (Figure 1B). Enzymatic reduction of the oxidized methionine residues, via two methionine sulfoxide reductase enzymes, thioredoxin, thioredoxin reductase and NADPH, allows reformation of liquid-like droplets (Figure 7B). We therefore propose that the LC domain of Pbp1 utilizes methionine residues, instead of tyrosine and/or phenylalanine residues prototypic of other LC domains, so that it can be converted back-and-forth between soluble and labile polymeric states in response to redox state.

Systematic mutagenesis was employed to probe the relative importance of specific methionine residues within the Pbp1 LC domain. Such studies allowed comparative analysis of Pbp1 function in both living yeast cells and defined biochemical reactions. A pattern of non-equivalence was observed, such that a group of eight methionine residues located within the amino-terminal half (M_7-14) of the LC domain were found to be more important for the in vivo and in vitro functions of Pbp1 than a group of eight methionine residues located within the carboxyl-terminal half (M_17-24) of the domain. Concomitant conversion of the former residues to serine, yielding the MS8a variant, significantly impeded the in vivo function of Pbp1 (Figures 2, 5, 6). In parallel, this mutated variant of the protein failed to polymerize into labile, cross-β polymers, and produced liquid-like droplets having aberrantly rapid exchange properties (Figures 2B and 2D). By contrast, amino acid substitutions of eight methionine residues within the C-terminal half of the Pbp1 LC domain yielded the MS8b protein variant that was fully functional in vivo, fully capable of forming labile cross-β polymers, and fully capable of forming liquid-like droplets endowed with the same, slow-exchange kinetics as droplets formed from the intact LC domain (Figure 2).

Unlike the loss-of-function phenotype observed for the MS8a variant, concomitant conversion of the same eight methionine residues to tyrosine yielded a protein variant that exhibited a dominant gain-of-function phenotype. Upon introduction into yeast cells, this MY8a variant allowed for increased autophagy (Figure 5). As studied in vitro, the MY8a variant formed liquid-like droplets that were resistant to H₂O₂-mediated dissolution (Figure 4B). In combination, these in vivo and in vitro experimental observations pinpoint methionine residues 591, 595, 605, 606, 614, 616, 618 and 625 (M_7-14) as components of a redox sensor. In simplistic terms, this device is interpreted to constitute a receptor for reactive oxygen species differentially produced by mitochondria as a function of nutritional environment. We propose that the Pbp1 receptor can exist in one of two states - polymeric or monomeric – with modification by reactive oxygen species favoring the latter state. Consistent with this idea, methionine oxidation has been reported to impede fibril formation of the α-synuclein and Aβ proteins (Johansson et al., 2007; Uversky et al., 2002). We further propose that the oligomeric state of the Pbp1 receptor is used to control activity of the TORC1 complex such that cells properly choose whether to be autophagic or not.

A vivid example of the ability of cells to exist in different metabolic states is described by the yeast metabolic cycle (YMC). When prototrophic (virile) strains of yeast are grown in a chemostat under nutrient-limiting conditions, the entire culture synchronously cycles back and forth from states of intense versus reduced respiration. The remarkably robust YMC is 4-5 hours in length and defined by precise and wholesale changes in gene expression, metabolite abundance and oscillating adaptation in many fundamental aspects of cell biology including cell division (Tu et al., 2005; Tu et al., 2007). The precise coordination of the YMC is reliant upon the ability of regulatory proteins to both perceive metabolic state and adaptively exact regulatory changes to optimize cell fitness. The methionine-rich LC domain of Pbp1 represents a regulatory protein endowed with a simple redox sensor. This sensor allows cells to continuously monitor the respiratory status of mitochondria and adapt other aspects of cell physiology accordingly. Therefore, Pbp1 directly controls cellular fitness at the “ground zero” level of being the precise molecular receptor that senses reactive oxygen species.

The human ataxin-2 protein contains a C-terminal LC domain roughly 400 amino acids in length. The human ataxin-2 LC domain includes 18 evolutionarily conserved methionine residues, eight of which are paired as unusual doublets analogous to the five methionine doublets within the Pbp1 LC domain that we have investigated in the present study. We thus speculate that the human ataxin-2 protein may perform an analogous mechanistic function to that described herein for Pbp1. The human ataxin-2 protein is of central relevance to neurodegenerative disease. Variants of the protein that contain a polyglutamine expansion are associated with enhanced susceptibility to spinal cerebellar atrophy or amyotrophic lateral sclerosis (Elden et al., 2010; Ross et al., 2011). The poly-glutamine expansion in the human ataxin-2 protein might abet self-association of the protein in a manner analogous to the MY8a variant of Pbp1. If so, disease pathophysiology might reflect recalcitrance of glutamine-expanded ataxin-2 to properly sense the metabolic activity of mitochondria. Alternatively, the glutamine-expansion might impede the ability of the human ataxin-2 protein to properly enter its mitochondria-proximal “cloud” of loosely-assembled cross-β polymers – perhaps favoring its aggregation in another part of the cell. In this case, the phenotypic deficits of the poly-glutamine expansion might mimic the loss-of-function phenotype described herein for the MS8a variant of Pbp1. In either scenario, proper communication between mitochondrial activity and cellular metabolic state would be anticipated to be impeded.

It is likewise notable that early human genetic studies on ALS gave evidence of the involvement of a copper:zinc superoxide dismutase enzyme now termed SOD1 (Rosen et al., 1993). This SOD1 enzyme, going back to the early origins of enzymology and biochemistry, is understood to catalyze dismutation of superoxide radical anions into H₂O₂ (McCord and Fridovich, 1988). Transgenic expression of mutated versions of human SOD1 in mice prompts various forms of neurodegenerative disease (Gurney et al., 1994). Whether dominant gain-of-function mutations in SOD1 modify the ability of cells to detoxify oxygen radicals is unclear. We hope that the studies reported herein might offer a new and simplified perspective concerning both the cellular function of the ataxin-2 protein and its role in health and disease.

We close by comparing the LC domain of Pbp1 with that of the fused in sarcoma (FUS) RNA binding protein, an intensely studied LC domain. These LC domains are of similar size, and both contain subregions having a relatively stronger propensity to form labile cross-β interactions. The polymer-forming core of the FUS LC domain is roughly 60 amino acids in length and is localized within its amino-terminal half. The atomic structure of polymers formed from this FUS LC domain core has been resolved by solid state NMR spectroscopy (Murray et al., 2017). It is also the case that the polymer forming core of Pbp1 is sub-localized within the amino-terminal half of its LC domain. Both of these cores form labile, cross-β structures, and both polymers can be readily disassembled via post-translational modification. Phosphorylation of the FUS LC domain core by DNA-dependent protein kinase (DNAPK) causes polymer disassembly resulting in the melting of liquid-like droplets (Murray et al., 2017). As reported herein, oxidation of methionine residues within the Pbp1 LC domain core causes polymer disassembly also resulting in the melting of liquid-like droplets (Figure 1). Mutated variants of FUS that block DNAPK-mediated phosphorylation of the LC domain core specify the formation of liquid-like droplets immune to phosphorylation-mediated melting (Murray et al., 2017). Likewise, mutated variants of Pbp1 that cannot be oxidized within its LC domain core cannot be melted by H₂O₂ (Figure 4). Thus, although organized in architecturally similar ways, the FUS and Pbp1 LC domains appear to have evolved so as to be susceptible to entirely different sorts of regulatory input. That both proteins dynamically interconvert between loosely folded polymers and unstructured monomers enables facile side chain access for post-translational modification, such that FUS is exceedingly susceptible to DNAPK-mediated phosphorylation and Pbp1 is exceedingly susceptible to methionine oxidation. We offer that these sorts of studies are beginning to frame the organizational and informational logic as to how LC domains function to abet the robustness of biological systems.

STAR METHODS

CONTACT FOR REAGENT AND RESOURCE SHARING

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Benjamin Tu (benjamin.tu@utsouthwestern.edu)

EXPERIMENTAL MODEL AND SUBJECT DETAILS

Yeast strains

The prototrophic Saccharomyces cerevisiae CEN.PK strain was used in all experiments (van Dijken et al., 2000). All strains used in this study are listed in the KEY RESOURCES TABLE with some from a previous study (Wu and Tu, 2011). Gene deletions were carried out using standard PCR-based strategies to amplify resistance cassettes with appropriate flanking sequences, and replacing the target gene by homologous recombination (Longtine et al., 1998). C-terminal tags were similarly made using PCR to amplify resistance cassettes with flanking sequences. The diploid strain was constructed by standard mating procedures. Pbp1 (yeast ataxin-2) mutants with various domain deletion or point mutations were first made using PCR and then integrated into the PBP1 locus in a pbp1Δ strain with different selection markers.

KEY RESOURCES TABLE

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Anti-Flag antibody	Sigma	Cat#F1804; RRID: AB_262044
Anti-GFP monoclonal antibody	Roche	Cat#11814460001; RRID: AB_390913
Anti-G6PDH antibody	Sigma	Cat#A9521; RRID: AB_258454
Anti-Flag antibody	Cell Signaling	Cat#2368; RRID: AB_2217020
Bacterial and Virus Strains
BL21(DE3)	Novagen	69450-3
Rossetta(DE3)	Novagen	70954-3
Chemicals, Peptides, and Recombinant Proteins
Ni-NTA agarose resin	GoldBio	Cat#H-320-500
30% H₂O₂ solution	Sigma	Cat#H1009-100ml
Dynabeads protein G	Life technologies	Cat#10004D
Critical Commercial Assays
BCA protein assay kit	Thermo	Cat#23227
Experimental Models: Organisms/Strains
S.cerevisiae strain CEN.PK: genotype: WT	van Dijken et al., 2000	N/A
S.cerevisiae strain CEN.PK: genotype: pbp1Δ::Hyg	This paper	N/A
S.cerevisiae strain CEN.PK: genotype: Idh1-GFP::KanMX	Wu and Tu, 2011	N/A
S.cerevisiae strain CEN.PK: genotype: Idh1-GFP::KanMX, pbp1Δ::Hyg	This paper	N/A
S.cerevisiae strain CEN.PK: genotype: Idh1-GFP::KanMX, pbp1Δ::pbp1M591, 595, 605, 606, 614,616, 618, 625S-3xFlag::Hyg (MS8a)	This paper	N/A
S.cerevisiae strain CEN.PK: genotype: Idh1-GFP::KanMX, pbp1Δ::pbp1M635, 638, 657, 667, 681,682, 689, 696S-3xFlag::Hyg (MS8b)	This paper	N/A
S.cerevisiae strain CEN.PK: genotype: Idh1-GFP::KanMX, pbp1Δ::pbp1M614,616Y-3xFlag::Hyg (M₁₁,₁₂Y)	This paper	N/A
S.cerevisiae strain CEN.PK: genotype: Idh1-GFP::KanMX, pbp1Δ::pbp1M614,616, 618, 625Y-3xFlag::Hyg (M_11-14Y)	This paper	N/A
S.cerevisiae strain CEN.PK: genotype: Idh1-GFP::KanMX, pbp1Δ::pbp1M605, 606, 614,616, 618, 625Y-3xFlag::Hyg (M_9-14Y)	This paper	N/A
S.cerevisiae strain CEN.PK: genotype: Idh1-GFP::KanMX, pbp1Δ::pbp1M591, 595, 605, 606, 614,616, 618, 625Y-3xFlag::Hyg (MY8a)	This paper	N/A
S.cerevisiae strain CEN.PK: genotype: Idh1-GFP::KanMX, pbp1Δ::pbp1M614,616F-3xFlag::Hyg (M₁₁,₁₂F)	This paper	N/A
S.cerevisiae strain CEN.PK: genotype: Idh1-GFP::KanMX, pbp1Δ::pbp1M614,616, 618, 625F-3xFlag::Hyg (M_11-14F)	This paper	N/A
S.cerevisiae strain CEN.PK: genotype: Idh1-GFP::KanMX, pbp1Δ::pbp1M605, 606, 614,616, 618, 625F-3xFlag::Hyg (M_9-14F)	This paper	N/A
S.cerevisiae strain CEN.PK: genotype: Idh1-GFP::KanMX, pbp1Δ::pbp1M591, 595, 605, 606, 614,616, 618, 625F-3xFlag::Hyg (MF8a)	This paper	N/A
S.cerevisiae strain CEN.PK: genotype: pbp1 Δ::pbp1M591, 595, 605, 606, 614,616, 618, 625S-3xFlag::Hyg (MS8a)	This paper	N/A
S.cerevisiae strain CEN.PK: genotype: pbp1 Δ::pbp1M635, 638, 657, 667, 681,682, 689, 696S-3xFlag::Hyg (MS8b)	This paper	N/A
S.cerevisiae strain CEN.PK: genotype: pbp1 Δ::pbp1M591, 595, 605, 606, 614,616, 618, 625Y-3xFlag::Hyg (MY8a)	This paper	N/A
S.cerevisiae strain CEN.PK: genotype: pbp1 Δ::pbp1M591, 595, 605, 606, 614,616, 618, 625F-3xFlag::Hyg (MF8a)	This paper	N/A
S.cerevisiae strain CEN.PK: genotype: Pbp1-yoEGFP::KanMX	This paper	N/A
Recombinant DNA
pHis-parallel-yATXN2 LC	Yang et al., 2019	N/A
pHis-parallel-FUS LC (human)	Kato et al., 2012	N/A
pHis-parallel-GFP-yATXN2 LC	Yang et al., 2019	N/A
pHis-parallel-yATXN2 LC MS8a	Yang et al., 2019	N/A
pHis-parallel-GFP-yATXN2 LC MS8a	Yang et al., 2019	N/A
pHis-parallel-yATXN2 LC MS8b	This paper	N/A
pHis-parallel-GFP-yATXN2 LC MS8b	This paper	N/A
pHis-parallel-yATXN2 LC MY8a	This paper	N/A
pHis-parallel-GFP-yATXN2 LC MY8a	This paper	N/A
yATXN2-Flag	Yang et al., 2019	N/A
yATXN2-Flag MS8a	Yang et al., 2019	N/A
yATXN2-Flag MS8b	Yang et al., 2019	N/A
yATXN2-Flag MY8a	Yang et al., 2019	N/A
yATXN2-Flag MF8a	Yang et al., 2019	N/A
pHis-parallel-MBP-MsrA	This paper	N/A
pHis-parallel-MBP-MsrA C25A/C176A	This paper	N/A
pHis-parallel-MBP-MsrB	This paper	N/A
pHis-parallel-MBP-MsrB C97A/C157A	This paper	N/A
pHis-parallel-Trx1	This paper	N/A
pHis-parallel-Trr1	This paper	N/A
Software and Algorithms
ImageJ	NIH	https://imagej.nih.gov/ij
Prism7	Graphpad software	https://www.graphpad.com/
QuickTime Player 7 Pro	Apple	https://support.apple.com/kb/DL923?locale=en_US
Code for image analysis	Yang et al., 2019	https://github.com/tulabutsw/image_analysis

Open in a new tab

Yeast growth media and procedures

Media used in this study

YPL (0.5% yeast extract (Bio Basic), 2% peptone (BD Biosciences) and 2% lactate (Sigma L1375)); SL (0.17% yeast nitrogen base without amino acids (BD Biosciences), 2% lactate).

Media switch

Cells from overnight YPL culture were inoculated into fresh YPL to 0.1 optical density (OD₆₀₀)/ml and grown for a few generations in log phase. The log-phase cells were then diluted again to very low OD₆₀₀, grown overnight till the OD₆₀₀ reached ~1. The cells were then spun down, washed, and re-suspended to OD₆₀₀ = 0.7 in SL medium. Samples were collected at 6 h after switch.

METHOD DETAILS

Protein expression and purification

His₆-tagged human FUS LC domain (residues 2-214) was prepared as described before (Kato et al., 2012). All Pbp1 LC domains (both GFP-fusion and His₆-tagged) were prepared as described in the accompanying article (Yang et al., 2019). DNA fragments of yeast MsrA, MsrB, Trx1 and Trr1 were amplified using PCR from yeast cDNA library. MsrA and B fragments were inserted into the multiple cloning sites of the pHis-MBP-parallel1 vector, and Trx1 and Trr1 fragments were cloned into pHis-parallel1 vector (Sheffield et al., 1999). Catalytically dead mutants of MBP-MsrA (C25A/C176A) and MBP-MsrB (C97A/C157A) (Ma et al., 2011) were generated by quickchange site-directed mutagenesis. All constructs were confirmed by sequencing.

MBP-MsrA and MBP-MsrB were overexpressed in E. coli Rosetta(DE3) cells with 0.5 mM IPTG at 16°C overnight. Trx1 and Trr1 were overexpressed in E. coli BL21(DE3) cells with 0.5 mM IPTG at 20°C for overnight. Cells were harvested by centrifugation, washed with PBS and stored at −80°C. For purification of these four proteins, harvested cells were lysed with 0.2 mg/ml lysozyme in a lysis buffer containing 50 mM Tris-HCl pH 7.5, 500 mM NaCl, 20 mM 2-mercaptoethanol (βME), 1% Triton X-100, 10% glycerol and a protein inhibitor tablet (Sigma) for 30 min on ice, and then sonicated for 2 min (10s on / 30s off). The cell lysate was centrifuged at 35,000 rpm for 1 h. The supernatant was mixed with Ni-NTA resin (QIAGEN) for 30 min at 4°C. The Ni-NTA resin was packed in a glass column and washed with a washing buffer containing 20 mM Tris-HCl pH 7.5, 500 mM NaCl, 20 mM imidazole, 20 mM βME, 0.1 mM PMSF and 10% glycerol. The bound proteins were eluted from the resin with an elution buffer containing 20 mM Tris-HCl (pH 7.5), 500 mM NaCl, 250 mM imidazole, 20 mM βME, 0.1 mM PMSF and 10% glycerol. EDTA was added to a final concentration of 0.5 mM to the eluted protein solutions. MBP-MsrB was further purified with a gel filtration column Superdex S-200 (GE Health Science) with the elution buffer. The purified proteins were concentrated with Amicon Ultra centrifugal filters (Millipore) to the final concentration of 1400 μM for Trx1, 850 μM for Trr1, and 200-400 μM for MBP-MsrA and MBP-MsrB. The protein solutions were stored at −80°C. The purities of the purified proteins were confirmed by SDS-PAGE, and the concentrations were determined by absorbance at UV₂₈₀.

H₂O₂-mediated melting of phase-separated liquid-like droplets

Liquid-like droplets of His₆-Pbp1_LC were formed as described in the accompanying article (Yang et al., 2019). Liquid-like droplets of His₆-FUS LC were formed as described before (Murray et al., 2017). The liquid-like droplet solutions (100 μl) were incubated for 1 h at room temperature (RT). H₂O₂ solution (1%, 3%, 10% or 30%) was added to the droplet solution by either 100-fold or 33.3-fold dilution to obtain final concentrations of 0.01%, 0.03%, 0.1%, 0.3% and 1%. After 3 h incubation at RT, images of the droplets were taken using Bio-Rad ZOE Fluorescent Cell Imager with a white light mode.

To visualize protein oxidation caused by H₂O₂, a portion (10 μl) of the reaction mixture was recovered from the chamber after 30 min incubation and then analyzed by SDS-PAGE. To analyze band shifts, the distances between the initial band and each band in different H₂O₂ concentrations were measured using the program ImageJ. The distances were then normalized by the maximum distances measured from the highest H₂O₂ concentration (1%). Plots were curve-fitted with a one-phase association equation using the program Prism7 (Graphpad, USA) to obtain half-maximal oxidation concentrations. For the MS8b variant, the distances were measured at the midpoints of discrete bands with bias for a denser band. Thus, the half-maximal oxidation concentration for MS8b was visually estimated from the plots.

TEM imaging of polymers

To inspect polymer formation of the wild type and mutants of GFP-fusion Pbp1_LC, the proteins (200-300 μM) were dialyzed in a gelation buffer at RT for overnight. Preparation of EM grids was described in the accompanying article (Yang et al., 2019). TEM images were obtained with a JEOL 1200EX electron microscope at 120 kV.

Fluorescence recovery after photobleaching (FRAP) and quantification

FRAP experiments were carried out as described in the accompanying article (Yang et al., 2019). Briefly, GFP-Pbp1_LC was first diluted in a gelation buffer at a final concentration of 1 μM. Immediately, His₆-Pbp1_LC was diluted into this solution at a final concentration of 2 mg/ml (~100 μM). The phase-separated droplet solution was dispensed in a glass-bottomed microscope dish and incubated at room temperature for 2 h. GFP labeled phase-separated droplets were visualized by confocal microscope (Zeiss LSM880 Airyscan). Time lapse images were recorded with a 5s interval for a total of 63 images. Photobleaching of GFP signals in selected droplets was carried out after the 3^rd image was recorded. GFP intensities of photobleached areas were measured for all the images using ImageJ. The intensities from individual images were normalized against those from the initial image. Plots were generated and curve-fitted using Prism 7 program (Graphpad, USA).

Imaging analysis of cells expressing Pbp1-yoEGFP

A strain expressing Pbp1 tagged with yeast optimized EGFP (yoEGFP) was made using homologous recombination (Lee et al., 2013). This strain was used to examine the localization of Pbp1 following treatment of cells with H₂O₂, which causes a decrease in overall fluorescence. The cell cytoplasm was determined through global intensity thresholding using Otsu’s method (Otsu, 1979). Using this optimum threshold the images were divided into foreground, i.e. cell cytoplasm, and background. Next, connected components smaller than 25 pixels were removed and small holes in the foreground, if any, were filled using morphological closure with a 5-by-5 square-structuring element. Next, we calculated the mean and the standard deviation of the pixel intensities on the foreground. The image analysis was implemented in MATLAB 2018a. Specifically, Otsu’s method was implemented using imbinarize function and morphological closing was implemented using imclose function. The code is available at https://github.com/tulabutsw/image_analysis.

Enzymatic reduction of H₂O₂-oxidized protein by methionine sulfoxide reductases

Liquid-like droplets of His₆-tagged Pbp1 LC (100 μl) were first formed in a buffer containing 20 mM Tris-HCl pH 7.5, 200 mM NaCl, 0.5 mM EDTA and 0.1 mM PMSF. The solution was transferred into a well of a 96-well plate, and then the droplets were melted by adding 1 μl of 10% H₂O₂ (final 0.1%). After liquid-like droplets completely disappeared (1-2 h), 3 μl of 2 M Na₂SO₃ solution was added to the mixture in order to quench residual H₂O₂. After incubation for 1 h, 10 μl of 25 mM NADPH solution was added (final ~2.5 mM). Subsequently, 10 μl of 10x enzyme cocktail that contained 1 μM MBP-MsrA, 10 μM MBP-MsrB, 200 μM His₆-tagged Trx1 and 20 μM His₆-tagged Trr1 was added to the mixture. The reaction mixture was incubated at RT and inspected by a light microscope for droplet revival. For controls, MsrA/B were replaced by the catalytically dead mutants, each component (Trx1, Trr1 or NADPH) was removed, or NADPH was replaced by NADP (+NADP). Images were taken with Bio-Rad ZOE Fluorescent Cell Imager with a white light mode after overnight reaction. To visualize protein reduction, a portion (10 μl) of the reaction mixture was recovered from the well and then analyzed by SDS-PAGE. See also Movie S1.

GFP cleavage assay for autophagy

A mitochondrial matrix protein, Idh1, was tagged with GFP. When mitophagy is induced, this protein accumulates in the vacuole and is degraded. The more stable GFP is detected by immunoblotting with anti-GFP monoclonal antibody (Roche, clone 7.1 and 13.1) as semi-quantitative evidence for mitophagy (Kanki et al., 2009). Cells were collected at 6 h after switch from YPL to SL. H₂O₂, Antimycin A, or KCN were added to the SL medium at time 0 at indicated concentration.

Fractionation to assess intracellular assembly formation

Following switch from YPL to SL for 3 h, 20 OD of cells were harvested, flash frozen with liquid nitrogen, and stored at −80°C until cell lysis. The cell pellet was resuspended with 250 μl of lysis buffer (150 mM HEPES pH 7.5, 150 mM NaCl, 1 mM EDTA, 0.2% NP-40, 2X protease inhibitor cocktail, 1 mM PMSF, 1 mM DTT). After adding ~100 μl of glass beads, cells were lysed by bead beating 4 times: 30s of beating / 30s of cooling on ice. Cell lysates were clarified by centrifugation at 650 g for 2 min at 4°C. For total protein sample, 100 μl of lysates were taken and incubated for 1 h at 4°C during the separation of soluble and pellet fractions. For soluble and pellet samples, 100 μl of lysates were centrifuged at 100,000 g for 1 h at 4°C. The supernatant was transferred to another tube and the pellet was resuspended with 100 μl of 0.2% NP-40 lysis buffer. SDS sample dye was then added to total, soluble, and pellet samples and boiled for 5 min at 95°C. There is some unavoidable degradation due to cellular proteases in the lysate, especially in the total samples.

Disc diffusion assay for sensitivity to H₂O₂ or Antimycin A

Cells were grown in YPL and then switched to SL as described in EXPERIMENTAL MODEL AND SUBJECT DETAILS. Cell growth in SL was constantly measured to monitor entry into stationary phase. At stationary phase, 5 OD of cells were plated onto SL plate. Grade 1 filter paper with 1 cm in diameter (Whatman) was placed onto the SL plate and 10 μl of H₂O₂ (1 M) or Antimycin A (200 μg/mL) was soaked into the filter paper. Cells were grown at 30°C for 48 h.

Immunoprecipitation and subsequent reduction of oxidized Pbp1 from yeast cells

Log phase YPL cultures were switched to SL media with or without 0.1% H₂O₂ and grown for 3 h at which time 10 OD of cells were harvested, flash frozen with liquid nitrogen, and stored at −80°C until cell lysis. The cell pellet was resuspended with 175 μl of lysis buffer A (50 mM HEPES pH 7.5, 150 mM NaCl, 10 mM MgCl₂, 0.5% NP-40, 2X protease inhibitor cocktail, 1 mM PMSF, 1 mM sodium orthovanadate, 5 mM NaF, 10 nM leupeptin, 5 nM pepstatin A, 1 mM DTT). After adding ~150 μl of glass beads, cells were lysed by bead beating 6 times: 30 s of beating / 2 min of cooling on ice. The lysed cells were then separated from glass beads by centrifugation at 6000 rpm for 2 min at 4°C, and diluted with 262.5 μL of lysis buffer B devoid of NP-40 (50 mM HEPES pH 7.5, 150 mM NaCl, 10 mM MgCl₂, 2X protease inhibitor cocktail, 1 mM PMSF, 1 mM sodium orthovanadate, 5 mM NaF, 10 nM leupeptin, 5 nM pepstatin A, 1 mM DTT). The crude cell extracts were then clarified by two successive centrifugations at maximum speed for 10 min at 4°C. The protein concentration of the cleared lysates was then measured using Bradford assay (Bio-Rad), and adjusted to be equal among all samples in 420 μl reaction volume. For each coimmunoprecipitation reaction, 12.5 μl of dynabeads protein G (Life technologies) were washed with the IP lysis buffer and incubated with 2.5 μg of anti-Flag antibody (Sigma, M2) for 1 h at 4°C. The unbound antibodies were then removed by centrifugation at 500 x g for 1 min at 4°C, and the conjugated dynabeads-antibody were then added to the cleared lysates. After incubating for 2 h at 4°C, the dynabeads were washed 3 times with wash buffer (50 mM HEPES pH 7.5, 150 mM NaCl, 10 mM MgCl₂, 2X protease inhibitor cocktail, 0.2% NP-40, 1 mM DTT), washed 1 time with reduction reaction buffer (20 mM Tris-HCl pH 7.5, 200 mM NaCl, 0.5 mM EDTA, 1 mM DTT, 2.5 μM NADPH), and split into 4 tubes before adding 30 μl reduction reaction buffer containing 0.1 μM MBP-MsrA, 1 μM MBP-MsrB, 20 μM His-tagged Trx1, and 2 μM His-tagged Trr1. Catalytically dead mutants of MBP-MsrA and/or MBP-MsrB were substituted as indicated. The reaction was incubated at room temperature with mild agitation for 1 h. After removing the reaction mixture, 30 μl 2X SDS sample buffer was added to the dynabeads and denatured for 5 min at 95°C.

Supplementary Material

Figure S1. Comparison of SDS-PAGE migration of WT, MS8a, and MS8b variants of Pbp1 LC domain. Related to Figure 3. The indicated proteins were treated with or without 1% H₂O₂ for 30 min and then sized using SDS-PAGE.

NIHMS1522968-supplement-1.tif^{(1.3MB, tif)}

Figure S2. Replicates of the experiment shown in Figure 5E. Related to Figure 5. Note the increased proportion of full-length Pbp1 MF8a and MY8a variants in the pellet (P) relative to soluble (S) fraction compared to WT, consistent with the formation of more stable assemblies. In contrast, the full-length Pbp1 MS8a variant exhibits increased proportion in the S fraction relative to P compared to WT, consistent with the formation of less stable assemblies.

NIHMS1522968-supplement-2.tiff^{(3MB, tiff)}

Figure S3. Disc-diffusion assay showing sensitivity of strains expressing indicated full-length Pbp1 variants to Antimycin A. Related to Figure 6. Note that the relative sensitivity of indicated strains to Antimycin A is similar to their relative sensitivity to H₂O₂ (Figure 6A).

NIHMS1522968-supplement-3.tiff^{(3MB, tiff)}

Movie S1. Enzymatic revival of Pbp1 LC droplets after melting by H₂O₂. Related to Figures 1 and 7.

Phase separated liquid-like droplets of His-Pbp1 LC domain (2 mg/ml) were formed on a cover glass by a quick dilution method (see Star Methods). Initially, droplets were sinking down on the surface of the cover glass where they fused with other droplets (2 h period). After addition of H₂O₂ (0.1% final), the droplets started melting and disappeared after 1 h. The residual H₂O₂ was quenched by Na₂SO₃ (1 h period, not included in the movie). Subsequently, the methionine sulfoxide reductase cocktail was added, and revival of the droplets was monitored for 2 h. All steps were carried out in the same reaction mixture. Video speeds were different for each step but roughly set up at 45-150× faster than the normal speed. The width of the video image corresponds to 225 μm.

Download video file^{(50MB, mp4)}

Movie S2. Resistance of FUS LC droplets to melting by H₂O₂. Related to Figure 1.

Phase separated liquid-like droplets of His-FUS LC domain (2 mg/ml) were formed on a cover glass by a quick dilution method (see Star Methods). Initially, droplets were sinking down on the surface of the cover glass where they fused with other droplets (30 min period). After addition of H₂O₂ (0.1% final), the droplets stayed without any change for 2 h. Even after 24 h, the majority of the droplets remained. All steps were carried out in the same reaction mixture. A video speed was set up at 150× faster than the normal speed. The width of the video image corresponds to 225 μm.

Download video file^{(56.1MB, mp4)}

Highlights.

Phase separation of the yeast ataxin-2 methionine-rich LC domain is sensitive to H₂O₂
The dissolved droplets are specifically revived by methionine sulfoxide reductases
Methionine to tyrosine variants confer resistance to H₂O₂ in vitro and in cells
Yeast ataxin-2 senses cellular redox state to modulate TORC1 activity

Acknowledgments:

This research was enabled by funding support to B.P.T. from the NIH (R01GM094314), the Welch Foundation (I-1797), HHMI-Simons Foundation, and the Packard Foundation; and funding support to S.L.M. from NIH (U01GM107623) and unrestricted support to S.L.M. provided by an anonymous donor. We thank Y. Lin for assistance with some in vitro experiments, and N.E. Wood for assistance with some image analysis.

Footnotes

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Declaration of Interests:

The authors declare no competing interests.

SUPPLEMENTAL INFORMATION

Supplemental Information includes 3 figures and 2 movies.

References:

Achilli C, Ciana A, and Minetti G (2015). The discovery of methionine sulfoxide reductase enzymes: An historical account and future perspectives. Biofactors 41, 135–152. [DOI] [PubMed] [Google Scholar]
Bakthavachalu B, Huelsmeier J, Sudhakaran IP, Hillebrand J, Singh A, Petrauskas A, Thiagarajan D, Sankaranarayanan M., Mizoue L., Anderson EN., et al. (2018). RNP-Granule Assembly via Ataxin-2 Disordered Domains Is Required for Long-Term Memory and Neurodegeneration. Neuron 98, 754–766 e754. [DOI] [PubMed] [Google Scholar]
Buchan JR, Muhlrad D, and Parker R (2008). P bodies promote stress granule assembly in Saccharomyces cerevisiae. The Journal of cell biology 183, 441–455. [DOI] [PMC free article] [PubMed] [Google Scholar]
Chen J, Sutter BM, Shi L, and Tu BP (2017). GATOR1 regulates nitrogenic cataplerotic reactions of the mitochondrial TCA cycle. Nat Chem Biol 13, 1179–1186. [DOI] [PMC free article] [PubMed] [Google Scholar]
Chong PA, and Forman-Kay JD (2016). Liquid-liquid phase separation in cellular signaling systems. Curr Opin Struct Biol 41, 180–186. [DOI] [PubMed] [Google Scholar]
Elden AC, Kim HJ, Hart MP, Chen-Plotkin AS, Johnson BS, Fang X, Armakola M, Geser F, Greene R, Lu MM, et al. (2010). Ataxin-2 intermediate-length polyglutamine expansions are associated with increased risk for ALS. Nature 466, 1069–1075. [DOI] [PMC free article] [PubMed] [Google Scholar]
Gurney ME, Pu H, Chiu AY, Dal Canto MC, Polchow CY, Alexander DD, Caliendo J, Hentati A, Kwon YW, Deng HX, et al. (1994). Motor neuron degeneration in mice that express a human Cu,Zn superoxide dismutase mutation. Science 264, 1772–1775. [DOI] [PubMed] [Google Scholar]
Han TW, Kato M, Xie S, Wu LC, Mirzaei H, Pei J, Chen M, Xie Y, Allen J, Xiao G et al. (2012). Cell-free formation of RNA granules: bound RNAs identify features and components of cellular assemblies. Cell 149, 768–779. [DOI] [PubMed] [Google Scholar]
Johansson AS, Bergquist J, Volbracht C, Paivio A, Leist M, Lannfelt L, and Westlind-Danielsson A (2007). Attenuated amyloid-beta aggregation and neurotoxicity owing to methionine oxidation. Neuroreport 18, 559–563. [DOI] [PubMed] [Google Scholar]
Kanki T, Kang D, and Klionsky DJ (2009). Monitoring mitophagy in yeast: the Om45-GFP processing assay. Autophagy 5, 1186–1189. [DOI] [PMC free article] [PubMed] [Google Scholar]
Kato M, Han TW, Xie S, Shi K, Du X, Wu LC, Mirzaei H, Goldsmith EJ, Longgood J, Pei J, et al. (2012). Cell-free formation of RNA granules: low complexity sequence domains form dynamic fibers within hydrogels. Cell 149, 753–767. [DOI] [PMC free article] [PubMed] [Google Scholar]
Khan NA, Nikkanen J, Yatsuga S, Jackson C, Wang L, Pradhan S, Kivela R, Pessia A, Velagapudi V, and Suomalainen A (2017). mTORC1 Regulates Mitochondrial Integrated Stress Response and Mitochondrial Myopathy Progression. Cell Metab 26, 419–428 e415. [DOI] [PubMed] [Google Scholar]
Laxman S, Sutter BM, Shi L, and Tu BP (2014). Npr2 inhibits TORC1 to prevent inappropriate utilization of glutamine for biosynthesis of nitrogen-containing metabolites. Science signaling 7, ra120. [DOI] [PMC free article] [PubMed] [Google Scholar]
Lee S, Lim WA, and Thorn KS (2013). Improved blue, green, and red fluorescent protein tagging vectors for S. cerevisiae. PLoS One 8, e67902. [DOI] [PMC free article] [PubMed] [Google Scholar]
Lin Y, Currie SL, and Rosen MK (2017). Intrinsically disordered sequences enable modulation of protein phase separation through distributed tyrosine motifs. J Biol Chem 292, 19110–19120. [DOI] [PMC free article] [PubMed] [Google Scholar]
Lin Y, Protter DS, Rosen MK, and Parker R (2015). Formation and Maturation of Phase-Separated Liquid Droplets by RNA-Binding Proteins. Mol Cell 60, 208–219. [DOI] [PMC free article] [PubMed] [Google Scholar]
Liu Z, and Butow RA (2006). Mitochondrial retrograde signaling. Annu Rev Genet 40, 159–185. [DOI] [PubMed] [Google Scholar]
Longtine MS, McKenzie A 3rd, Demarini DJ, Shah NG, Wach A, Brachat A, Philippsen P, and Pringle JR (1998). Additional modules for versatile and economical PCR-based gene deletion and modification in Saccharomyces cerevisiae. Yeast 14, 953–961. [DOI] [PubMed] [Google Scholar]
Ma XX, Guo PC, Shi WW, Luo M, Tan XF, Chen Y, and Zhou CZ (2011). Structural plasticity of the thioredoxin recognition site of yeast methionine S-sulfoxide reductase Mxr1. J Biol Chem 286, 13430–13437. [DOI] [PMC free article] [PubMed] [Google Scholar]
Mangus DA, Amrani N, and Jacobson A (1998). Pbp1p, a factor interacting with Saccharomyces cerevisiae poly(A)-binding protein, regulates polyadenylation. Mol Cell Biol 18, 7383–7396. [DOI] [PMC free article] [PubMed] [Google Scholar]
McCord JM, and Fridovich I (1988). Superoxide dismutase: the first twenty years (1968-1988). Free Radic Biol Med 5, 363–369. [DOI] [PubMed] [Google Scholar]
Molliex A, Temirov J, Lee J, Coughlin M, Kanagaraj AP, Kim HJ, Mittag T, and Taylor JP (2015). Phase separation by low complexity domains promotes stress granule assembly and drives pathological fibrillization. Cell 163, 123–133. [DOI] [PMC free article] [PubMed] [Google Scholar]
Moskovitz J, Berlett BS, Poston JM, and Stadtman ER (1997). The yeast peptide-methionine sulfoxide reductase functions as an antioxidant in vivo. Proc Natl Acad Sci U S A 94, 9585–9589. [DOI] [PMC free article] [PubMed] [Google Scholar]
Moskovitz J, Rahman MA, Strassman J, Yancey SO, Kushner SR, Brot N, and Weissbach H (1995). Escherichia coli peptide methionine sulfoxide reductase gene: regulation of expression and role in protecting against oxidative damage. J Bacteriol 177, 502–507. [DOI] [PMC free article] [PubMed] [Google Scholar]
Murray DT, Kato M, Lin Y, Thurber KR, Hung I, McKnight SL, and Tycko R (2017). Structure of FUS Protein Fibrils and Its Relevance to Self-Assembly and Phase Separation of Low-Complexity Domains. Cell 171, 615–627 e616. [DOI] [PMC free article] [PubMed] [Google Scholar]
Nonhoff U, Ralser M, Welzel F, Piccini I, Balzereit D, Yaspo ML, Lehrach H, and Krobitsch S (2007). Ataxin-2 interacts with the DEAD/H-box RNA helicase DDX6 and interferes with P-bodies and stress granules. Mol Biol Cell 18, 1385–1396. [DOI] [PMC free article] [PubMed] [Google Scholar]
Otsu N (1979). A threshold selection method from gray-level histograms. IEEE transactions on systems, man, and cybernetics 9, 62–66. [Google Scholar]
Ralser M, Albrecht M, Nonhoff U, Lengauer T, Lehrach H, and Krobitsch S (2005). An integrative approach to gain insights into the cellular function of human ataxin-2. J Mol Biol 346, 203–214. [DOI] [PubMed] [Google Scholar]
Rosen DR, Siddique T, Patterson D, Figlewicz DA, Sapp P, Hentati A, Donaldson D, Goto J, O’Regan JP, Deng HX, et al. (1993). Mutations in Cu/Zn superoxide dismutase gene are associated with familial amyotrophic lateral sclerosis. Nature 362, 59–62. [DOI] [PubMed] [Google Scholar]
Ross OA, Rutherford NJ, Baker M, Soto-Ortolaza AI, Carrasquillo MM, DeJesus-Hernandez M, Adamson J, Li M, Volkening K, Finger E, et al. (2011). Ataxin-2 repeat-length variation and neurodegeneration. Hum Mol Genet 20, 3207–3212. [DOI] [PMC free article] [PubMed] [Google Scholar]
Sheffield P, Garrard S, and Derewenda Z (1999). Overcoming expression and purification problems of RhoGDI using a family of “parallel” expression vectors. Protein Expr Purif 15, 34–39. [DOI] [PubMed] [Google Scholar]
Tu BP, Kudlicki A, Rowicka M, and McKnight SL (2005). Logic of the yeast metabolic cycle: temporal compartmentalization of cellular processes. Science 310, 1152–1158. [DOI] [PubMed] [Google Scholar]
Tu BP, Mohler RE, Liu JC, Dombek KM, Young ET, Synovec RE, and McKnight SL (2007). Cyclic changes in metabolic state during the life of a yeast cell. Proc Natl Acad Sci U S A 104, 16886–16891. [DOI] [PMC free article] [PubMed] [Google Scholar]
Uversky VN, Yamin G, Souillac PO, Goers J, Glaser CB, and Fink AL (2002). Methionine oxidation inhibits fibrillation of human alpha-synuclein in vitro. FEBS Lett 517, 239–244. [DOI] [PubMed] [Google Scholar]
van Dijken JP, Bauer J, Brambilla L, Duboc P, Francois JM, Gancedo C, Giuseppin ML, Heijnen JJ, Hoare M, Lange HC, et al. (2000). An interlaboratory comparison of physiological and genetic properties of four Saccharomyces cerevisiae strains. Enzyme Microb Technol 26, 706–714. [DOI] [PubMed] [Google Scholar]
Wang J, Choi JM, Holehouse AS, Lee HO, Zhang X, Jahnel M, Maharana S, Lemaitre R, Pozniakovsky A, Drechsel D, et al. (2018). A Molecular Grammar Governing the Driving Forces for Phase Separation of Prion-like RNA Binding Proteins. Cell 174, 688–699 e616. [DOI] [PMC free article] [PubMed] [Google Scholar]
Wu X, and Tu BP (2011). Selective regulation of autophagy by the Iml1-Npr2-Npr3 complex in the absence of nitrogen starvation. Mol Biol Cell 22, 4124–4133. [DOI] [PMC free article] [PubMed] [Google Scholar]
Xiang S, Kato M, Wu LC, Lin Y, Ding M, Zhang Y, Yu Y, and McKnight SL (2015). The LC Domain of hnRNPA2 Adopts Similar Conformations in Hydrogel Polymers, Liquid-like Droplets, and Nuclei. Cell 163, 829–839. [DOI] [PMC free article] [PubMed] [Google Scholar]
Yang Y-S, Kato M, Wu X, Litsios A, Sutter BM, Wang Y, Hsu C-H, Wood NE, Lemoff A, Mirzaei H, et al. (2019). Yeast ataxin-2 (Pbp1) forms an intracellular condensate required for the inhibition of TORC1 signaling during respiratory growth. accompanying manuscript, in revision. [DOI] [PMC free article] [PubMed]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

NIHMS1522968-supplement-1.tif^{(1.3MB, tif)}

NIHMS1522968-supplement-2.tiff^{(3MB, tiff)}

NIHMS1522968-supplement-3.tiff^{(3MB, tiff)}

Movie S1. Enzymatic revival of Pbp1 LC droplets after melting by H₂O₂. Related to Figures 1 and 7.

Download video file^{(50MB, mp4)}

Movie S2. Resistance of FUS LC droplets to melting by H₂O₂. Related to Figure 1.

Download video file^{(56.1MB, mp4)}

PERMALINK

Redox State Controls Phase Separation of the Yeast Ataxin-2 Protein via Reversible Oxidation of its Methionine-Rich Low Complexity Domain

Masato Kato

Yu-San Yang

Benjamin M Sutter

Yun Wang

Steven L McKnight

Benjamin P Tu