Multivalent binding of the tardigrade Dsup protein to chromatin promotes yeast survival and longevity upon exposure to oxidative damage

Tardigrades are remarkable in their ability to survive extreme environments. The damage suppressor (Dsup) protein is thought responsible for their extreme resistance to reactive oxygen species (ROS) generated by irradiation. Here we show that expression of Ramazzottius varieornatus Dsup in Saccharomyces cerevisiae reduces oxidative DNA damage and extends the lifespan of budding yeast exposed to chronic oxidative genotoxicity. This protection from ROS requires either the Dsup HMGN-like domain or sequences C-terminal to same. Dsup associates with no apparent bias across the yeast genome, using multiple modes of nucleosome binding; the HMGN-like region interacts with both the H2A/H2B acidic patch and H3/H4 histone tails, while the C-terminal region binds DNA. These findings give precedent for engineering an organism by physically shielding its genome to promote survival and longevity in the face of oxidative damage.


INTRODUCTION
Tardigrades (also termed water bears) are an invertebrate phylum of > 1,200 species with broad-reaching habitats. Many can survive desiccation, extreme temperatures, high pressure, severe irradiation, and exposure to space 1 . The mechanisms by which tardigrade species resist such extreme stressors are poorly understood. Ramazzottius varieornatus is highly resistant to ionizing radiation (IR); capable of surviving > 48 hours after a dose of 4000 Gy 2 , compared to the human LD50 of ~ 4.5 Gy 3 . The R. varieornatus Dsup (Damage suppressor) protein is chromatin associated and predicted to promote IR resistance, being absent from IR sensitive tardigrade species 4 . Indeed, when expressed in human cells, Dsup localizes to nuclear DNA and confers IRresistance accompanied by reduced levels of single-and double-strand DNA breaks (SSBs and DSBs) 4 ; it also confers protection from radiation when expressed in tobacco plants 5 .
While IR can directly induce SSBs and DSBs, much of its genotoxicity is mediated by hydroxyl radicals (OH • ), the most powerful oxidant among the reactive oxygen species (ROS) and generated when radiation interacts with water molecules 6 . Consistent with Dsup protecting against hydroxyl radicals, it also reduces the number of DNA breaks in human cells exposed to hydrogen peroxide (H2O2) 4,7 . The high-energy hydroxyl radicals react with DNA bases to form lesions (including 8-oxoguanine; 8-oxo-G), while oxidation of the deoxyribose backbone dissociates sugar-phosphate bonds leading to DNA breaks 8 . Throughout life, oxidative DNA damage is generated from aerobic metabolism, with the resulting mutations thought to contribute to the ageing process 9 and the development of age-related diseases 10 , such as neurodegeneration 11 and cancer 12,13 . Most cancer treatments cause oxidative DNA damage and strand breaks, and thus contributes to long-term side effects in survivors 14 . As such, the means by which proteins such as R. varieornatus Dsup protect the genome from oxidative damage are of extreme interest.

R. varieornatus
Dsup is a 445 amino acid protein predicted to be intrinsically disordered 15 .
Of note, disorder at the N-and C-termini is an important feature of proteins that scan and engage DNA, consistent with a DNA-binding role for Dsup 16,17 . C-terminal deletion (D aa 208-445) abrogates Dsup binding to naked DNA or human chromatin 4 . Indeed, Dsup binds with higher affinity to reconstituted chromatin over free DNA, and sequences within aa 360-445 are required for the association with chromatin and protection from DSBs caused by hydroxyl radicals 18 . While Dsup induction in human cells upregulates the expression of DNA repair genes 7 ; the protein also physically prevents DNA damage via chromatin binding, as this ability is observed in a reconstituted system lacking DNA repair factors 18 .
Within the Dsup C-terminal region, an eight amino acid stretch (aa 363-370, RRSSRLTS) has homology to the core consensus (RSRARLSA) of the nucleosome binding domain of vertebrate High Mobility Group-N (HMGN) proteins 18,19,20 . The chromatin binding of HMGN proteins influences a wide variety of functions (including embryogenesis, development and disease protection) across diverse cell types and species 21 . Mutation of the Dsup HMGN-like domain or deletion of its entire C-terminus respectively reduces or ablates binding to reconstituted chromatin and DNA protection from hydroxyl radicals 18 . As such, in the prevailing model revealed by use of the reconstituted system, Dsup protects the genome from DNA damage by physically shielding chromatin from hydroxyl radicals, involving the Dsup HMGN-like domain within its Cterminal sequences 18 . Whether the Dsup HMGN-like domain functions in vivo to mediate the interaction with chromatin and protect it from oxidative DNA damage is unknown.
Here, we show that when highly expressed in budding yeast R. varieornatus Dsup uses its HMGN-like domain and an additional region in the adjacent C-terminal sequences to bind chromatin and protect the genome from oxidative DNA damage in a manner dependent on chromatin engagement but independent of scavenging hydroxyl radicals. Dsup expression also extends yeast replicative lifespan in the face of chronic endogenous oxidative DNA damage. A detailed analysis of [Dsup : nucleosome] engagement finds that its HMGN-like domain mediates interaction with both the H2A/H2B acidic patch on the nucleosome surface and the H3/H4 Nterminal tails, while the distal C-terminal sequences binds DNA. Of note such a binding mechanism supports a broad engagement with in vivo chromatin independent of the landscape of histone post-translational modifications (PTMs). Our studies indicate that tardigrade Dsup can be introduced to a heterologous in vivo system and confer viability and longevity. This is achieved by physically coating the chromatinized genome via multivalent interactions to prevent hydroxyl radicals from damaging genomic DNA.

Heterologous expression of R. varieornatus Dsup in budding yeast protects against oxidative damage and promotes longevity in the face of increased oxidative stress
To initiate this study we expressed epitope tagged 6His-Dsup-FLAG (hereafter Dsup-FLAG) in yeast under the constitutive high output TDH3 promoter 22 , with the goal of achieving in vivo protein levels sufficient to coat the genome. Of note this yielded Dsup-FLAG of similar abundance to H2B-FLAG (Fig. 1a). To investigate the response of Dsup-FLAG yeast to chronic oxidative damage, we performed serial dilution assays on plates containing H2O2, observing a ~25-fold increased survival relative to yeast lacking Dsup (Fig. 1b). This did not extend to general protection from genotoxic insult, since Dsup-FLAG slightly decreased yeast survival in response to non-oxidative DNA-damaging agents such as alkylating methyl methanesulfonate (MMS), radiomimetic Zeocin, or UV (Fig. 1b).
In reconstituted assays recombinant Dsup protects chromatin from DSBs caused by hydroxyl radicals 18 , so we asked if Dsup expression protected the yeast genome from oxidative DNA damage. 8-oxoguanine (8-OHdG) is generated when ROS species react with DNA 23 , so we quantified the base modification after transient exposure to H2O2 and observed a significant reduction in 8-OHdG in the presence of Dsup (Fig. 1c).
ROS and oxidative damage increase with age, and reducing oxidative damage extends the lifespan of multiple species (yeast, worms, fruit flies, mice 24 ), while elevated ROS production shortens lifespan 25 . We thus asked if Dsup expression could extend yeast lifespan. In otherwise WT yeast Dsup has a negligible impact on chronological lifespan (the length of time a cell survives in a non-dividing state; Suppl. Fig. 1a), while the replicative lifespan (the maximum number of times a cell can divide), was slightly reduced (Suppl. Fig. 1b). Cells lacking the superoxide dismutase (SOD) genes are deficient in their ability to process both endogenous and exogenous ROS. As a result, they accumulate oxidative stress and damage, such that yeast lacking SOD1 have a shortened replicative lifespan 26 . When expressed in sod1Δ yeast, Dsup significantly increased their replicative lifespan (Fig. 1d), suggesting enhanced survival and longevity in the face of chronic oxidative damage.
Dsup ∆C was thus omitted from further in vivo study. Importantly, Dsup 3R/3E and Dsup ∆C+NLS proteins were expressed at least as well as Dsup (WT) in yeast (Fig. 2c), and the presence of each did not significantly impact cell growth (Fig. 2d).
We next examined the ability of Dsup mutants to enhance survival after chronic H2O2 exposure. Mutation of the HMGN-like domain (Dsup 3R/3E) protected cells comparably to Dsup (WT), while Dsup ΔC+NLS yielded no protection, with similar growth to an empty-vector strain ( Fig. 3a). As such, the entire C-terminus of Dsup is important for protecting yeast against oxidative DNA damage, while the included HMGN-like domain is dispensable for this function. The observed sensitivity of yeast expressing Dsup (WT) to MMS, Zeocin and UV (Fig. 1b) was not seen upon expression of Dsup 3R/3E or Dsup ∆C+NLS (not shown), indicating that while Dsup 3R/3E can protect from oxidative damage, it does not fully mimic the WT protein.
To examine whether Dsup expression has any influence on growth following acute oxidative stress, we exposed cells in liquid culture to H2O2 for 1.5 hours, before allowing them to recover on plates with no oxidizing agent. Here Dsup or Dsup 3R/3E expression significantly (and indistinguishably) increased survival following acute oxidative stress, while Dsup ΔC+NLS conferred no protection (Fig. 3b). As such the findings from chronic and acute H2O2 exposure analyses are consistent with expression of Dsup or Dsup 3R/3E, but not Dsup ΔC+NLS, promoting yeast survival in response to oxidative stress.
Free-radical scavengers are effective at protecting yeast from oxidative stress and extending lifespan 30 . Therefore, we investigated whether Dsup acts as a free-radical scavenger.
Redox-sensitive GFPs are excited at 405 nm in an oxidizing environment but 488 nm in reducing conditions, so emissions from excitation at [405/488 nm] allows the measurement of relative changes in redox state. To make a nuclear reporter for this study we added a C-terminal NLS to a roGFP2-Grx1 (glutathione reductase enzyme Grx1 31 ) fusion, and confirmed the desired subcellular localization (Suppl. Fig. 2). Using this approach, we found that the redox state of the nucleus increased upon H2O2 treatment, but this was not impacted by any Dsup alleles (Fig. 3c).
Therefore, Dsup expression had no influence on the yeast nucleus redox state, indicating it uses a mechanism distinct from ROS scavenging to protect the genome from oxidative damage.

Dsup binds chromatin throughout the yeast genome, in a manner dependent on sequences within the C terminus
Dsup was first isolated from the chromatin fraction of Tardigrade cells 4 , and shown to bind preferentially to nucleosomes over free DNA in vitro 18 . Therefore, we investigated if Dsup binds yeast chromatin in vivo. After cellular fractionation to separate chromatin-bound from soluble proteins, Dsup and Dsup 3R/3E were enriched in the chromatin-bound fraction (Fig. 4a). By contrast, Dsup ΔC+NLS was entirely in the soluble fraction (Fig. 4a), suggesting that despite nuclear localization (Fig. 2b), it does not bind chromatin. Of note, the chromatin localization of Dsup and Dsup 3R/3E, but not Dsup ΔC+NLS, parallels their ability to promote cell survival in the face of oxidative damage (Fig. 3a,b), suggesting that chromatin binding is key.
Tardigrade Dsup expression in human and plant cells alters transcription factor binding and gene expression in response to DNA damage 5,7 . This suggests that Dsup may bind 8 preferentially to certain areas of the genome to influence gene expression. Alternatively, to have the largest physically protective effect from oxidative DNA damage, Dsup might uniformly coat the genome. To investigate these possibilities, we used Cleavage Under Targets & Release Using Nuclease (CUT&RUN) 32 to map 6His-Dsup-FLAG localization (by anti-FLAG) across the yeast genome, and observed that Dsup (WT) associated with all regions, with little noticeable bias or selectivity (i.e. without forming peaks / domains; Fig. 4b). Of note, the ability of CUT&RUN to map transcriptionally active promoters with anti-H3K4me3 was unaffected by Dsup (compare Empty vector and Dsup (WT)), indicating a minimal impact on local chromatin structure (Fig. 4b). In agreement, on titrated MNase digestion of yeast cells we observed no significant difference in chromatin accessibility between strains -/+ Dsup expression (not shown).
We next compared CUT&RUN across Dsup alleles, first noting that the relative DNA yield post MNase digestion (prior to adapter ligation) was consistently Dsup (WT) >> Dsup 3R/3E > Dsup ΔC+NLS > Empty vector (EV) (Suppl. Fig. 3). This suggests Dsup 3R/3E has weaker (or higher turnover) binding relative to Dsup (WT) during the CUT&RUN steps prior to MNase activation. Their relative yield is mirrored in the CUT&RUN data, where Dsup 3R/3E showed less enrichment than Dsup (WT) across all genomic regions, while Dsup ΔC+NLS resembled empty vector ( Fig. 4b; all data group scaled after normalizing to E. coli spike-in to allow comparisons of global changes in factor binding). It would appear CUT&RUN is a more stringent analysis of chromatin interaction (presumably due [at least in part] to the long incubation times) as compared to chromatin fractionation where Dsup (WT) and Dsup 3R/3E were indistinguishable (Fig. 4a).
Taken together, these data indicate that Dsup binds without obvious bias across the genome in a manner that is dependent on its C-terminus (which includes the HMGN-like domain), while mutation within the HMGN-like domain (3R/3E) reduces chromatin binding relative to wild type Dsup, but not enough to confer loss of protection from oxidative DNA damage (Fig. 3a,b).

Dsup binds nucleosomes via multivalent interactions with the histone tails, acidic patch, and DNA
To rigorously interrogate the mode of interaction of Dsup with nucleosomes or free DNA, we used the dCypher in vitro chemiluminescent assay 33 . Here the biotinylated target (e.g., free DNA or fully defined mononucleosome) couples to streptavidin-donor beads while epitope-tagged query (here WT or mutant forms of 6His-Dsup-FLAG (Suppl. Fig. 4) 18 ) couples to anti-tag acceptor beads. After mixing potential reactants the donor beads are excited at 680 nm, releasing a singlet oxygen that causes emission (520-620 nm) in proximal acceptor beads: this luminescent signal is directly correlated to interaction / binding affinity (Fig. 5a). To compare across each [Query : Target], data is presented as their relative concentration effective in producing 50% of the maximal response (EC50 rel ) by plotting Alpha Counts (fluorescence) as a function of protein concentration (see Suppl. Table 3 for all EC50 rel from this study).
To begin these studies, we titrated salt (sodium chloride) to examine the potential complication of non-specific ionic interactions (Suppl.  The dCypher platform allowed us to query a diversity of fully defined mononucleosomes (Suppl . Table 1D) to ascertain which surfaces are most important for Dsup binding to chromatin ( Fig. 5d-g). Here the apparent affinity of Dsup for mononucleosomes was minimally impacted by a variety of lysine acylations or methylations (Fig. 5d, showed noticeably reduced binding to intact and tail-less nucleosome relative to free DNA (compare to WT: Fig. 6a,b). The addition of competitor DNA (to conditions optimized for Dsup WT) then reduced nucleosome binding by Dsup 3R/3E below the level of detection (Fig. 6a,b). In profound contrast, Dsup ΔC (which lacks the HMGN-like domain and C-terminal sequences) showed no interaction with nucleosomes or free DNA under conditions optimized for Dsup (WT) ( Fig. 6c), only binding at reduced NaCl (Suppl. Fig. 6

Dsup interaction with either the histones or DNA is sufficient to survive oxidative damage
Our finding that the Dsup-DNA interaction remained intact after mutation of the HMGNlike domain (aa 363-370: Fig. 6b), but was lost on deletion of the entire C-terminus (D360-445:  . 7a). This allele is expressed in yeast at slightly reduced levels relative to the other forms of Dsup ( Fig. 7b) but was notably able to promote yeast survival in the face of chronic H2O2 exposure (Fig. 7c). These data show that either an intact HMGN-like domain or intact C-terminal downstream sequences, by respectively binding to the nucleosome or DNA, are sufficient for Dsup to protect the genome against oxidative damage ( Figure 8).

DISCUSSION
To understand the molecular basis of the extreme radioresistance of tardigrades, we investigated if, and how, their Dsup protein protects against oxidative damage in vivo. When expressed in budding yeast, Dsup coated the entire genome without apparent bias, using two Cterminal regions to associate with chromatin via multivalent interactions involving several nucleosome surfaces and DNA. Functionally, this engagement prevents oxidative DNA damage in a manner independent of ROS scavenging. Our data supports a model where Dsup mediates multivalent interactions with chromatin to protect the underlying genome from oxidative DNA damage (Fig. 8), thus promoting yeast survival and longevity after exposure to elevated levels of hydroxyl radicals (Fig. 1b,d).
HMGN proteins are primarily described in vertebrates 38 Fig. 5g). However, we additionally find that the Dsup HMGN-like domain binds histone tails, as this interaction is again lost in Dsup 3R/3E (Fig. 5e,f) The observation that Dsup binding to the nucleosome was largely agnostic to most histone PTMs in vitro (Fig. 5d,e) is consistent with our finding that Dsup covers the entire in vivo genome rather than being enriched / excluded from certain regions with their particular histone modifications (Fig. 4b). Importantly, Ramazzottius varieomatus histones are highly conserved with human histones (as used in the dCypher assay) (Suppl . Table 1E), while yeast and human histones are even more similar. As a result, we consider that the observed interactions between Dsup and human or yeast histones are relevant for how the protein helps to protect tardigrades from irradiation.
In initial testing we expressed Dsup from a range of yeast promoters of various strengths, but only the very strong TDH3 promoter enabled protection from oxidative damage (not shown and Fig. 1b). Of note, this Dsup expression level was equivalent to that of histone H2B (Fig. 1a), suggesting Dsup may be in sufficient abundance for at least two molecules per yeast nucleosome.
Given that highly expressed Dsup protects the genome from oxidative DNA damage (Fig. 1c), is bound to chromatin genome-wide (Fig. 4a,b), and redundantly interacts with multiple nucleosome surfaces (Fig. 5), it is likely that Dsup non-specifically coats the in vivo genome to physically protect from oxidative damage, as was proposed from the previous in vitro studies 18 . It may be relevant to note that when we yeast-codon optimized the tardigrade Dsup protein in an attempt to promote still higher expression levels, the resulting yeast were inviable, suggesting that too much It is intriguing that Dsup expression protected yeast from oxidative damage, but not from MMS, bleomycin, or UV: indeed, it actually increased sensitivity to these agents (Fig. 1b). Future studies should examine whether there is delayed repair of the DNA lesions generated by these genotoxins, potentially due to Dsup hindering access of the repair machinery. We note, however, that the growth rate of Dsup yeast was not reduced (Fig. 2d), indicating they are fully capable of transcriptional regulation, DNA replication and mitosis -other events one could imagine might also be prone to complications from the genome being coated with Dsup protein -but that did not appear to be the case.
These findings provide precedent for the development of organisms that can survive and live longer in the face of oxidative damage, potentially expanding the range of applications for developing therapeutic interventions by biotechnology, and furthering efforts towards human resistance to extraterrestrial effects.
Yeast culture and handling was performed using standard methods. Growth of strains expressing Dsup was in SC-ura media (unless otherwise indicated). All strains were isogenic to BY4741 49 (Suppl . Table 1B).

Growth curve analysis
Yeast were grown to saturation overnight in YPD at 30°C and diluted to ODl600 0.1-0.2.
Growth measurements (ODl600) of cultures grown from three independent colonies were taken every 30 minutes and plotted over time. Growth curves were fitted with an exponential regression using Microsoft Excel, and doubling times calculated as the slope of the curve during exponential phase. Doubling times of independent growth curves were compared using a student's t-test.

Acute and chronic damage sensitivity analysis
To measure resistance to acute hydrogen peroxide (H2O2) exposure, cells were grown in liquid YPD media until mid-log, harvested by centrifugation, and resuspended to 0.6 OD in fresh media containing H2O2 (0, 4, 6, or 8 mM). After 90 minutes growth (30°C, with shaking), cultures were diluted and spread on SC-ura agar plates. After two days at 30°C, colonies were counted and averaged across three technical replicates. Three experiments were performed from separate starting colonies, and statistical analysis performed using a student's t-test.
The response to chronic H2O2 exposure was examined using a serial dilution assay. Cells were grown in liquid culture until mid-log (ODl600 0.5-1.0), harvested by centrifugation, and resuspended in sterile water to ODl600 1.0. Five-fold serial dilutions were made in a 96-well plate, and yeast spotted using a sterile 6x8-prong inoculating manifold onto YPD agar plates containing indicated concentrations of H2O2. Similar methods were used to evaluate sensitivity to methyl methanesulfonate (at the indicated concentrations in YPD) and Zeocin (at the indicated concentrations in YPD). For ultraviolet light sensitivity, yeast serial dilutions were onto YPD plates and exposed to UV (at the doses (J/cm 2 ) indicated in figure legends) using a crosslinker [Stratalinker]. Plates were incubated for 3 days at 30°C.

Replicative lifespan analysis
Cells were grown overnight to early-mid (ODl600 0.2-0.6) and diluted to OD 0.1 in freshlyfiltered YPD. This innoculum was added to an iBiochips automated dissection chip to achieve single cell loading as per manufacturer's instructions. Light microscopy images of cells were acquired every 20 minutes over four days using an Evos FL Auto two-cell imaging microscope and associated software (ImageJ). At least 50 cells were counted per condition, with survival curves calculated on Graphpad Prism 9, and statistical analysis performed with a log-rank test.

Chronological lifespan analysis
Chronological lifespan was measured according to published methods 53 . Data is presented as average and standard deviation across three independent cultures, each of which is an average of two technical replicates.
Fluorescence at 405 nm and 488 nm was measured on a flow cytometer (BD Biosciences BD ® LSR II) immediately before direct addition of H2O2 (2 mM or 10 mM). Subsequent fluorescence measurements were taken every 20 minutes over 80 minutes.
The mean of the 405/488 nm values for each timepoint was calculated using FlowJo, with the value at time 0 normalized to 1 for each strain. Data is presented as the mean and standard deviation of three independent cultures and compared using a student's t-test.

ELISA for 8-OHdG
30 mL yeast cultures were grown at 30°C in shaking flasks until ODl600 0.6. Cells were harvested by centrifugation, and half of each culture resuspended in either 15 mL of fresh SC-ura media or that containing 10 mM H2O2. After two hours growth at 30°C, cells were again harvested by centrifugation and genomic DNA isolated (Thermo Scientific Yeast DNA extraction kit).
Genomic DNA was resuspended in 50 μL of nuclease-free water and stored overnight at 4°C.
DNA concentrations were measured using a NanoDrop spectrometer, diluted in water to 2 mg/mL, boiled for five minutes at 95°C, then immediately placed on ice for 10 minutes (to denature double-stranded DNA). 50 μg of DNA (25 μL) were sequentially incubated with Nuclease P1 (NEB: 1 unit for 2 hours at 37°C in provided buffer) and alkaline phosphatase (NEB Quick CIP: 10 units for 1 hour at 37°C in provided buffer supplemented with 100 mM Tris pH 8). Samples were incubated to denature enzymes (10 minutes at 95°C), then spun at 6000 x g for 5 minutes.
DNA concentrations were measured on a NanoDrop spectrometer to ensure even loading onto the ELISA plate.
ELISA to 8-hydroxy 2 deoxyguanosine was performed as per kit instructions (Abcam ab201734). 15 μg of DNA was loaded into each of three triplicate wells for each sample (with three independent cultures measured for each condition). Absorbance at 450 nm was measured using a plate reader.

CUT&RUN analysis
Nuclei from yeast cells expressing Dsup alleles (Suppl . Table 1B) were purified according to published methods 54 with slight modifications. Yeast were grown in 500 mL of SC-ura media to ODl600 0.6-0.8. Cells were spheroplasted using 500 μL of 2 mg/mL Zymolyase 100T (37°C for ~ 30 mins; until a 50 μL aliquot mixed with 1 mL of 10% SDS had an ODl600 ~10% of the starting value). Remainder of the nuclei isolation was performed as previously 54 , and 1 mL aliquots containing 5 x 10 7 nuclei were slow-frozen in an isopropanol chamber at -80°C overnight.
For CUT&RUN nuclei were rapidly thawed (2-3   21 an average of ~1.1 million paired-end reads per reaction (Suppl. Table 2). Paired-end fastq files were aligned to the sacCer3 reference genome using Bowtie2. Duplicate (SAMtools) and multialigned (Picard) reads were filtered, and the resulting unique reads for comparable reactions normalized by an E. coli scaling factor (1/ % E. coli Reads) (bedtools), and further normalized to RPKM bigwig files (DeepTools). Integrative Genomics Viewer (IGV) was utilized for the visualization of peaks from bigwig files. All sequencing data has been deposited in the NCBI Gene Expression Omnibus (GEO) with accession number GSE237436.