An RNA granule for translation quality control in Saccharomyces cerevisiae

Ribosome stalling at oxidized bases and other lesions of mRNAs is problematic because it generates truncated and aggregation-prone polypeptides, which can impair cellular homeostasis (Shan et al., 2003; Simms et al., 2014). Consequently, mRNAs with stalled ribosomes are recognized and cleared by the translation quality control pathways, no-go decay (NGD) and ribosome-associated quality control (RQC). In translation quality control, collided ribosomes at a translation blockage are recognized, whereupon NGD separates their subunits and targets the aberrant mRNA for degradation (Doma and Parker, 2006; Ikeuchi et al., 2019; Juszkiewicz et al., 2018; Shoemaker et al., 2010; Simms et al., 2017). RQC extracts the nascent polypeptide from the 60S ribosomal subunit and targets it for degradation through the ubiquitin-proteasome system (Bengtson and Joazeiro, 2010; Defenouillère et al., 2013; Dimitrova et al., 2009; Shen et al., 2015; Sitron and Brandman, 2019). These pathways are assumed to occur throughout the cytoplasm, based on intracellular distributions of fluorescent protein-tagged factors in genome-wide localization screens (Huh et al., 2003). However, to our knowledge, any further compartmentalization has not yet been demonstrated.

DNA and protein quality control pathways are localized to specialized intracellular compartments (Kaganovich et al., 2008; Paull et al., 2000; Rogakou et al., 1999). NGD has been proposed to occur in processing bodies (P-bodies), and RQC factors are involved in partitioning mRNAs from stalled ribosomes to stress granules (Cole et al., 2009; Moon et al., 2020). However, the proteomes of P-bodies and stress granules do not contain translation quality control proteins, with the exception of Rqc2 in yeast stress granules (Hubstenberger et al., 2017; Jain et al., 2016). Therefore, additional work is required to determine whether translation quality control is compartmentalized and, if so, the nature of the compartment(s) involved.

An avenue to explore RNA quality control was suggested by our identification of cytoplasmic bodies that contain oxidized RNA in human cells (Zhan et al., 2015). These ‘oxidized RNA bodies’ (ORBs) were identified by immunofluorescence (IF) staining of HeLa cells using an antibody against an oxidized form of guanine, 8-oxoguanosine (8-oxoG). ORBs were shown to be distinct from P-bodies and stress granules, but their composition and function(s) remain unknown.

Here, we identify ORBs in Saccharomyces cerevisiae and show that they are RNA granules. In addition, we provide the following evidence that ORBs compartmentalize NGD and RQC: (1) factors of NGD and RQC localize to ORBs in situ; (2) two substrates of NGD and RQC localize to ORBs – oxidized RNA and a stalled mRNA–ribosome–nascent-chain complex (RNC); (3) effects of pharmacological inhibitors reveal that translation is required for ORB formation; and (4) certain mutants deficient for NGD, RQC or both have altered ORB numbers per cell. Finally, 68 ORB proteins were identified by IF microscopy, directed by proteomic analyses of affinity-purified ORBs. These proteins include additional known factors of NGD and RQC and novel candidate factors for future analyses of ORBs and translation quality control.

To determine whether ORBs exist in S. cerevisiae, we visualized the in situ distribution of 8-oxoG by IF microscopy, the approach that revealed ORBs in human cells (Zhan et al., 2015). Cells from cultures in exponential growth phase showed the 8-oxoG IF signal in cytoplasmic foci that numbered 5.4 per cell and measured 450 nm in diameter (Fig. 1A,B). Although 8-oxoG can be in RNA, DNA and nucleotides (Wurtmann and Wolin, 2009), these foci contain oxidized RNA because they were eliminated by RNase A treatment of fixed and permeabilized cells, they stained positive for RNA with an RNA-specific fluorescent dye, and they did not stain for DNA with DAPI (Fig. 1C–E). The specificity of the antibody for 8-oxoG RNA was confirmed by the abovementioned RNase A sensitivity of ORBs and the loss of IF signal upon preincubation of the antibody with 8-oxoG (Fig. S1; Park et al., 1992). Finally, ORBs are none of the organelles that are known to contain RNA and appear as foci in fluorescence microscopy (Fig. S2).

ORBs are RNA granules, that is membrane-less organelles that contain RNA and form by liquid–liquid phase separation (Shin and Brangwynne, 2017). The ORB number per cell was reduced by treatment in vivo with 1,6-hexanediol, which dissolves RNA granules by disrupting the intermolecular interactions that underlie their formation by liquid-liquid phase-separation (Fig. 1C; Alberti et al., 2019; Kroschwald et al., 2017). Like the other RNA granules P-bodies and stress granules, ORBs were stable in lysates, that is, ex vivo (Fig. 1D,F; Jain et al., 2016; Teixeira and Parker, 2007). In addition, these foci were resistant to treatment with the non-ionic detergent Triton X-100 ex vivo, a property of RNA granules but not membranous organelles (Fig. 1F; Fujimura et al., 2009; Shelkovnikova et al., 2014). These ex vivo foci were ORBs because they were the same diameter, stained positive for RNA with an RNA-specific fluorescent dye and were eliminated by RNase A (Fig. 1B,D,F). These results reveal that ORBs exist in yeast and are RNA granules.

Oxidized bases in mRNA stall ribosomes, which need to be resolved by the translation quality control pathways NGD and RQC (Inada, 2020; Yan and Zaher, 2019). Stalled ribosomes are recognized by Hel2, followed by the Dom34- and Hbs1-dependent splitting of their subunits, whereupon Rqc2 extracts the truncated nascent polypeptide from the 60S subunit. Moreover, translation quality control is required for the clearance of oxidized mRNA (Simms et al., 2014). To test whether ORBs compartmentalize these pathways, cells were IF stained for GFP fused to each of the abovementioned translation quality control factors and co-IF stained for 8-oxoG to visualize ORBs. ORBs were seen to have each of these translation quality control proteins (Fig. 2A). Quantification of the overlap (see Materials and Methods) with ORBs was 24% for Dom34–GFP, 25% for Hbs1–GFP, 24% for Hel2–GFP, and 32% for Rqc2–GFP (Fig. 2; Table S1). Fortuitous overlap is highly improbable because ORBs also co-IF-stained for Dom34–GFP or Hbs1–GFP ex vivo, that is, after the dispersal of intracellular material (Fig. 2B). That only subpopulations of ORBs have these proteins is consistent with the heterogeneity of translation-related RNA granules (Anderson and Kedersha, 2006; Buchan et al., 2011; Hoyle et al., 2007; Lui et al., 2014). Translation quality control factor localization to ORBs was surprising because, to our knowledge, they have not been previously reported to localize to foci.

In live cells, Dom34–GFP was in foci, consistent with its localization to ORBs (Fig. S4). Our attempts to determine whether other GFP-tagged NGD factors similarly localize to foci in live cells was severely hampered by our observation that cells without GFP have autofluorescent foci (Fig. S4). This autofluorescence arose in cells undergoing oxidative stress, as it was induced by the ROS hydrogen peroxide (Fig. S4). These foci also formed during treatment with formaldehyde, which induces oxidative stress in S. cerevisiae (Fig. S4; North et al., 2016). Importantly, in situ (i.e. in fixed cells), ORBs were not autofluorescent foci because they were not detected when primary antibodies were excluded (Fig. S1). The discrepancy between our detection of ORBs with Hbs1–GFP in situ but not in vivo could be explained if this localization were induced by stress during formaldehyde fixation in the IF protocol. Supporting this possibility, exposure of cells to various stressors prior to fixation was not seen to increase the number or size of ORBs, consistent with the induction of stress and ORB formation during cell fixation in all cells (Fig. S3). Importantly, under non-stress conditions, the Dom34–GFP foci mentioned above were present in a higher number per cell and in a greater percentage of cells than were the autofluorescent foci (Fig. S4), consistent with ORB localization of Dom34.

If ORBs compartmentalize translation quality control, they would likely be affected by the inhibition of translation. Translational roles of P-bodies and stress granules were revealed, in part, by the effects of translation inhibition by cycloheximide and puromycin (Eulalio et al., 2007; Kedersha et al., 2000). We found that treatment of cells with either drug reduced the number of ORBs per cell by over 50%, supporting a connection between ORBs and translation (Fig. 3A,B). The intensity of the 8-oxoG IF signal throughout cells increased by over 2-fold during translation inhibition (Fig. 3B). This is consistent with the known role of translation in clearing oxidized mRNAs through the NGD pathway (Simms et al., 2014). These results support further a role of ORBs in the compartmentalization of translation quality control.

If ORBs compartmentalize translation quality control, they should be enriched in a translationally arrested RNC, a translation quality control-specific substrate (Brandman et al., 2012; Navickas et al., 2020). An arrested RNC was generated on a URA3 mRNA with a self-cleaving hammerhead ribozyme inserted into its coding region (Fig. S5A,B; Navickas et al., 2020). Ribozyme-catalyzed self-cleavage generates a 3′ truncation, which arrests translating ribosomes. Our rationale was to determine whether the arrested RNC localizes to ORBs by IF staining an HA-epitope tag at the N-terminus of the nascent Ura3 polypeptide. The truncated (Rz) HA–Ura3 nascent polypeptide in a wild-type (WT) translation quality control background, although detected by immunoblot analysis, accumulated only to the levels of non-specific background bands, which would likely interfere with detection of the RNC in situ (Fig. 3C). However, in a dom34Δ background, the defect in NGD enhanced the accumulation of the truncated nascent polypeptide by ∼40-fold, such that the HA–Ura3 truncated nascent polypeptide was the predominant species (Fig. 3C; Doma and Parker, 2006). Moreover, it was shown previously that mRNA1Rz in dom34Δ is bound by an array of colliding ribosomes (Navickas et al., 2020; Tsuboi et al., 2012). Therefore, the truncated HA–Ura3 nascent polypeptide in dom34Δ served as a marker for the arrested RNC in situ.

IF microscopy images showed the HA signal from the arrested RNC localized to ORBs in dom34Δ cells (Fig. 3D). Significantly less ORB localization was seen for the arrested RNC in the WT strain and for the full-length HA–Ura3 polypeptide in either background (Fig. 3D–F). These results are consistent with the levels of the truncated HA–Ura3 nascent polypeptide in Fig. 3C. Given that ribosomes are present in arrested RNCs, we also asked whether ORBs contain ribosomal subunits by IF staining GFP fused to ribosomal proteins. ORBs IF stained for both ribosomal proteins tested; one of each of the subunits (Fig. 3G). Additional evidence of ribosomal protein localization to ORBs is presented below. Therefore, ORBs contain both ribosome subunits, as would be expected for a location of translation quality control. Together, these results demonstrate that a translation quality control-specific substrate localizes to ORBs.

As another test of a role of ORBs in the compartmentalization of translation quality control, we asked whether their size and number per cell are altered in mutants deficient for NGD, RQC or both (Fig. 4A,B). Relative to in WT, hel2Δ cells had 50% more ORBs; hbs1Δ cells had 18% fewer ORBs; dom34Δ-rqc2Δ double mutant cells had 46% fewer ORBs. The latter is a synthetic phenotype; it was not seen in dom34Δ or rqc2Δ single mutants. ORBs in hbs1Δ cells were 30% larger than those in the WT strain. These phenotypes provide functional evidence that ORBs compartmentalize translation quality control.

We used proteomics to explore ORB composition and to further explore their functions. ORBs were immunoaffinity-purified by a procedure developed for stress granules, but using the antibody against 8-oxoG (Jain et al., 2016). 8-oxoG immunoprecipitations were followed by proteomic analyses that compared two conditions, one in which ORBs are observed (no cycloheximide) versus one in which ORBs are depleted (with cycloheximide). Our rationale is that these analyses will help identify candidate proteins for ORBs (i.e. proteins significantly enriched in the ORB-preserving condition). Proteomic analyses identified 822 candidate proteins by at least two unique peptides, and which were not depleted by cycloheximide treatment (Fig. 5; Table S2). By co-IF staining 109 of these GFP-tagged candidate proteins with 8-oxoG in situ, we identified 68 proteins that localize to ORBs (Fig. 5; Table S1; Cdc42 was IF-stained with an antibody against it). These proteins tested positive for ORB localization as determined through a custom-written ImageJ macro (Table S1). To the list of validated ORB proteins, we added eight proteins that were found to localize to ORBs prior to the proteomic analysis but, nonetheless, were supported by one unique peptide and non-depletion in the cycloheximide control. The resulting partial ORB proteome contains 68 proteins, all of which were validated by IF staining of GFP-tagged proteins in situ (Table S1).

This proteome supports further ORBs as being RNA granules with functions in RNA metabolism. Results of GO analysis revealed that, relative to the yeast proteome, the ORB proteome was enriched in protein classes found in cytoplasmic RNA granules (e.g. ATPases, RNA-binding proteins and RNA helicases; Fig. 5A; Hubstenberger et al., 2017; Jain et al., 2016). We identified additional RNA-binding proteins by comparison to published databases (Table S1; Beckmann et al., 2015; Mitchell et al., 2013). Like other RNA granule proteomes, the ORB proteome was enriched in proteins with intrinsically disordered regions and prion-like domains (Fig. 5A). ORB proteins formed networks of physical and genetic interactions that were denser than would occur at random with 4.7 physical interactions per protein (P=3.83×10⁻⁹) and 6.1 genetic interactions (P=1.76×10⁻⁶) (Fig. 5B). Finally, ORBs contained several members of the Ccr4–Not complex, which contributes to most aspects of RNA metabolism (Table S1; Collart, 2016).

The compartmentalization of translation quality control in ORBs is supported further by the proteome data. These analyses identified four additional translation quality control factors in ORBs – Not4, Cdc48, Cue3 and Rtt101 (Fujii et al., 2009; Yan and Zaher, 2019). Additionally, a total of six ribosomal proteins of both subunits were identified, as expected for a translation quality control compartment. In summary, the validated proteome revealed that ORBs contain proteins with concerted biochemical activities, as would be expected in a membrane-less organelle.

This proteome allowed us to compare and contrast ORBs with P-bodies and stress granules, whose proteomes have been reported (Hubstenberger et al., 2017; Jain et al., 2016). Most proteins in the ORB proteome are in neither P-bodies nor stress granules (Fig. 5C). During validation of the proteome, ORBs did not IF stain positive for two additional P-body proteins (Ccr4 and Xrn1) and six additional stress granule proteins (Hrp1, Pab1, Rbg1, Rio2, Rpo21 and Rvb1) (Table S1). Therefore, ORBs are distinct from stress granules and P-bodies.

The ORB proteome shares a minority of proteins with P-bodies and stress granules (Fig. 5C). The ORB proteome has eight stress granule proteins, nine P-body proteins, and two proteins that are common to both (Dhh1 and Scd6) (Fig. 5C; Table S1). Finally, analyses of the ORB proteome combined with either the stress granule proteome or the P-body proteome revealed dense physical interaction networks, which involved many inter-proteome interactions (Fig. 5D). Therefore, ORBs appear to be functionally related to stress granules and P-bodies, consistent with our evidence for ORB roles in mRNA metabolism and translation.

We present evidence that ORBs are RNA granules, that is, membrane-less cytoplasmic bodies that contain RNA and form by liquid–liquid phase separation. ORBs share functional classes of proteins with P-bodies and stress granules, for example, factors in RNA metabolism and the remodeling of RNP complexes, RNAs or proteins (Fig. 5; Hubstenberger et al., 2017; Jain et al., 2016). ORBs are enriched in proteins with intrinsically disordered regions and prion-like domains, hallmarks of RNA granules. Finally, proteins in our ORB proteome form a dense network of physical interactions, consistent with their being components of a membrane-less organelle.

We found evidence that a subpopulation of ORBs compartmentalize consecutive steps in translation quality control. First, translation is required for both translation quality control and the normal ORB number per cell (Fig. 3A,B). Second, NGD substrates localize to ORBs, that is, oxidized RNA and an arrested RNC (Figs S1 and S3). Third, ORBs contain factors related to NGD and RQC processes, and both ribosome subunits (Figs 2 and 3). Fourth, certain translation quality control-deficient mutants have altered ORB size or number per cell (Fig. 4).

The translation quality control factors in ORBs function in the recognition of arrested RNCs (Hel2 and Cue3), the separation of the stalled ribosome subunits (Dom34, Hbs1 and Rli1), the extraction of the nascent polypeptide from the 60S subunit (by CAT-tailing) or its ubiquitylation (Rqc2, Cdc48 and Not4) (Inada, 2020; Yan and Zaher, 2019). Notably, we did not detect proteins with roles in the endonucleolytic cleavage of the aberrant mRNA (Cue2) or the exonucleolytic degradation of the cleavage products (Xrn1 or most exosome subunits; Table S1; Doma and Parker, 2006; D'Orazio et al., 2019). The possibility that aberrant rRNAs undergo quality control in ORBs is suggested by our detection of Rtt101 and ribosome subunits (Fujii et al., 2009).

ORBs might facilitate translation quality control by maintaining elevated local concentrations of substrates and intermediates to favor forward reactions, and by sequestering aberrant mRNAs and truncated nascent chains to prevent them from undergoing deleterious side reactions. For example, sustained translation initiation on an oxidized mRNA would likely lead to the production of aggregation-prone polypeptides and toxicity (Defenouillère et al., 2016; Jamar et al., 2018; Shan et al., 2003; Yonashiro et al., 2016).

The ORB proteome reveals candidate translation quality control proteins. For example, it contains ribosome nuclear export factors, which prevent association of unassembled ribosomal subunits until they are competent for translation (Woolford and Baserga, 2013). Subunit separation is required in translation quality control when the mRNA and nascent polypeptide are extracted from the disassembled ribosome (Brandman and Hegde, 2016). The ORB proteome includes four nuclear export factors; two for each ribosome subunit (40S, Nob1 and Tsr1; 60S, Lsg1 and Nmd3; Table S1). These could be acting like Rqc2 by separating ribosome subunits during RQC (Lyumkis et al., 2014; Shao et al., 2015; Shen et al., 2015). Therefore, our results open avenues to further dissect translation quality control.

Our finding that only a proportion of ORBs contains translation quality control factors and the arrested RNC, reveal that they are functionally heterogenous. Our proteomic results reveal candidate proteins and hence likely functions of ORB subpopulations involved in other processes.

ORBs are in a rapidly growing class of cytoplasmic RNA granules with functions related to translation (Decker and Parker, 2012; Lui et al., 2014; Panasenko et al., 2019). Our results support a role of ORBs in translation quality control of non-translatable defective mRNAs with arrested ribosomes. By contrast, translation granules are sites of active translation, Not1-containing assemblysomes contain temporarily paused ribosomes, and P-bodies and stress granules handle translationally repressed mRNAs (Ivanov et al., 2018; Lui et al., 2014; Panasenko et al., 2019). Distinctions between ORBs and both P-bodies and stress granules are revealed by our results. This is particularly important because P-bodies are a proposed location of NGD, and stress granules have a relationship with a non-canonical stress-activated RQC pathway (Cole et al., 2009; Moon et al., 2020). Only one translation quality control factor was identified in the proteome of stress granules (Rqc2) and none were identified in the P-body proteome (Hubstenberger et al., 2017; Jain et al., 2016). We show that at least three P-body proteins and seven stress granule proteins were not detected in ORBs (Table S1). These include canonical proteins of P-bodies (Dcp2 and Xrn1) and stress granules (Pab1 and Pub1). In addition, the large majority of the proteins in the ORB proteome were absent from the proteomes of P-bodies and stress granules (Fig. 5C). Ribosomal subunit localization also differs between these RNA granule types; P-bodies lack both subunits, stress granules have only the small subunit, and ORBs have both subunits (Fig. 3G; Table S1; Hubstenberger et al., 2017; Jain et al., 2016). Puromycin treatment stabilizes P-bodies and stress granules but decreased the number of ORBs per cell (Fig. 3A,B; Eulalio et al., 2007; Kedersha et al., 2000). ORBs probably are distinct from translation bodies, RNA granules where gene-specific mRNAs are translated, because cycloheximide has opposite effects on their respective numbers per cell (Fig. 3A,B; Lui et al., 2014). P-bodies and stress granules have roles in the dynamic handling of mRNAs cycling to and from the translated mRNA pool on polysomes. These roles were revealed, in part, by the opposite effects on their presence when the initiation or elongation phases of translation were inhibited, by puromycin or cycloheximide, respectively (Eulalio et al., 2007; Kedersha et al., 2000). By contrast, our findings that both cycloheximide and puromycin reduced ORB number per cell and increased 8-oxoG fluorescence intensity suggest that defective mRNAs do not exit ORBs and return to polysomes, consistent with their degradation by NGD, as expected.

Despite these differences, similarities between ORBs and both stress granules and P-bodies were revealed by the partial overlap of their proteomes. ORB localization was found for nine P-body proteins and eight stress granule proteins (Table S1; Fig. 5C). Two proteins are common to P-bodies, stress granules and ORBs – Dhh1 and Scd6. Analysis of the ORB proteome combined with the proteome of either P-bodies or stress granules revealed dense networks of physical interactions (11.8 average interactions between ORB and stress granule proteins, and 8.5 between ORB and P-body proteins), supporting further functional relationships between ORBs and these RNA granules (Fig. 5D). The high density of interactions between the proteomes of ORB and stress granule is particularly interesting given that stress granules receive mRNAs released from stalled ribosomes in a non-canonical stress-activated RQC pathway (Moon et al., 2020). Therefore, an intriguing possibility is that stress granules receive mRNAs released from translation quality control in ORBs, analogous to the exchange of components that occurs between stress granules and P-bodies (Buchan et al., 2008).

Yeast plasmids were constructed using standard molecular biology procedures starting from p415ADH1 (Mumberg et al., 1995) and were described previously (Navickas et al., 2020). pLB138 expresses a truncated mRNA (Rz) from a 2HA-tagged URA3-RZ (Ribozyme) construct. pLB126 expresses a full length 2HA–URA3 mRNA (FL). The WT strain was BY4741 (MATa his3Δ1 leu2Δ0 met15Δ0 ura3Δ0). All single mutant strains were obtained from the yeast knockout collection in the BY4741 background (Winzeler et al., 1999). All strains with GFP-tagged proteins were obtained from the GFP clone collection in the BY4741 background (Huh et al., 2003). The dom34Δ-rqc2Δ double mutant was generated by one-step gene replacement using PCR fragment of the NatMX6 cassette amplified from plasmid pFA6a-natMX6 (Hentges et al., 2005).

Where indicated, cells were exposed to 100 µg/ml cycloheximide (BioShop), 100 µg/ml puromycin (Bioshop), or 5% 1,6-hexanediol (Sigma) with 10 µg/ml digitonin (Bioshop) for 30 min. For stress conditions, cells were exposed for 30 min to either 5.0 mM hydrogen peroxide, 0.5% sodium azide or medium lacking glucose. To detect autophagosomes, we induced autophagy by transferring cells in mid-log phase to nitrogen starvation minimal medium (SD-N; 0.17% yeast nitrogen base without ammonium sulfate and amino acids and with 2% glucose); for 2 h (Fig. S2). Where indicated, cells were treated with 3.7% formaldehyde for 30 min.

Yeast cells from mid-log phase cultures were incubated on ice for 10 min in lysis buffer (0.2 M NaOH and 0.2% β-mercaptoethanol), whereupon 5% trichloroacetic acid was added followed by a 10-min incubation on ice. Precipitated proteins were pelleted by centrifugation at 12,000 g for 5 min and resuspended in 35 μl of dissociation buffer (4% SDS, 0.1 M Tris-HCl pH 6.8, 4 mM EDTA, 20% glycerol, 2% β-mercaptoethanol and 0.02% Bromophenol Blue). Tris-HCl pH 6.8 was then added to a final concentration of 0.3 M and samples were incubated at 37°C for 10 min. Total protein extracts were subjected to SDS-PAGE and immunoblot analysis. Membranes were reacted with 1:2000 mouse anti-HA primary antibody (Covance, # MMS-101P). Secondary staining used 1:10,000 goat anti-mouse-IgG antibody (KPL).

RNA extraction and northern blotting were performed as described previously (Navickas et al., 2020). Blots were exposed to PhosphorImager screens, scanned using a Typhoon FLA89500 (Fuji), and quantified with ImageJ software.

Yeast cells from mid-log phase cultures were fixed in 3.7% formaldehyde for 15 min at room temperature, washed in phosphate buffer (0.1 M potassium phosphate, pH 6.5), and incubated with 10 mg/ml lyticase (Sigma; in phosphate buffer containing 1.2 M sorbitol) for 30 min at 30°C. Spheroplasts were adhered to 0.1% poly-L-lysine-coated slides for 10 min. Slides were then submerged in ice-cold methanol for 5 min followed by room-temperature acetone for 30 s. Cells were blocked in 2% BSA, 1× PBS and 0.1% Tween-20 for 10 min and incubated with primary antibody [1:500 mouse anti-8-oxoG, QED Bioscience (#12501); 1:5000 rabbit anti-GFP, Thermo Fisher Scientific (#A-11122); 1:5000 mouse anti-HA, Covance (#MMS-101P)] overnight at 4°C in a humidity chamber. Secondary antibodies were Alexa Fluor 647 donkey anti-goat IgG (#A-21447), Alexa Fluor 568 goat anti-mouse IgG (#A-11004), and Alexa Fluor 488 goat anti-rabbit IgG (#A-11008), each diluted 1:300 (Thermo Fisher Scientific). The presence of RNA in ORBs was detected by staining with a 1:500 dilution of SYTO RNASelect™ (Thermo Fisher Scientific, #S32703) for 10 min prior to mounting. Where indicated, fixed and permeabilized cells were treated for 1 h at room temperature with 10 µg/ml RNase A (Fermentas) prior to blocking.

The specificity of the anti-8-oxoG antibody was confirmed by incubating with its antigen 8-hydroxy-2′-deoxyguanosine (abcam; 10 µg/µg antibody) for 2 h before IF staining; the anti-HA antibody is specific because the signal was not seen in a non-transformed strain or when the primary anti-HA antibody is omitted in IF of a pLB126 transformant, and it was significantly lower in the dom34Δ mutant. The GFP antibody is specific because no signal was seen in a non-transformed strain (Fig. S1).

ORBs were prepared from BY4741 cells or the indicated strains from the GFP fusion library (Huh et al., 2003), as described previously for stress granule cores (Jain et al., 2016). Briefly, cell pellets from 50 ml cultures grown to mid-log phase were resuspended in 500 µl lysis buffer [50 mM Tris-HCl pH 7.5, 50 mM NaCl, 5 mM MgCl₂, 50 μg/ml heparin, 0.5 mM DTT, 0.5% Nonidet P-40, fungal protease inhibitor (BioShop) and 1:5000 Antifoam B (Sigma)] and lysed by vortexing with acid-washed glass beads for 2 min followed by 2 min on ice for three cycles. The initial lysate was cleared of unbroken material by centrifugation at 800 g for 2 min at 4°C, and then centrifuged at 17,200 g for 10 min at 4°C. This ORB-enriched pellet fraction was then resuspended in a final volume of 500 µl lysis buffer on ice. To test RNase A and Triton X-100 sensitivity of ORBs ex vivo, the final pellet was resuspended in lysis buffer containing either 10 μg/ml RNase A without heparin or 2% Triton X-100 and incubated for 30 min at room temperature on a nutator before a second centrifugation at 17,200 g for 10 min at 4°C and resuspension in fresh lysis buffer. ORBs were detected by adhering the ORB-enriched fraction to poly-L-lysine-coated slides and probing for 8-oxoG as described above.

Microscopy images were acquired using a Leica DMI 6000 epifluorescence microscope (Leica Microsystems) with a 63×/1.4NA objective, a Hamamatsu OrcaR2 camera and Volocity acquisition software (Perkin-Elmer). Z-stacks were taken by series capture at a thickness of 0.2 µm per section. Stacks were deconvoluted with AutoQuant X3 (Media Cybernetics Inc.). To quantify ORBs in situ we used a custom-written macro in ImageJ. Briefly, each cell was identified in a central Z-slice, and its average 8-oxoG signal intensity determined. ORBs were defined as foci between 9 and 198 pixels² with a fluorescence intensity 1.5-fold higher than the cell average. The number of ORBs per cell, their size and fluorescence intensity were then quantified by using the macro.

To quantify ORBs ex vivo we used the 3D objects counter plugin in ImageJ. ORB colocalization of stalled RNCs or GFP-tagged proteins in situ was determined using a custom-written ImageJ macro. Briefly, foci in each channel were determined as described above, and two foci from different channels were considered colocalized if at least 25% of the area of each foci in each channel overlapped. For a candidate protein to be considered a validated ORB protein, three criteria were required: visual inspection of colocalization must be positive; at least 10% of identified ORBs must colocalize with a GFP foci; at least 18% of cells must have at least one colocalized foci. A minimum of three images were quantified for each biological replicate. One replicate from WT-FL in Fig. 3F was considered an outlier because it was above 2 s.e.m. and was not considered.

ImageJ macros used to quantify images are at https://github.com/Zergeslab/.

ORBs were purified as described previously for stress granule cores (Wheeler et al., 2017). Protein A Dynabeads (Invitrogen) were added to ORB-enriched fractions from three biological replicates of either untreated or cycloheximide-treated cultures for 1 h at room temperature to pre-clear the fraction of non-specific interactions. Cleared lysates were then incubated with anti-8-oxoG (1:300, QED Bioscience) at 4°C for 1 h to capture ORBs, followed by a 1 h incubation with Dynabeads to capture the ORB–antibody complexes. Purified ORBs were resuspended and boiled in SDS-loading buffer for 5 min, then loaded onto a 5% polyacrylamide gel for in-gel trypsin digestion.

Purified ORB proteins were concentrated in a stacking gel using SDS-PAGE stained with Coomassie Brilliant Blue R-250 (Bio-Rad). Proteins in the excised gel band were subjected to in-gel digestion as follows. Gel pieces were incubated for 30 min at room temperature in 50 mM NH₄HCO₃ (Sigma) plus 10 mM dithiothreitol (BioShop) to reduce the proteins and then for 30 min at room temperature with 50 mM NH₄HCO₃ plus 50 mM iodoacetamide (Sigma) in the dark to alkylate them. Gel dehydration was undertaken with a series of acetonitrile (ACN, BDH) washes. Gel pieces were rehydrated in trypsin digestion solution containing 25 mM NH₄HCO₃ and 10 ng/μl of trypsin (Sigma) followed by incubation overnight at 30°C. Tryptic peptides were extracted three times for 15 min at room temperature with extraction solution [60% acetonitrile plus 0.5% formic acid (FA, Thermo Fisher Scientific), four volumes of the digestion solution]. Peptides were dried using a Speedvac at 43°C and stored at −20°C until MS analysis. Liquid chromatography-tandem MS (LC-MS/MS) analyses were performed on a Thermo EASY nLC II LC system coupled to a Thermo LTQ Orbitrap Velos mass spectrometer equipped with a nanospray ion source. Tryptic peptides were resuspended in solubilization solution containing 97% of water, 2% of ACN and 1% of FA to give a peptide concentration of 100 ng/µl. 2 μl of each sample were injected into a 10 cm×75 μm column that was in-house packed with Michrom Magic C18 stationary phase (5 μm particle diameter and 300 Å pore size). Peptides were eluted using a 90-min gradient at a flow rate of 400 nl/min with mobile phase A (96.9% water, 3% ACN and 0.1% FA) and B (97% ACN, 2.9% water and 0.1% FA). The gradient started at 2% of B, linear gradients of B were achieved to 8% at 16 min, 16% at 53 min, 24% at 69 min, 32% at 74 min, 54% at 81 min, 87% at 84 min followed by an isocratic step at 87% for 3 min and at 2% for 3 min. A full MS spectrum (m/z 400–1400) was acquired in the Orbitrap at a resolution of 60,000, then the ten most abundant multiple charged ions were selected for MS/MS sequencing in linear trap with the option of dynamic exclusion. Peptide fragmentation was performed using collision induced dissociation at normalized collision energy of 35% with activation time of 10 ms. Spectra were internally calibrated using polycyclodimethylsiloxane (m/z 445.12003 Da) as a lock mass.

The MS data were processed using Thermo Proteome Discoverer software (v2.2) with the SEQUEST search engine. The enzyme for database search was chosen as trypsin (full) and maximum missed cleavage sites were set at 2. Mass tolerances of the precursor ion and fragment ion were set at 10 ppm and 0.6 Da, respectively. Static modification on cysteine (carbamidomethyl, +57.021 Da) and dynamic modifications on methionine (oxidation, +15.995 Da) and the N-terminus (acetyl, +42.011 Da) were allowed. The initial list contained 1048 proteins identified with high confidence (false discovery rate <1%) and from combining three biological replicates. Of this list we retained 822 proteins according to criteria stated in the results section.

The ORB proteome was classified according to GO Molecular Function and GO Biological Process information from the SGD database (https://www.yeastgenome.org/). RNA-binding activity was determined using GO analysis and published lists of yeast RNA binding proteins (Beckmann et al., 2015; Mitchell et al., 2013). Networks of physical and genetic interactions were generated and visualized using Cytoscape (version 3.7.2) with the GeneMania plugin (version 3.5.2) (Shannon et al., 2003; Warde-Farley et al., 2010). Proteins with intrinsically disordered regions or prion-like domains were predicted using the SLIDER and PLAAC tools, respectively (Lancaster et al., 2014; Peng et al., 2014).

We thank the following three centres and their specialists at Concordia University for technical advice and access to infrastructure: the Centre for Biological Applications of Mass Spectrometry and Dr Heng Jiang; the Centre for Microscopy and Cell Imaging and Dr Christopher Law, who also wrote the custom ImageJ macros; and the Centre for Structural and Functional Genomics. We also thank Dr Christopher Brett (Concordia University) and the members of his group for their technical advice, materials and access to equipment.

The peer review history is available online at https://journals.biologists.com/jcs/lookup/doi/10.1242/jcs.260388.reviewer-comments.pdf