Genetically induced redox stress occurs in a yeast model for Roberts syndrome

Abstract Roberts syndrome (RBS) is a multispectrum developmental disorder characterized by severe limb, craniofacial, and organ abnormalities and often intellectual disabilities. The genetic basis of RBS is rooted in loss-of-function mutations in the essential N-acetyltransferase ESCO2 which is conserved from yeast (Eco1/Ctf7) to humans. ESCO2/Eco1 regulate many cellular processes that impact chromatin structure, chromosome transmission, gene expression, and repair of the genome. The etiology of RBS remains contentious with current models that include transcriptional dysregulation or mitotic failure. Here, we report evidence that supports an emerging model rooted in defective DNA damage responses. First, the results reveal that redox stress is elevated in both eco1 and cohesion factor Saccharomyces cerevisiae mutant cells. Second, we provide evidence that Eco1 and cohesion factors are required for the repair of oxidative DNA damage such that ECO1 and cohesin gene mutations result in reduced cell viability and hyperactivation of DNA damage checkpoints that occur in response to oxidative stress. Moreover, we show that mutation of ECO1 is solely sufficient to induce endogenous redox stress and sensitizes mutant cells to exogenous genotoxic challenges. Remarkably, antioxidant treatment desensitizes eco1 mutant cells to a range of DNA damaging agents, raising the possibility that modulating the cellular redox state may represent an important avenue of treatment for RBS and tumors that bear ESCO2 mutations.


Introduction
Roberts syndrome (RBS) (MOM #268300, MIM #269000) is a severe autosomal recessive disorder characterized by phocomelia (flipper-like appendages), microcephaly, cleft palate, syndactyly, intellectual disabilities, seizures, abnormalities in the heart, urinary, genital, and alimentary tract, spontaneous abortion, stillbirth, and possible early mortality (Van Den Berg and Francke 1993;Vega et al. 2005). To date, no cure exists for RBS and treatment is limited to modalities that partially improve quality of life for affected individuals (Van Den Berg and Francke 1993; Gordillo et al. 1993).
ESCO2, and cohesin, function in a variety of cellular processes that include gene transcription, chromosome condensation, sister chromatid cohesion (SCC), and damage-induced cohesion (DIC). The function of the ESCO2/Eco1 family of acetyltransferases crosses all portions of the cell cycle: functioning during G1 to regulate transcription, S phase to promote SCC and chromosome condensation, and G2/M phases in which DIC promotes sister chromatid interactions during homologous recombination (HR) that occurs in response to DNA damage (Guacci et al. 1997;Michaelis et al. 1997;Losada et al. 1998;Skibbens et al. 1999;Tó th et al. 1999;Losada and Hirano 2001;Ivanov et al. 2002;Kim et al. 2002;Strö m et al. 2004Strö m et al. , 2007Ü nal et al. 2004Heidinger-Pauli et al. 2010;Mö nnich et al. 2011;Bose et al. 2012;Lu et al. 2014;Banerji et al. 2016Banerji et al. , 2017). An early model of RBS focused on mitotic failure based on the initial characterization of Ctf7/Eco1 in yeast as a cohesion factor (Skibbens et al. 1999;Tó th et al. 1999) and findings in vertebrate cells that loss of ESCO2 function coincided with mitotic failure (stemming from SCC defects) and apoptosis, which likely precedes loss of progenitor stem cell populations (Mö nnich et al. 2011;Morita et al. 2012;Whelan et al. 2012;Percival et al. 2015). However, a large body of work supports the model that RBS arises instead from transcriptional dysregulation. For instance, knockdown of ESCO2 results in zebrafish fin regeneration defects that occur independent of increased cell apoptosis (Banerji et al. 2016). Moreover, target genes downstream of ESCO2 (and cohesin) indeed function in bone growth and their exogenous expression suppresses phenotypes that otherwise arise from ESCO2 or cohesin (SMC3) reduction (Banerji et al. 2016(Banerji et al. , 2017. Early studies further linked cohesin reduction to rDNA defects (Skibbens et al. 2010), the first study to conceptually link cohesinopathies [RBS and Cornelia de Lange syndrome (CdLS)] to ribosomopathies (Treacher-Collins syndrome, Blackfan Anemi, Schwann-Diamond syndrome, etc.). Elegant studies document that rDNA transcription defects result in reduced ribosome biogenesis and translation (Bose et al. 2012;Xu et al. 2013;Lu et al. 2014). Importantly, many of the developmental defects that arise in zebrafish esco2 hypomorphs can be reduced simply through modulating translation through mTOR pathways (Xu et al. 2013(Xu et al. , 2016. These findings, and others obtained from zebrafish, tissue culture cells, and tumor models (Leem et al. 2011;Mö nnich et al. 2011;Xu et al. 2016;Guo et al. 2018) support a transcriptional dysregulation model of RBS. In this light, it is not surprising that CdLS, which arises through cohesin and cohesin regulator mutations, also results from transcription dysregulation and can occur in the absence of mitotic failure (Rollins et al. 1999;Ren et al. 2008;Kawauchi et al. 2009;Rhodes et al. 2010;Dorsett 2010Dorsett , 2011Mö nnich et al. 2011;Deardorff et al. 2012;Dorsett and Merkenschlager 2013;Dowen et al. 2013;Nolen et al. 2013;Remeseiro et al. 2013;Schaaf et al. 2013;Skibbens et al. 2013;Zuin et al. 2014;Mannini et al. 2015;Yuan et al. 2015;Banerji et al. 2016Banerji et al. , 2017Boyle et al. 2017;Boudaoud et al. 2017;Luna-Pelá ez et al. 2019;Gudmundsson et al. 2019;De Koninck et al. 2020;Dorval et al. 2020). The extent to which deficits in other ESCO2 functions, beyond transcription regulation, contribute to RBS birth defects remains unknown.
ESCO2 and cohesin are both critical for DNA damage repair. While there are examples that mutation of DNA damage defense genes leads to severe birth defects (e.g., Ataxia Telangiectasia, Bloom syndrome, Fanconi Anemia, Cockayne syndrome, etc.), a DNA damage model for RBS has been minimally explored. A DNA damage model of RBS is supported by findings that RBS cell lines exhibit increased sensitivity to a battery of genotoxins that include the alkylating agent mitomycin C, ionizing radiation, the topoisomerase I inhibitor camptothecin, and the topoisomerase II inhibitor etoposide (Gentner et al. 1985(Gentner et al. , 1986Burns and Tomkins 1989;Van Den Berg and Francke 1993;Gordillo et al. 2008;van der Lelij et al. 2009;McKay et al. 2019). Moreover, altered ESCO2 and cohesin activity are tightly correlated with tumorigenesis (Zou et al. 1999;Oikawa et al. 2004Oikawa et al. , 2008Ryu et al. 2007;Shepard et al. 2007;Barber et al. 2008;Solomon et al. 2011;Xu et al. 2011;Thol et al. 2014;Bailey et al. 2014;Xu et al. 2014;Chen et al. 2018;Guo et al. 2018;Wang and Liu 2020). Therefore, it becomes important to test whether ESCO2 mutations, by themselves, could induce DNA damage and/or upregulate ROS pathways and thus contribute to RBS phenotypes. Here, we test the extent to which RBS can be considered, in part, a disease caused by a defective DNA damage response (DDR). We use a budding yeast model to focus in part on oxidative DNA damage induced by reactive oxygen species (ROS) produced by hydrogen peroxide (H 2 O 2 ) and paraquot.
In addition to prior studies that document ROS sensitivities in cells deficient in cohesin function (Ren 2005;Ren et al. 2008;Cukrov et al. 2018), this work on Eco1 provides novel insights into the role that cohesion pathways play in the defense against redox stress and highlights a novel mechanism of dysregulated endogenous ROS that contributes to reduced cell vitality in the absence of Eco1.

Cell culture and cell cycle synchronization methods
In all experiments log phase cultures were used in an actively growing state by either growth overnight in YPD at 23 C with subculturing on day 2 followed by growth at 23 C for an additional 6 h. Or, cells on day 2 of growth were subcultured in YPD for overnight growth at 23 C to be used for experiments on day 3. Hydroxyurea (HU) (Sigma) was prepared at a stock concentration of 2 M in sterile water. Zeocin (Invitrogen) was added from a stock concentration at 100 mg/ml as prepared by the supplier. H 2 O 2 (Sigma) was added from a stock concentration of either 9.8 M when preparing plates or from a 50-mM stock in sterile PBS when treating liquid cultures. Paraquat (PQ) (Sigma) was prepared at a stock concentration of 100 mM in sterile water. N-acetylcysteine (Sigma) was prepared at a stock concentration of 1 M in sterile 4% DMSO.

Drug-sensitivity assays
For drop-tests, cultures were normalized to OD 600 ¼ 0.1 and diluted 10-fold six times in YPD. Four microliters of all seven dilutions were plated on YPD or YPD þ drug and grown at the indicated temperatures for 2-4 days in the dark. Plates containing drug were prepared by adding the indicated drugs directly to the cooled agar before pouring. Plates were incubated in dark conditions to dry and used for experiments 1 day after pouring to preserve drug activity in the media.
For survival assays, cultures were normalized to OD 600 ¼ 1.0 in PBS. Cells were then either left untreated or treated with H 2 O 2 for 1 h at 30 C with continuous mixing. The cells were then washed two times with PBS and serially diluted 10-fold five times. Two hundred to 250 ml of the lowest dilution were plated on YPD and survival was determined by counting colony-forming units after 2-4 days of growth at 23 C. For experiments involving N-acetylcysteine treatment, N-acetylcysteine was added to a final concentration of 25 mM to cells suspended in PBS and incubated with continuous mixing for 2 h at 23 C. Then, cells were washed in PBS two times before H 2 O 2 treatments as described above.

Western blot analysis
Whole cell extracts were made by TCA precipitation as previously described (Tong and Skibbens 2014). Rad53 and phosphoglycerate kinase (PGK) detection was carried out as previously described (Mfarej and Skibbens 2020).

Quantitative growth rate and cell cycle progression analyses
For growth rate analyses, cultures were normalized to OD 600 ¼ 1.0 in PBS. Cells went either untreated or treated with H 2 O 2 for 1 h at 30 C with continuous mixing. Cells were then washed two times with PBS, resuspended in YPD and diluted to OD 600 ¼ 0.1. Cells were incubated with continuous mixing at 23 C for 6 h with OD 600 readings being taken at the indicated time points.

ROS labeling analyses
Cells were normalized to OD 600 ¼ 1.0 in PBS and then went either untreated or treated with 1 mM H 2 O 2 diluted in PBS for 1 h with continuous mixing at 30 C. Cells were then washed two times with PBS. 2 0 ,7 0 -Dichlorofluorescein diacetate (H 2 DCFDA; ThermoFisher) was added to cells from a 1 mg/ml stock in DMSO to a final concentration of 5-10 mg/ml and incubated for 0.5-2 h at 23 C with continuous mixing. Dihydroethidium (DHE; ThermoFisher) was added to cells from 1 mg/ml stock in DMSO to a final concentration of 5 mg/ml and incubated for 2 h at 23 C with continuous mixing. Following dye incubations, cells were washed two times with PBS then used for analyses through either fluorescence microscopy or flow cytometry.
For fluorescence microscopy analysis, cells were seeded on poly-L-lysine-coated microscope slides prior to imaging [Nikon Eclipse E800, Hamatsu microscope equipped with a cooled CD (Coolsnapfx, Photometrics) and IPLab software (Scanolytics)]. All images within one biological replicate were taken using the same exposure time and a cell-sized ROI. Quantifications were performed with ImageJ by measuring the integrated density across the total surface area of each cell that was located fully in the field of view. For images where one measurement per cell could not be accurately taken due to high number of cells in the field of view, at least 100 cells were measured. For each image, 10-100 background fluorescence measurements were taken using the same cell-sized ROI and the average background intensity subtracted from the average cell fluorescence intensities to yield the normalized average cell fluorescence intensity for each image. All of the image quantifications for each sample were then averaged to obtain the average normalized fluorescence intensity for that population under the indicated treatment conditions. Fold changes in fluorescence were determined by dividing treated/ mutant samples by the WT untreated sample.
For flow cytometry analyses, no dye controls consisted of an aliquot of cells from untreated samples that went for 2 h during the H 2 DCFDA incubation in PBS without labeling dye. The samples were then washed as previously described. Background levels in no dye controls were measured using the FITC channel on a BD Canto II flow cytometer with a threshold of 300, a low flow rate and counting 10,000 events. The voltage level was chosen in order to restrict the fluorescence peak to 10 2 on the x-axis. All subsequent measurements in dye-treated samples were assayed using the same settings established for the no-dye controls.

Results
Eco1 and function are critical during redox stress Eco1 and cohesin function are required for DIC in response to DSBs (Strö m et al. 2004Ü nal et al. 2004. The ironresponsive transcription factor Yap5, which protects against metal-catalyzed oxidation reactions, regulates ECO1 expression in response to iron stress and DNA damage (Pimentel et al. 2012;Mfarej and Skibbens 2020). These findings suggest that Eco1dependent cohesion establishment also may be important during oxidative stress. Note that eco1 W216G is homologous to the mutation in ESCO2 that results in RBS in humans (Gordillo et al. 2008) while RAD21 (MCD1 in yeast) mutation gives rise to CdLS (Deardorff 2012;Minor 2014). Both eco1 W216G and mcd1-1 alleles produce cell temperature-sensitive growth and exhibit defects in the DDR (Ü nal et al. 2007;Lu et al. 2010;McAleenan et al. 2012). We first confirmed the temperature sensitivity of eco1 W216G and mcd1-1 mutant strains and then further identified an intermediate temperature (33 C for eco1 W216G, 31 C for mcd1-1) in which mutant cell growth appears sensitized ( Figure 1A). All remaining experiments were performed at these experientially derived allele-sensitizing temperatures. Next, eco1 W216G and mcd1-1 mutant cells, each with a matched wildtype control strain, were exposed to rich media plates or rich medium that contained either zeocin (a DSB inducer) or HU (a replication fork staller). In the absence of exogenous genotoxic agents, wildtype cells exhibit robust growth at 33 C with the eco1 W216G mutant strain exhibiting only moderately reduced growth at 33 C ( Figure 1, A and B). eco1 W216G mutant cells, however, exhibited severely reduced growth on plates supplemented with zeocin and were near-inviable on plates supplemented with HU ( Figure 1B). These findings greatly extend prior reports that the eco1 W216G allele renders cells sensitive to bleomycin and X-ray irradiation (Lu et al. 2010) in that HU sensitivity is unique from these genotoxic agents which induce DSBs. Additionally, while mcd1-1 cells exhibit reduced growth at 31 C, mcd1-1 cell growth is almost completely abolished when grown on plates supplemented with zeocin or HU ( Figure 1B). Similar sensitivity to HU was observed in other ts cohesion mutant strains bearing the smc3-42 mutant allele (Supplementary Figure S1A) and the scc2-4 mutant allele (Supplementary Figure S1B).
ROS are a source of free radical electrons that cause oxidative damage to macromolecules including lipids, proteins, and DNA (reviewed in Cooke et al. 2003). Although oxidative DNA damage was once understood to be repaired mainly through baseexcision repair (BER) and nucleotide-excision repair (NER) pathways (Gellon et al. 2001;reviewed in Melis et al. 2013), studies in both budding yeast and mammalian cell lines highlight the importance of the HR pathway in promoting genome integrity in response to ROS (Otterlei et al. 2000;Winn et al. 2003;Salmon et al. 2004;Yi et al. 2016;Choi et al. 2018;Novarina et al. 2020). Despite the importance of Eco1 in HR (Strö m et al. 2007;Ü nal et al. 2007;Lu et al. 2010;Lyons et al. 2013;Mfarej and Skibbens 2020a), few reports exist regarding the role of Eco1 function in ROS response pathways (Xu et al. 2013(Xu et al. , 2016Cukrov et al. 2018). To further test if Eco1 and cohesins co-functions are important during redox stress, we tested eco1 W216G, mcd1-1, smc3-42, and scc2-4 mutant sensitivity to H 2 O 2 , which induces oxidative stress. The results reveal that eco1 W216G, mcd1-1, smc3-42, and scc2-4 mutant cells are highly sensitive to H 2 O 2 compared with wildtype cells ( Figure 1C; Supplementary Figures S1, A and B, and S2), providing evidence for the importance of cohesion pathways in the defense against oxidative stress. In parallel, we tested eco1 W216G and mcd1-1 mutant strain sensitivity to the superoxide anion inducer PQ which generates reductive stress. Interestingly, while eco1 W216G mutant cells exhibit severely compromised growth in the presence of PQ ( Figure 1C), mcd1-1 mutant cells are only mildly sensitive to PQ ( Figure 1C). These results suggest that Eco1 function may be required for a broader range of redox stresses compared with cohesin.

Oxidative stress hyperactivates DNA damage checkpoints in the absence of Eco1
If oxidative stress induces DNA damage, then the slowed growth rate exhibited by eco1 mutant cells might be due to hyperactivation of DNA damage checkpoints. To test this prediction, log-phase WT and eco1 W216G cells were exposed to 0.25 mM H 2 O 2 or 0.5 mM H 2 O 2 for 1 h at 30 C, washed and resuspended in drug-free media. Cell density (OD 600 ) was then measured over a 6-h time course. The results show that eco1 W216G cells exhibit approximately a 35% reduction in growth rate when exposed to medium containing 0.5 mM H 2 O 2 , relative to the growth rate of ECO1 cells (Figure 2, A and B). We reasoned that the reduced growth rate of eco1 W216G cells, in the presence of 0.5 mM H 2 O 2 , may result from hyperactivation of a DNA damage checkpoint. To test this possibility, we assessed the phosphorylation state of Rad53, which becomes phosphorylated in response to DNA damage (Allen et al. 1994;Sun et al. 1998). In the absence of H 2 O 2 , Rad53 remained unphosphorylated in both wildtype and eco1 W216G mutant cells. As expected, wildtype cells exposed to 0.5 mM H 2 O 2 contained phosphorylated Rad53, but this level of H 2 O 2 was only sufficient to induce roughly 50% of Rad53 phosphorylation. In contrast, all of Rad53 appeared phosphorylated in eco1 W216G mutant cells in response to 0.5 mM H 2 O 2 ( Figure 2C).

ECO1 mutation elevates endogenous ROS levels
DNA damage upregulates endogenous ROS pathways which in turn are capable of further damaging DNA and other cellular macromolecules Rowe et al. 2008;Kang et al. 2012;Rowe et al. 2012;Marullo et al. 2013;Yi et al. 2016;Choi et al. 2018). Given eco1 W216G mutant cell hypersensitivity to ROS and genotoxic agents, we wondered whether cells with reduced Eco1 function might exhibit increased ROS-even in the absence of exogenous stresses. To test this hypothesis, we quantified ROS levels in eco1 W216G and mcd1-1 mutant cells. Log phase eco1 W216G and mcd1-1 mutant strains, and matched wildtype strains, were treated with 1 mM H 2 O 2 for 1 h at 30 C, washed and then incubated in the presence of H 2 DCFDA. H 2 DCDFA is a cell permeable nonfluorescent reagent, that upon oxidation by ROS and reactive nitrogen species, fluoresces and becomes trapped in the cell (reviewed in Kalyanaraman et al. 2012). Importantly, eco1 W216G mutant cells produced an H 2 DCFDA signal over two times greater than matched wildtype cells, in the absence of H 2 O 2 -induced stress when ROS levels were analyzed by fluorescence microscopy (Figure 3, A and B; Supplementary Figure S3A). Similarly, increased H 2 DCFDA fluorescence levels in eco1 W216G cells were also detected when measured by flow cytometry (Supplementary Figure S4, A-C). mcd1-1 mutant cells exhibited over a 2.5-fold increase in H 2 DCFDA signal, compared with matched control wildtype cells, in the absence of H 2 O 2 -induced stress although this increase was below the threshold of statistical significance (Figure 3, A   Eco1-and cohesin-dependent DNA repair pathways promote cell growth in response to redox stress; (A-C) 10-fold serial dilutions of ECO1, eco1 W216G, MCD1, and mcd1-1 mutant strains at the indicated conditions. Cells seeded on drug-treated plates were tested at a single semipermissive temperature as indicated. Representative results shown from a total of N 2 biological replicates. Plates were imaged after two days of growth at the indicated temperatures.
ROS in the absence of exogenous stress, it became important to test the ability to increase ROS production upon H 2 O 2 exposure.
Neither eco1 W216G or mcd1-1 mutant cells, however, increased ROS levels in response to H 2 O 2 treatment beyond that induced by the mutation alone (Supplementary Figures S4, A-C  and S5, A-C). This is likely due to ROS levels that remain higher in eco1 W216G and mcd1-1, with or without H 2 O 2 , than respective wildtype strains exposed to H 2 O 2 (   reacts with superoxide anions to produce a fluorescent and cellimpermeable marker (reviewed in Kalyanaraman et al. 2012). Results obtained from DHE in untreated cells were overall consistent with those obtained using H 2 DCFDA. For instance, eco1-W216G exhibited increased superoxide anion levels, relative to matched controls, in the absence of H 2 O 2 -induced stress (Figure 3, C and D; Supplementary Figure S4B). mcd1-1 mutant cells also exhibited an increase DHE intensity, but below the threshold of statistical significance ( Figure  3, C and D; Supplementary Figure S4B). DHE staining, in response to H 2 O 2 treatment, also revealed notable differences from H 2 DCFDA results. For instance, the DHE signal was not significantly different between ECO1 and eco1 W216G in the presence of H 2 O 2 -induced stress (Supplementary Figure S5, D-F). eco1 mutant cells in fact appear to have slightly diminished fluorescent intensity, although this difference was not statistically significant from ECO1 wildtype samples (Supplementary Figure S5, D-F). Moreover, DHE signal in the MCD1 or mcd1-1 strains did not change in response to H 2 O 2 stress although DHE fluorescence levels remained higher in the mcd1-1 relative to wildtype controls in response to H 2 O 2 (Supplementary Figure S5, D-F).

ROS contribute to genotoxicity in the absence of Eco1
Our findings that mutation of either ECO1 or MCD1 is solely sufficient to increase endogenous ROS levels, in the absence of exogenous oxidative stress, suggests strategies through which mutant cell phenotypes could be ameliorated. To test this, we used the ROS-scavenging antioxidant N-acetyl cysteine (NAC) (Aruoma et al. 1989;Carter et al. 2005;Reliene et al. 2009;Tang et al. 2015).
Log-phase ECO1, eco1 W216G, MCD1, and mcd1-1 mutant cell cultures were split with half untreated and the other half pretreated with 25 mM NAC for 2 h at 23 C. Cells were then washed, resuspended in PBS and either remained untreated or exposed to 1 mM H 2 O 2 at 30 C for an additional 1 h ( Figure 4A). The resulting cultures were then plated onto fresh medium, maintained at 23 C and viability assessed 48-96 h later. All viability data are normalized to the untreated cultures.
We reasoned that, if NAC neutralizes exogenous H 2 O 2 that generates oxidative stress, then cells pretreated with NAC should reduce the genetically induced ROS load and promote increased cell viability after subsequent oxidative stress. Following a 1-h treatment with 1 mM H 2 O 2 , ECO1, eco1 W216G, MCD1, and mcd1-1 cells exhibited reductions (roughly 40%, 75%, 30%, and 65%) in cell viability ( Figure 4A). When pretreated with NAC, however, both eco1 W216G and mcd1-1 cells exhibited a significant increase in cell viability ( Figure 4A). While matched wildtype controls appeared to benefit from NAC pretreatment, the results were not statistically significantly which highlights the effects of reduced cohesin function in promoting ROS production ( Figure 4A).
The increase in mutant cell viability, when pretreated with NAC and subsequently followed by a short exposure to H 2 O 2 , suggested that increased endogenous ROS levels might play a key role in eco1 W216G and mcd1-1 decreased resistance to DNA damage. To formally test whether reducing ROS levels through NAC would protect cells against exogenous DNA damage, we supplemented rich medium with HU or zeocin as well as NAC. Note that these genotoxic agents produce DNA damage defects that do not directly rely on ROS. As before, the results show that both eco1 Figure 4 Rescue of Eco1 and cohesin mutation-associated defects with antioxidant treatment. (A) Quantification of ECO1, eco1W216G, MCD1, and mcd1-1 strain viability at 2-4 days following exposures to either no drug or 25 mM NAC at 23 C for 2 h followed by subsequent incubations at 30 C for 1 h either with or without 1 mM H 2 O 2 . Percent viability was determined by averaging the ratios of viability (drug-treated colonies/untreated colonies) across at least N 4 biological replicates. Error bars indicate standard deviation. Statistical analysis was performed using one-tailed, Student's T-test (a 0.05); (B and C) 10-fold serial dilutions of ECO1, eco1-W216G, MCD1, and mcd1-1. Each yeast strain was grown for 2 days at the indicated temperatures on YPD or YPD supplemented with either 25 mM NAC, 10 mg/ml zeocin, 100 mM HU, NAC þ zeocin, or NAC þ HU.
W216G and mcd1-1 cells exhibit hypersensitivity to HU and zeocin (Figure 4, B and C). When medium is supplemented with NAC, in addition to either zeocin or HU, eco1 W216G mutant cells exhibited reduced genotoxic sensitivity and a partial rescue of growth defects ( Figure 4B). NAC supplementation similarly provided a growth benefit to mcd1-1 mutant cells simultaneously exposed to zeocin. This protection, however, did not extend to mcd1-1 mutant cells concurrently exposed to NAC and HU ( Figure 4C).

Discussion
Oxidative stress is implicated in aging and numerous human maladies that include neurodegeneration, heart disease, and cancer (reviews: Finkel and Holbrook 2000;Cooke et al. 2003;Barnham et al. 2004;Valko et al. 2006). Oxidative reactive species damage proteins, lipids, and produce mutagenic DNA adducts such as modified bases, apurinic/apyrimidinic sites, single strand breaks (SSBs), and DSBs (reviewed in Kryston et al. 2011). NER repair and BER are considered primary defenses against oxidative DNA damage. The first major revelation of the current study is that Eco1 and cohesin function as physiologically relevant defenses against oxidative stress. Our studies extend prior work that demonstrated functional overlap between HR and oxidative stress regulators (Yi et al. 2016;Choi et al. 2018;Novarina et al. 2020). The current study thus provides strong support for a model that Eco1/cohesin are important regulators of HR during oxidative stress (Xu et al. 2013(Xu et al. , 2016Cukrov et al. 2018). Given the variety of mutagenic modifications induced by oxidative DNA damage, one limitation of this study is that the specific types of oxidative DNA damage repair that Eco1 and cohesin promote remain unknown. It is known, however, that DIC plays a critical role during DSB repair. Future studies may provide important insights as to the physiological contributions that Eco1/cohesin provide to repair other oxidative lesions (e.g., base modifications, SSBs, etc.). While DIC is likely one pathway through which Eco1/ cohesin resolve oxidative stress, we further note that Eco1/cohesin are critical regulators of gene transcription and that this parallel pathway may be, in parallel, important in oxidative stress responses.
What is the mechanism of Eco1/cohesin sensitivity to DNA damage? DIC is required for bringing sister chromatids into close physical proximity to promote strand invasion reactions during high fidelity HR (Birkenbihl and Subramani 1992;Kadyk and Hartwell 1992;Sjö gren and Nasmyth 2001;Strö m et al. 2004Strö m et al. , 2007Ü nal et al. 2004. DNA damage, however, also induces ROS upregulation, providing a feedback mechanism that activates repair processes. The second major revelation of the current work is that Eco1/cohesin regulates endogenous ROS levels during DNA damage. ROS are upregulated in the cell in response to a variety of stresses, including DNA damage, which functions to post-translationally modify DDR regulators like Yap1 in budding yeast and ATM in mammals through cysteine side chain oxidation (Moye-Rowley et al. 1989;Coleman et al. 1999;Delaunay et al. 2000;Delaunay et al. 2002;reviewed in Rodrigues-Pousada et al. 2004;Guo et al. 2010). In cells with functional DIC, endogenous sources of DNA damage are repaired through NER, BER, and HR pathways-each of which keeps ROS levels in check ( Figure 5A). In the absence of DIC via ECO1 or cohesin mutation, ROS are upregulated and background levels of oxidative stress produce increased genotoxic stress ( Figure 5B). In combination with other studies that implicate deficiencies in Esco2 or cohesins in driving elevated ROS levels (Xu et al. 2013(Xu et al. , 2016Cukrov et al. 2018), our findings raise important implications regarding the etiology of RBS. For example, elevated ROS levels present in cells deficient in Esco2 and cohesin functions are likely to incur additional levels of macromolecular damage to DNA, proteins, and lipids. This model suggests that RBS cytotoxicity may arise through several synergistic ROS-dependent mechanisms that include oxidation of DNA and proteins that may contribute to defects in transcription, translation, and SCC ( Figure 5B) (reviewed in Mfarej and Skibbens 2020b).
The molecular description of Eco1 function is likely still in its infancy. For instance, Eco1 not only acetylates Smc3 (SCC establishment), but also Mcd1 (DIC), Mps3 (nuclear organization), and the inner surface of the PCNA sliding clamp that regulates repair polymerase processivity during HR (Antoniacci et al. 2004;Heidinger-Pauli et al. 2009;Billon et al. 2017). The final revelation of the current study is that Eco1 and cohesin exhibit different redox stress characteristics. For example, both eco1 and mcd1 mutants are sensitive to H 2 O 2 -induced stress. Between the two, however, strains lacking full Eco1 function appear more sensitive to the superoxide anion generator PQ. This suggests that Eco1-dependent pathways function in response to a wider range of redox stresses than cohesin-dependent pathways and may reflect a broader role for Eco1 in the maintenance of genomic integrity. Similarly, while both eco1 and mcd1 mutant strains exhibit increased endogenous levels of oxidative ROS, eco1 mutant cells exhibit higher levels of superoxide anion in the absence of challenges whereas the mcd1 mutant exhibits higher superoxide anion levels in response to H 2 O 2 -induced stress. The effects of superoxide anion levels in eco1 and mcd1 mutants is unclear given data that superoxide anion serves as a cell death inducer in response to endogenous DNA damage and UV-C treatment (Rowe et al. 2008(Rowe et al. , 2012Marullo et al. 2013). In contrast, other studies suggest that superoxide anions confer protective effects against H 2 O 2 -induced stress (Thorpe et al. 2013). The higher basal levels of superoxide anion in eco1 mutant cells may explain PQ hypersensitivity of eco1 mutant cells, compared with mcd1 mutant cells.
A failure to modulate the cellular redox state in cells with reduced ESCO2/Eco1 function may expose an Achilles heel in both RBS and cancer. For example, RBS (a.k.a. pseudothalidomide syndrome) is phenocopied by the birth defects caused by exposure to the teratogen thalidomide (Herrmann and Opitz 1977;Waldenmaier et al. 1978;Sherer et al. 1991;Holden et al. 1992). ROS-neutralizing chemicals rescue teratogenic birth defects in the offspring of thalidomide-treated pregnant rabbits by reducing oxidative DNA damage (Parman et al. 1999;Wani et al. 2017). Moreover, genetic modulation of superoxide defense systems in Drosophila oocytes harboring cohesin mutations influences the rate of chromosome mis-segregation errors (Perkins et al. 2016(Perkins et al. , 2019. These results raise the possibility that increasing antioxidant levels may represent a treatment option to ameliorate the severity of birth defects otherwise present in RBS individuals. Separately, synthetic lethal approaches, through inhibition of either poly-ADP-ribose polymerase (PARP) or Wnt signaling, are effective treatments for tumors that exhibit marked changes in ESCO2-dependent pathways (McLellan et al. 2012;reviewed in O'Neil et al. 2013;Liu et al. 2018;Mondal et al. 2019;Waldman 2020;Chin et al. 2020). This raises the possibility that oxidative stress pathways may serve as an additional synthetic lethality target in treating tumors with mutated ESCO2 or cohesin.

Data availability
Strains are available upon request. The authors affirm that all data necessary for confirming the conclusions of the article are present within the article. A strain table (Supplemental Table 1), figures, and tables with raw and complete images are available at figshare (https://figshare.com/authors/Robert_Skibbens/8138124).

Acknowledgments
We thank the Skibbens lab members for their helpful discussions throughout this process. We also thank Paul Doetsch for his insights regarding the experimental design of our ROS labeling experiments.

Funding
This work was supported by an award to R.V.S. from the National Institutes of Health [R15GM110631] and a Nemes research fellowship award to M.G.M. The funders had no role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Conflicts of interest
The authors declare that there is no conflict of interest.