Yeast chaperones and ubiquitin ligases contribute to proteostasis during arsenite stress by preventing or clearing protein aggregates

Arsenite causes proteotoxicity by targeting nascent proteins for misfolding and aggregation. Here, we assessed how selected yeast chaperones and ubiquitin ligases contribute to proteostasis during arsenite stress. Loss of the ribosome‐associated chaperones Zuo1, Ssz1, and Ssb1/Ssb2 reduced global translation and protein aggregation, and increased arsenite resistance. Loss of cytosolic GimC/prefoldin function led to defective aggregate clearance and arsenite sensitivity. Arsenite did not induce ribosomal stalling or impair ribosome quality control, and ribosome‐associated ubiquitin ligases contributed little to proteostasis. Instead, the cytosolic ubiquitin ligase Rsp5 was important for aggregate clearance and resistance. Our study suggests that damage prevention, by decreased aggregate formation, and damage elimination, by enhanced aggregate clearance, are important protective mechanisms that maintain proteostasis during arsenite stress.

Protein homeostasis (proteostasis) ensures a functional proteome within cells and involves protein quality control (PQC) systems that regulate the fate of proteins from synthesis and folding to degradation [1]. Molecular chaperones are crucial for PQC as they assist in the folding of nascent and misfolded proteins, the disaggregation and refolding of aggregated proteins, and the degradation of proteins that are beyond repair [2,3]. Misfolded proteins that cannot be refolded are degraded by the ubiquitin-proteasome system (UPS) and the autophagy-lysosome pathway [4,5]. Defective PQC and environmental stress, such as heat shock or toxic metals, disrupt proteostasis and result in the cellular accumulation of misfolded and aggregated proteins [6]. Aggregated proteins may be cytotoxic, and protein aggregation underlies many neurodegenerative and age-related disorders [1,5,6].
Folding of nascent proteins can proceed on the ribosome during protein synthesis, in the cytoplasm after ribosomal release, or in specific compartments such as the mitochondrion and the endoplasmic reticulum. Eukaryotes have two ribosome-associated chaperone systems that support de novo protein folding: RAC (ribosome-associated complex) and NAC (nascent polypeptide-associated complex) [7]. In budding yeast Saccharomyces cerevisiae, RAC consists of the Hsp40 protein Zuo1 and the Hsp70 homolog Ssz1 that act together with the ribosome-attached Hsp70s Ssb1 and Ssb2 (SSB) [8,9] while NAC is composed of heterodimers containing Egd2 together with either Egd1 or Btt1 [10]. Studies in yeast have shown that RAC-SSB and NAC associate with virtually every newly translated polypeptide [11,12], and cells defective in NAC and RAC-SSB function accumulate aggregated proteins and lose viability during protein folding stress [11,[13][14][15]. Chaperone systems that act downstream of the ribosome include the GimC/prefoldin complex, the chaperonin TRiC/CCT complex, Hsp90, and the Hsp40-Hsp70 system [6].
Co-translational PQC also involves the ribosomeassociated quality control (RQC) pathway that detects and degrades nascent polypeptides and mRNAs when ribosomes stall during translation [16,17]. Ribosome stalling is toxic for cells, not only due to the production of aberrant polypeptides, but also because ribosomal subunits need to be recycled to enable consecutive rounds of protein synthesis [18][19][20]. Ubiquitin (Ub) ligases play key roles in the RQC pathway. Hel2 recognizes stalled ribosomes and ubiquitinates the small (40S) ribosomal subunit, resulting in ribosome dissociation and subunit recycling [18,19,[21][22][23]. Ltn1 binds to the large ribosomal subunit (60S) and ubiquitinates the incomplete nascent chain, targeting it for degradation [16,17,24]. Not4 has also been implicated in ribosome ubiquitination upon stalling [25] and in ubiquitination of aberrant nascent polypeptides located in stalled 60S particles [26]. Besides ribosomeassociated Ub ligases, yeast cells also possess Ub ligases implicated in PQC in the nucleus, cytoplasm, and various organelles [15,27,28]. Of these, Rsp5 plays a central role in targeting misfolded proteins at the plasma membrane and the cytosol for degradation [29,30].
Exposure to heavy metals and metalloids is associated with several diseases such as cancer and various protein misfolding disorders [31][32][33] and several lines of evidence indicate that cadmium and arsenic affect proteostasis [34][35][36]. Studies in yeast showed that trivalent arsenite [As(III)] induces misfolding and aggregation of nascent or non-native proteins [37][38][39]. As (III) inhibits protein folding primarily by binding to cysteine residues in non-native proteins [35,[40][41][42]. Chaperone-assisted folding or disaggregation may also be affected, either by direct binding to the chaperone or by affecting aggregate structure such that efficient binding of the chaperone to the aggregate is impaired [37,38,42,43]. Additionally, As(III)-misfolded proteins can seed the misfolding and aggregation of other proteins that have not encountered the metalloid [37]. A broad network of cellular systems is implicated in proteostasis during As(III) stress including functions in transcription, translation, and protein folding, turnover, and degradation [39]. These systems protect the proteome during As(III) stress but the role of individual factors is only partially understood. Here, we systematically assessed the contribution of selected PQC systems to proteostasis during As(III) stress focusing on ribosomal and non-ribosomal chaperones and Ub ligases.

Materials and methods
Yeast strains, plasmids, and growth conditions Saccharomyces cerevisiae strains used in this study (Table S1) are based on BY4741 [44], the yeast deletion collection [45], and the collection of temperature-sensitive mutants of essential yeast genes [46].
Plasmids used are described in Table S2. pAG416GPD-Sis1-GFP was constructed via Gateway Recombination Cloning (Thermo Fisher Scientific, Waltham, MA, USA) according to the manufacturer's instructions. The SIS1 sequence was amplified by PCR from pAG413GPD-Sis1-GFP [47] followed by insertion into the destination vector and sequencing.
Yeast cells were cultivated at 30°C on rich YPD medium (1% yeast extract, 2% peptone, 2% glucose) or on minimal synthetic complete (SC) medium (0.67% yeast nitrogen base) supplemented with auxotrophic requirements and 2% glucose as a carbon source. Sodium arsenite (NaAsO 2 ; Sigma-Aldrich, St. Louise, MO, USA) was added to plates or cell cultures at the indicated concentrations. For plate growth assays, cells were grown in SC medium until log phase. Cells corresponding to an optical density (OD) at 600 nm of 0.1 were diluted in 10-fold steps and plated using a sterilized stamp. Plates were incubated for 2-3 days at 30°C.

Fluorescence microscopy
Yeast cells expressing a genomic copy of Hsp104-GFP or Sis1-GFP on a plasmid were grown to mid-log phase in SC medium and exposed to As(III). Cells were collected, fixed with 37% formaldehyde (30 min, room temperature), and washed with 19 PBS. GFP signals were observed using a Zeiss Axiovert 200 M (Carl Zeiss Microscopy, M€ unchen, Germany) fluorescence microscope equipped with Plan-Apochromat 1.40 objectives and appropriate fluorescence light filter sets. Images were taken with a digital camera (AxioCamMR3). The ZEISS ZEN PRO software (Carl Zeiss Microscopy) was used to capture the images. To quantify protein aggregation, the fraction of cells with aggregates (Hsp104-GFP or Sis1-GFP foci) was determined using the IMAGEJ-FIJI software [48].

Flow cytometry
Yeast cells carrying the indicated plasmids were grown in SC medium to mid-log phase, collected before and after addition of 0.5 mM As(III), and fixed (37% formaldehyde, 30 min, room temperature). Around 300 000 cells were collected for each sample and gated on FSC-A and FSC-H to ensure the presence of single cells. The intensity of RFP (PE channel) and GFP (FITC channel) was measured on an LSR FortessaTM X-20 Flow Cytometer (BD Biosciences, Franklin Lakes, NJ, USA). Data were analyzed as described previously [49].

Global translation assays
Global translation was assessed by measuring 35 S-methionine incorporation into newly synthesized proteins as described [50]. Briefly, cells grown to OD 600 of 0.8-1 in YPD were washed and concentrated to an OD 600 of 10 in 5 mL of SC medium lacking methionine. Cells were methionine-starved for 1 h whereafter the culture was split: 0.75 mM As(III) was added to one half, whereas the other half was left untreated. Both cultures were incubated for an additional hour and OD 600 measured. Cells were pulsed with 35 S-methionine (10 lCiÁmL À1 ; Perkin Elmer, Waltham, MA, USA) for 5 min and cycloheximide (300 lgÁmL À1 ; Sigma-Aldrich) added on ice. Ice-cold 50% trichloroacetic acid (TCA; Thermo Fisher Scientific) was added to 1 mL of each culture followed by 20 min at 70°C. The samples were centrifuged and filtered using 25-mm glass fiber filters (Whatman GF/C, 25 mm diameter, 1.2 lm pore size; Sigm-Aldrich) under vacuum. Filters were washed twice with 5% TCA, twice with 95% EtOH, and dried. The radioactivity retained on the filters was determined by liquid scintillation counting and normalized to the OD 600 value before the pulse.
Proteins were boiled (95°C, 5 min), separated on a 4-20% CriterionTM TGX-Stain Free PreCast Gel (Bio-Rad Laboratories, Hercules, CA, USA), and visualized using Chemidoc XRS+ (Bio-Rad) with UV-activation. For western blot analysis, proteins were transferred to a PVDF membrane using TransBlot Turbo transfer system (Bio-Rad). Membranes were blocked with 5% bovine serum albumin in Tris-buffered saline (TBS) containing 0.05% Tween 20 (TBS-T) for 1 h at room temperature followed by over-night incubation at 4°C with an anti-ubiquitin K48 linkage-specific antibody (1 : 1000, rabbit, ab1460601; Abcam, Cambridge, UK) or a K63 linkage-specific antibody (1 : 1000, rabbit, 05-1308; EMD Millipore Corp., Burlington, MA, USA). Membranes were washed 39 with TBS-T, incubated for 2 h with StarBright700 anti-mouse-IgG secondary antibody (1 : 10 000, 10000068185; Bio-Rad) or StarBright700 anti-rabbit-IgG secondary antibody (1 : 5000, 10000068187; Bio-Rad), and washed with TBS-T followed by signal detection using the ChemiDoc XRS+ imaging system (Bio-Rad). Images were analyzed using IMA-GEJ software and quantified as described previously [42]. Briefly, the signal in the aggregate fraction was first normalized to the signal in the corresponding total lysate and then to the total signal over all samples to allow comparison between gels. To quantify K48 and K63 Ub levels in the aggregate fraction, the signal in the aggregate fraction was first normalized to the corresponding total lysate and then normalized to the total signal over all samples to allow comparison between blots.

Intracellular arsenic measurements
Intracellular arsenic was measured as described [39]. Briefly, exponentially growing cells were exposed to 0.5 mM As(III) for 1 h, collected, washed in ice-cold water, and disrupted by boiling for 10 min. After centrifugation, the supernatant was collected, and the arsenic content of each sample was measured by inductively coupled plasma-mass spectrometry (ICP-MS) using an ICAP Q ICP-MS (Thermo Fisher Scientific) with an SC-FAST automated sample introduction system (Elemental Scientific, Omaha, NE, USA).

Results
Loss of ribosome-associated chaperones results in diminished protein aggregation and enhanced As(III) resistance To address how S. cerevisiae cells maintain proteostasis during As(III) stress, we monitored the formation and clearance of protein aggregates using the established green fluorescence protein-tagged marker proteins Sis1-GFP and Hsp104-GFP [37,39,47,[52][53][54][55]. The disaggregase Hsp104 [55] and the Hsp40 Sis1 [54] are present in the cytosol and the nucleus where they are engaged in protein rescue, protein degradation, and gene regulation [56][57][58]. We previously showed that Hsp104-GFP and Sis1-GFP relocalize to distinct foci and associate with aggregated proteins in As(III)exposed cells [37]. Herein, we used cells with a genomic copy of HSP104-GFP or cells transformed with a plasmid harboring SIS1-GFP.
Exposing wild-type cells to As(III) resulted in Sis1-GFP redistribution to distinct foci (protein aggregates) in~90% of the cells within 1 h ( Fig. 1A-C). The fraction of wild-type cells with aggregates/Sis1-GFP foci decreased over time (~50% after 3 h,~30% after 5 h) during exposure, indicating that aggregates were cleared [37,39,42]. Cell deficient in RAC function (zuo1D or ssz1D) showed reduced protein aggregation during As(III) stress (Fig. 1A). The fraction of ssz1D cells with aggregates was lower at the 1 h time point than for wild-type cells, suggesting that less aggregates were formed. The fraction of ssz1D cells with aggregates decreased over time, indicative of clearance. Remarkably, zuo1D cells did not display a detectable increase in protein aggregation (Fig. 1A), raising the possibility that ZUO1 deletion abrogates aggregate formation during As(III) stress. Aggregate formation was also reduced in cells lacking SSB1 and SSB2 (ssb1D ssb2D; hereafter SSBD), and the aggregates formed were efficiently cleared (Fig. 1B). These findings implicate RAC and SSB in proteostasis during As (III) stress. For cells lacking NAC (egd1D egd2D btt1D; hereafter NACD), aggregation levels after 1 h of exposure were similar to that in wild-type cells but somewhat lower at the 5 h time point (Fig. 1C), suggesting a moderate increase in clearance capacity. Notably, NACD, SSBD, zuo1D, and ssz1D cells were all As(III) resistant (Fig. 1D).

Cells lacking RAC and SSB have reduced global translation
The finding that mutants defective in co-translational folding accumulate less aggregates than wild-type cells during As(III) exposure was unexpected given that loss of NAC and RAC-SSB function causes widespread protein aggregation and loss of cell viability during protein folding stress induced by azetidine-2-carboxylic acid (AZC) [11,[13][14][15], a proline analog that, when incorporated, prevents correct folding of nascent proteins. Since intracellular arsenic and protein aggregation levels often correlate [39], we reasoned that the mutants might accumulate less arsenic. However, intracellular arsenic levels were comparable in wildtype and the mutants (Fig. S1).
The RAC-SSB system has been reported to affect ribosome biogenesis, and global translation is reduced in SSBD cells [13,59]. To address whether translation is affected in the chaperone mutants, we measured global translation via incorporation of 35 S-methionine into newly synthesized proteins [50]. Wild-type cells repressed translation in response to As(III) [39,60], as shown by reduced 35 S-methionine incorporation during exposure (Fig. 1E). Importantly, 35 S-methionine incorporation was substantially lower in zuo1D, ssz1D, and SSBD cells compared with the wild-type in the absence of As(III), indicative of diminished global translation. In contrast, 35 S-methionine incorporation was comparable in wildtype and NACD cells. Like the wild-type, all mutants repressed translation in response to As(III). Together, these data suggest that reduced protein synthesis might protect SSBD, zuo1D, and ssz1D from protein aggregation and As(III) toxicity. For NACD, another mechanism is likely to account for its As(III) resistance.
GimC/prefoldin contributes to aggregate clearance during As(III) stress We next assessed the contribution of chaperone systems that operate downstream of the ribosome. GimC/prefoldin (PFD) is a hexameric protein complex that acts as a co-chaperone of TRiC/CCT [61,62], and PFD mutants are As(III) sensitive [43,63]. Deletion of PAC10 (Gim2) or YKE2 (Gim1) resulted in a higher fraction of cells with aggregates (Hsp104-GFP foci) at the 3 and 5 h time points ( Fig. 2A), indicating defective clearance. The pre-stress aggregation levels in pac10D and yke2D were also elevated and the mutants were As (III) sensitive (Fig. 2D) [43,63]. Intracellular arsenic was similar in wild-type and the PFD mutants (Fig. S1). Hence, the As(III) sensitivity of PFD mutants may be a consequence of defective aggregate clearance.
TRiC/CCT is composed of eight paralogous subunits (Cct1 to Cct8 in yeast), and the complex is required for the folding of an important subset of cytosolic proteins, including actin, tubulin, and cell cycle regulators [61,64]. TRiC/CCT activity is inhibited by As(III), and TRiC/ CCT contributes to As(III) resistance [43]. Since deletion of any of the CCT genes causes lethality, we addressed proteostasis in a strain carrying a temperature-sensitive allele of CCT5, cct5-5001. Although cct5-5001 cells had elevated protein aggregation levels (Sis1-GFP foci) in the absence of stress, aggregate formation and clearance were similar in cct5-5001 and wild-type cells during As (III) stress (Fig. 2B), as was intracellular arsenic concentrations (Fig. S1). Thus, the As(III) sensitivity of TRiC/ CCT defective cells (Fig. 2D) [43] is probably not caused by excessive protein aggregation. Hsp90 functions in the folding of a diverse set of proteins involved in different cellular pathways [6]. S. cerevisiae has two cytosolic Hsp90 isoforms encoded by HSC82 and HSP82. Deletion of HSC82 increased protein aggregation (Sis1-GFP foci) levels in the absence of stress (Fig. 2C). However, during As(III) stress, hsp82D and hsc82D cells accumulated and cleared aggregates comparable to the wild-type and their growth were unaffected (Fig. 2D). Hence, the Hsp90 system does not appear to contribute to proteostasis during As(III) stress.

Ribosome-associated ubiquitin ligases have a minor role in proteostasis during As(III) stress
The ribosome-associated Ub ligases Hel2, Ltn1, and Not4 play important roles in RQC and cotranslational PQC [15][16][17][18][19][20]27,28]. As shown previously  [53], a substantial fraction (~60%) of ltn1D cells contained aggregates (Hsp104-GFP foci) in the absence of stress (Fig. 3A). The fraction of ltn1D cells with aggregates increased to~100% within 1 h of exposure and then diminished to pre-stress levels, suggesting that aggregate clearance is not dependent on Ltn1 during As(III) stress. Similarly, the fraction of cells with aggregates (Sis1-GFP foci) was higher in hel2D compared with wild-type before exposure but aggregate formation and clearance were similar in wild-type and hel2D during As(III) stress (Fig. 3B). Aggregation levels (Sis1-GFP foci) were comparable in not4D and wild-type cells except at the 5 h time point where the aggregate levels were somewhat higher in not4D

As(III) does not induce ribosomal stalling or impair RQC
The data above suggested that As(III) may not induce ribosomal stalling or impair RQC. To substantiate this, we made use of established stalling and RQC reporters [49]. To address stalling, we used two different constructs: one that contains a polyarginine (R12) sequence that induces stalling (pTDH3-GFP-T2A2-GFP-linker-R12-T2A2-RFP) and a non-stalling substrate (pTDH3-GFP-T2A2-GFP-linker-ST6-T2A2-RFP). The reporters were introduced into wild-type and hel2D cells, the latter serving as a positive control for reporter read-through, and the RFP and GFP signal intensities were obtained via flow cytometry. The RFP:GFP signal ratio was calculated and normalized to the RFP:GFP ratio of the non-stall reporter. As shown previously [65], HEL2 deletion alleviated stalling leading to an increased RFP:GFP ratio compared with wild-type cells (Fig. 3E). The presence of As(III) did not affect the RFP : GFP ratio, neither in wildtype nor hel2D cells, suggesting that As(III) does not induce ribosome stalling to any major extent.
Next, we evaluated whether As(III) impairs degradation of stalled nascent chains that are RQC substrates. For this, we used two reporter constructs: a GST expression control containing RFP and GST (pTDH3-RFP-T2A2-GST-long-linker-R12-myc), and a construct containing RFP and GFP (pTDH3-RFP-T2A2-GFP-long-linker-R12-myc) [49]. Both constructs were inserted into wild-type and ltn1D cells, the latter serving as a positive control for reduced clearance of RQC substrates. The RFP signal from the GST expression control was subtracted from the RFP signal from the long-linker RQC construct and the GFP : RFP ratio was calculated. As expected, ltn1D cells had a higher GFP : RFP ratio than the wild-type, confirming reduced degradation of the RQC substrate (Fig. 3F) [49]. As(III) exposure did not alter the GFP : RFP ratio neither in wild-type nor in ltn1D, suggesting that As(III) does not impair RQC function or the degradation of stalled nascent chains.
Rsp5 contributes to aggregate clearance during As(III) stress As(III)-treated yeast cells accumulate aggregated proteins with K48-linked Ub chains that are substrates for proteosomal degradation [42]. To identify Ub ligases implicated in this process, we monitored protein aggregation (Sis1-GFP) in mutants that lack Ubr1 (nuclear and cytoplasmic PQC), Hul5 (cytoplasmic PQC), and San1 (nuclear and cytoplasmic PQC), and in a strain harboring a temperature-sensitive allele of Rsp5 (cytoplasmic PQC), rsp5-1 (RSP5 is essential for viability) [27]. Although the fraction of cells with aggregates was somewhat elevated in hul5D and san1D in the absence of stress, aggregate formation and clearance during As(III) exposure was comparable in wildtype and these mutants (Fig. 4A). Aggregate formation was comparable also in wild-type and ubr1D cells, although aggregate levels were somewhat higher in ubr1D at the 5 h time point. Growth of these mutants was unaffected by As(III) (Fig. 4B). Hence, Ubr1, San1, and Hul5 contribute little to proteostasis and growth during As(III) stress. In contrast, rsp5-1 cells showed a clear defect in aggregate clearance (Fig. 4A) and poor growth (Fig. 4B) [66] during As(III) exposure. Rsp5-mediated clearance required its Ub ligase activity, since catalytically inactive Rsp5-C777A could complement neither the clearance defect (Fig. 4C) nor the growth defect of rsp5-1 (Fig. 4B). These results implicate Rsp5 in proteostasis during As(III) stress.
Rsp5 regulates turnover of specific hexose transporters during As(III) stress and their degradation protects cells from As(III) toxicity [66]. We reasoned that stabilization of these hexose transporters in rsp5-1 [66] may be accompanied by increased intracellular arsenic which, in turn, would cause enhanced protein misfolding and aggregation. However, intracellular arsenic levels were comparable in wild-type and rsp5-1 (Fig. S1).
Rsp5 is the main Ub ligase that targets cytosolic proteins that misfold following heat stress for ubiquitination and proteasomal degradation [29]. Since Rsp5 assembles K48-linked Ub chains on heat-misfolded proteins [29,67], we tested its involvement in K48linked ubiquitination of As(III)-aggregated proteins. We biochemically isolated total and aggregated proteins from wild-type and rsp5-1 cells, separated the proteins in each fraction using SDS/PAGE, and subjected the samples to anti-K48 immunoblotting. As shown previously, aggregate formation and clearance generally followed the same dynamics in the biochemical and microscopic assays [42]. Wild-type cells accumulated aggregated proteins (Fig. 4D) with K48linked Ub chains in response to As(III) (Fig. 4E). In the absence of stress, the amount of aggregated proteins with K48-linked Ub chains was higher in rsp5-1 than in the wild-type (Fig. 4D). This increase was not visible using Sis1-GFP as aggregation marker ( Fig. 4A,C), the reason for this discrepancy is unknown. The amount of aggregated proteins with K48-linked chains increased further in rsp5-1 during As(III) exposure. Since K48-linked ubiquitination is not reduced in rsp5-1, Rsp5 is probably not the main Ub ligase mediating K48-linked ubiquitination of As  (III)-aggregated proteins. Rsp5 preferentially assembles K63-linked Ub chains on its substrates [68,69], and we asked whether Rsp5 would mediate K63-linked ubiquitination of As(III)-aggregated proteins. Probing the same samples with an anti-K63 antibody showed a small increase in K63-linked ubiquitination of the As (III)-aggregated proteins in the wild-type (Fig. 4F).
For rsp5-1, the amounts of K63-linked chains in the aggregated protein fraction was high in the absence of stress and did not increase further during exposure (Fig. 4F). Thus, Rsp5 appears dispensable for assembling K63-linked Ub chains on aggregated proteins.

Discussion
Arsenic has a profound impact on proteostasis and is associated with protein misfolding disorders. Therefore, understanding the mechanisms that contribute to proteostasis during As(III) exposure could provide important insights into cellular stress responses and disease processes. Here, we show that loss of the ribosome-associated chaperones RAC and SSB results in reduced global translation and protein aggregation, and increased As(III) resistance (Fig. 1). Similarly, reduced global translation also protected rpb4D cells, lacking an RNA polymerase II subunit, from As(III)induced proteotoxicity [39]. Thus, damage prevention by decreased protein synthesis and hence, protein misfolding and aggregation, appears to be an efficient strategy to safeguard proteostasis and cell proliferation during As(III) stress. This is substantiated by several observations: yeast and mammals respond to As(III) by repressing translation at the initiation stage [39,60,70,71], which reduces protein aggregation [39] and increases survival [39,70]; functions in protein biosynthesis are enriched among mutants that show reduced protein aggregation during As(III) stress and mutants that are As(III) resistance [39]; during As(III) stress, cells down-regulate ribosomal protein levels [60] and expression of genes encoding aggregation-prone proteins [38] as well as RAC, SSB, and NAC components [72]. For NACD cells, global translation was similar to that in wild-type cells, while aggregate clearance was slightly improved. Whether As(III) resistance of NACD is elicited by this modest increase in clearance or through another mechanism, remains unresolved. The impact of As(III) on aggregate formation and clearance, global translation, and growth was not identical in NACD, SSBD, zuo1D, and ssz1D cells. For instance, zuo1D and ssz1D had similar translation and growth but accumulated different amounts of aggregates (Fig. 1). Thus, these chaperones may differ in their protective functions or have different sets of client proteins. Moreover, As(III) sensitivity or resistance could have reasons other than proteotoxicity since As (III) causes toxicity through several additional mechanisms, including oxidative stress, changes to the epigenome, genotoxicity, and altered protein function and activity [73,74]. In yeast, the UPS, autophagy, and the protein disaggregation machinery, each contribute to the clearance of protein aggregates formed during As(III) stress [42]. Here, we show that the PFD complex is implicated in aggregate clearance (Fig. 2) [43,63]. PFD acts as a cochaperone of TRiC/CCT [61,62], but in contrast to the tested PFD mutants, aggregate clearance was unaffected in cct5-5001 cells, even though PFD and TRiC/ CCT defective cells are highly As(III) sensitive [43,63]. These results suggest that PFD and TRiC/CCT provide protection by distinct mechanisms. PFD has, in addition to its role in nascent protein folding, been implicated in transcriptional regulation and protein turnover [62,75]. A role of PFD in protein turnover agrees with the clearance defect observed for pac10D and yke2D. The As(III) sensitivity of TRiC/CCT defective cells may not be a result of widespread protein aggregation but perhaps a result of defective folding of a specific TRiC/CCT client protein.
While ribosome-associated Ub ligases are important for RQC and co-translational ubiquitination of Fig. 3. Ribosome-associated ubiquitin ligases have a minor role in proteostasis during As(III) stress. (A-C) Hsp104-GFP (A) or Sis1-GFP (B, C) distribution was scored by fluorescence microscopy in cells before and after exposure to 0.5 mM As(III). The fractions of cells containing aggregates (Hsp104-GFP or Sis1-GFP foci) were determined by visual inspection of 151-411 (A), 272-384 (B) or 147-420 (C) cells per condition and time point. Data are expressed as mean AE SD from at least three independent biological replicates. * indicates a significant difference (P < 0.05) compared with wild-type (unpaired, two-tailed Student's t-test). (D) 10-fold serial dilutions of the indicated strains were placed onto agar plates with or without As(III). Growth was recorded after 3 days at 30°C. Growth assays were performed with three biological replicates and a representative image is shown. (E) Ribosome stalling. RFP and GFP mean intensity was measured by flow cytometry in wild-type and hel2D cells exposed to 0.5 mM As(III). Data are expressed as the RFP:GFP ratio for the stalling reporter as a percentage of the non-stalling reporter. Data are expressed as mean AE SD from three biological replicates. (F) RQC substrate degradation. GFP and RFP mean intensity was measured via flow cytometry in wild-type and ltn1D cells before and after exposure to 0.5 mM As(III). Data represents the GFP to RFP ratio after subtraction of an RFP expression control. Data are expressed as mean AE SD from three biological replicates.      nascent polypeptides [15][16][17], RQC is not a major protective mechanism or target during As(III) stress (Fig. 3). In mammalian cells, As(III) causes decreased ubiquitination of specific ribosomal proteins and increased read-through of a stalling reporter. Additionally, As(III) binds to ZNF598, an orthologue of yeast Hel2, and the increased read-through of the stalling reporter is abolished in ZNF598 knockout cells [76]. These findings suggest that As(III) directly inhibits ZNF598 activity to alleviate ribosome stalling. Using a similar stalling reporter, we did not detect reporter read-through during As(III) stress, suggesting that Hel2 is not inhibited by As(III). Accordingly, HEL2 deletion did not affect protein aggregation or growth during As(III) stress. It is unclear why As(III) would affect yeast Hel2 and mammalian ZNF598 differently since Hel2 has 19 cysteine residues of which 12 are present in CxxC motifs that are favored As(III) binding sites. Possibly, the As(III) concentration used here might not be sufficient to induce ribosome stalling since mammalian cells are more As(III) sensitive than yeast cells, mainly due to the presence of dedicated As (III) detoxification systems in yeast [77]. Alternatively, since As(III) exposure inhibits translation at the initiation step [39,60], the frequency of ribosome stalling and collision events might be low, and hence, Hel2mediated RQC might not be required. Our data implicate the cytosolic Ub ligase Rsp5 in proteostasis during As(III) stress. Reduced Rsp5 activity (using the rsp5-1 allele) affected aggregate clearance and cell growth during As(III) stress, and these defects were not rescued by catalytically inactive Rsp5-C777A (Fig. 4). While the Ub ligase activity of Rsp5 was important for clearance, it was dispensable for ubiquitination of aggregated proteins since neither K48-linked nor K63-linked ubiquitination of As(III)-aggregated proteins decreased in rsp5-1 cells. This contrasts with the role of Rsp5 in targeting cytosolic proteins for K48-linked ubiquitination during heat stress [29]. We previously showed that As(III) can affect the structure of an aggregated model protein, thereby limiting chaperone binding to the aggregate [42]. It is tempting to speculate that also Rsp5 has limited access to As(III)-aggregated substrates. Rsp5 did not regulate intracellular arsenic levels, indicating that it affects proteostasis independently of hexose transporter turnover [66]. The mechanism by which Rsp5 contributes to proteostasis during As(III) stress remains unknown. Likewise, the Ub ligases that mediate K48-and K63linked ubiquitination of the As(III)-aggregated proteins remain to be identified. Fig. 4. Rsp5 contributes to aggregate clearance during As(III) stress. (A) Sis1-GFP distribution was scored by fluorescence microscopy in cells before and after exposure to 0.5 mM As(III). The fractions of cells containing aggregates/Sis1-GFP foci were determined by visual inspection of 223-434 (left panel) or 197-377 (right panel) cells per condition and time point. Data are expressed as mean AE SD from three independent biological replicates. * indicates a significant difference (P < 0.05) compared with wild-type (unpaired, two-tailed Student's ttest). (B) 10-fold serial dilutions of the indicated strains were placed onto agar plates with or without As(III). Growth was recorded after 3 days at 30°C. Growth assays were performed with three biological replicates and a representative image is shown. (C) Sis1-GFP distribution was scored as in A in rsp5-1 cells transformed with an empty plasmids or with plasmids carrying RSP5 or RSP5-C777A before and after exposure to 0.5 mM As(III). The fractions of cells containing aggregates/Sis1-GFP foci were determined by visual inspection of 143-364 cells per condition and time point. Data are expressed as mean AE SD from three independent biological replicates. * indicates a significant difference (P < 0.05) compared with wild-type (unpaired, two-tailed Student's t-test). (D-F) Protein aggregation (D), K48-linked ubiquitination (E), and K63-linked ubiquitination (F). Wild-type (WT) and rsp5-1 (D) were exposed to 0.5 mM As(III) and the proteins in the total lysate and aggregate fractions were isolated at the indicated time points, separated on SDS/PAGE and visualized as described in Materials and methods. Immunoblotting was performed using an antibody recognizing K48-linked (E) or K63-linked (F) Ub chains. Shown is a representative gel and an immunoblot (upper panels) from two independent biological replicates. For quantification, images were analyzed using IMAGEJ. The signals given by the stain-free gels and the antibody from a blot were first normalized to the total signal per sample and then to the total signal over all samples. Data shown (lower panels) represent the average of two independent biological replicates with SD. * indicates a significant difference (P < 0.05) compared with wild-type (unpaired, two-tailed Student's t-test).

Data accessibility
All data described are available in the article and its Supporting Information.

Supporting information
Additional supporting information may be found online in the Supporting Information section at the end of the article. Fig. S1. Intracellular arsenic. Table S1. Yeast strains used in this study. Table S2. Plasmids used in this study.