Monitoring Schizosaccharomyces pombe genome stress by visualizing end-binding protein Ku

ABSTRACT Studies of genome stability have exploited visualization of fluorescently tagged proteins in live cells to characterize DNA damage, checkpoint, and repair responses. In this report, we describe a new tool for fission yeast, a tagged version of the end-binding protein Pku70 which is part of the KU protein complex. We compare Pku70 localization to other markers upon treatment to various genotoxins, and identify a unique pattern of distribution. Pku70 provides a new tool to define and characterize DNA lesions and the repair response.


INTRODUCTION
The response to genome stress and DNA repair can be observed in living cells in real time, by monitoring fluorescently-tagged DNA damage response proteins (Lisby et al., 2004;Lukas et al., 2005;Nagy and Soutoglou, 2009;Polo and Jackson, 2011). This has allowed characterization of dynamic processes that respond to damage and preserve genome integrity, including cell cycle, checkpoint, repair, and recovery pathways. In the fission yeast Schizosaccharomyces pombe, accumulation of foci of the singlestrand DNA binding protein Ssb1 (a subunit of Replication Protein A /RPA), and of the recombination protein Rad52, have been used to characterize intrinsic genome stresses as well as the response to external genotoxins (Meister et al., 2003;Kilkenny et al., 2008;Carneiro et al., 2010;Bass et al., 2012;Sabatinos et al., 2012). These proteins recognize and respond to single strand DNA accumulation, which can result from exonuclease activity, resection, processing of replication forks, and recombination intermediates, or R-loop or D-loop formation (Zeman and Cimprich, 2014;Sabatinos and Forsburg, 2015). Importantly, this has led to identification of distinct patterns of accumulation that can serve as fingerprints for different forms of genome stress (e.g. Sabatinos et al., 2012;Sabatinos and Forsburg, 2015;Ranatunga and Forsburg, 2016).
In this report, we describe the development of a new fluorescent marker for fission yeast, the Pku70 subunit of the Ku protein complex. We constructed a pku70 + -citrine fusion and integrated into the genome in wild-type fission yeast under the endogenous promoter. We examined its behavior and accumulation in treated and untreated wildtype cells in response to different genotoxins. We compared localization of Ku to Rad52, RPA, and Mre11 markers and observe a pattern of foci that is distinct from other proteins. This provides a new tool to characterize responses to different forms of genotoxic stress and a useful addition to the fission yeast tool kit for investigation of the 3-Rs of DNA replication, repair, and recombination.

Construction of strains with fluorescently tagged Pku70 and Mre11
Ku (a heterodimer of Pku70/80) and MRN (Mre11/Rad50/Nbs1) protein complexes are known for high affinity for binding DNA ends (Fell and Schild-Poulter, 2015;Shibata et al., 2018). We tagged Pku70 on its C-terminal end with Citrine fluorescent protein and integrated into the endogenous locus (see the Materials and Methods). Using a similar strategy, we also tagged Mre11 on its C terminal end with mCherry fluorescent protein. The resulting strains were compared to wild-type, pku70Δ, mre11Δ, and rad51Δ for their growth on four typical genotoxic drugs: methyl methanesulfonate (MMS), which creates alkylation damage that inhibits DNA replication fork progression; camptothecin (CPT), which blocks Topoisomerase I cleavage; hydroxyurea (HU), which causes nucleotide starvation and fork pausing; and Phleomycin (phleo), a radio-mimetic that causes single-and DSBs. Both the Mre11-mCherry and Pku70-Citrine tagged strains behaved the same as wild-type under normal growth and genotoxic stress. The Δpku70 strain also shows no sign of genotoxin sensitivity, as reported previously (Manolis et al., 2001;Sánchez and Russell, 2015) (Fig. 1A). In order to confirm that Pku70-Citrine is active, we performed a plasmid religation assay showing the tagged construct retains NHEJ function. (Fig. S1). We also find that deletion of Pku80 abolishes Pku70-citrine localization, consistent with proper assembly of the Ku heterodimer (Fig. S2).
Pku70 and Mre11 have increased nuclear signal following genotoxic stress In normal growth conditions, the tagged strains show a few scattered foci in Pku70-citrine cells and diffuse nuclear fluorescence in Mre11-mCherry cells (Fig. 1B). We examined the distribution of signal in cells treated with MMS, CPT, Phleo, or HU at 32°C after 4 h. There is a significant increase of cells with individual Pku70 nuclear foci in MMS, CPT, and to a lesser extent Phleo. Cells treated with HU did not show any significant difference from wild-type (Fig. 1C). In contrast, the Mre11-mCherry signal showed diffuse pan-nuclear staining in untreated cells (Fig. 1B). Following 4 h of treatment with the four genotoxic drugs, Mre11-mCherry did not show obvious foci. Rather, we observed generalized areas of increased fluorescence over threshold, but these typically were not well-defined discrete puncta as seen with other markers.

Colocalization of Pku70 and Mre11 with other markers of DNA damage
Previous studies of genome instability in fission yeast have imaged the single stranded binding protein Ssb1 (Rad11, RPA) and the homologous recombination protein Rad52 in response to different forms of replication stress (Meister et al., 2003;Kilkenny et al., 2008;Carneiro et al., 2010;Bass et al., 2012;Sabatinos et al., 2012). We examined co-localization using CPT, MMS, HU and Phleo in a strain with Pku70-citrine, Rad52-mCherry, and RPA-CFP. Four hours after drug addition at 32°C, we determined frequency of colocalization among all three tagged proteins. While there was partial overlap, we observed that Pku70 is not completely concordant with the other markers ( Fig. 2A). The number of foci per nucleus was calculated and binned as either 1or ≥2 foci using an automatic foci counter in ImageJ as described in the Materials and Methods (Fig. 2B). We observed that CPT 20 µM contained the highest frequency of Pku70 foci, then MMS, Phleo, and HU. The difference from prior observation likely reflects a somewhat different drug dosage: CPT levels were raised from 10 µM to 20 µM in order to produce an enhanced response and MMS was lowered from 0.9 mM to 0.45 mM to better resolve single foci.
Colocalization was determined using the objects-based method in the ImageJ plug-in JACoP (see the Materials and Methods). Fig. 2C shows the proportion of Pku70-Citrine foci that overlap with a thresholded region for Rad52-mCherry or RPA-CFP. For CPT, MMS, and Phleo, these proportions vary from 60-90%. In contrast, the scattered foci in HU showed only about 30% of Ku co-associating with another marker. Fig. 2D shows the proportion of Rad52-mCherry foci that have a colocalizing Pku70-Citrine focus. CPT contained the highest proportion of Rad52 as well as RPA with overlapping Pku70 foci, whereas HU contained the lowest.
We performed a similar study with Mre11-mCherry but could not perform the same quantitation because Mre11-mCherry does not form discrete foci. We observed areas of generally increased fluorescence but never clear puncta as with Pku70, Rad52, or RPA. Observing these cells in three-dimensional reconstruction showed no obvious colocalization between Rad52-YFP/RPA-CFP and Mre11-mCherry in live cell video microscopy, or in static images (Fig. 3A,B; Fig. S3).

Pku dynamics in S phase specific damage
The genotoxin MMS causes alkylation damage, generating lesions that block DNA polymerase (Lundin et al., 2005). This typically results in replication template switching (Barbour and Xiao 2003;Andersen et al. 2008). Previous work has suggested that Ku is recruited by blocked and regressed replication forks (Teixeira-Silva et al., 2017). Therefore, we investigated the dynamics of Ku response to MMS treatment as a model for disruptions in replication fork progression. We used live cell video microscopy to observe cells containing Rad52-mCherry and Pku70-Citrine over a 5 h period of MMS (0.45 mM) treatment at 28°C. We observe distinct dynamics for Rad52 and Pku70 recruitment during treatment. While absolute timing differs in individual cells, typically a Ku focus appears for a short time and partially co-localizes with Rad52. Fig. 4A shows a representative newborn cell that is likely in mid S phase, 1 h 20 m after drug treatment. The diffuse Rad52-mCherry signal is distributed in smaller foci which then coalesce into two large foci. Pku70-Citrine colocalizes at the center of these large foci for about 20-40 min. The large Rad52-mCherry foci persist for another 60 min and then begin to dissipate. Retention time of Pku-Citrine foci in MMS is ≤20 min with a fraction of cells maintaining it longer between 20 and 40 min. In contrast, Rad52 foci extend over a much longer period of time ranging from 20 all the way up to 160 min (Fig. 4B). Overall Rad52 tends to appear slightly earlier than Pku70 in most cells and disappears much later (Fig. 4C). (Additional time-lapse images found in Figs S4 and S5).

DISCUSSION
Localization of repair puncta in fission yeast has been a wellestablished means of observing DNA damage, quantified by counting foci, determining pixel intensity or size of foci, and threedimensional position in the nucleus (Green et al., 2015). The most frequently used fluorescent tags used in S. pombe for observing DNA lesions are the recombination protein Rad52 and single strand DNA binding protein Rad11, a subunit of RPA (Meister et al., 2003;Carneiro et al., 2010;Sabatinos et al., 2012). Studies have shown that in cycling wild-type cells, approximately 10-20% of cells show evidence of single RPA or Rad52 foci, likely due to sporadic S phase damage. The tagged proteins show distinct patterns in response to genotoxic stresses induced by mutations in the replication or repair pathways (Sabatinos et al., 2012;Sabatinos and Forsburg, 2015;Ranatunga and Forsburg, 2016), or in response to exogenous agents such as HU, which causes replication fork stalling (Thelander and Reichard, 1979); MMS, an alkylating agent that generates lesions that block the replication fork (Lundin et al., 2005); camptothecin (CPT), a topoisomerase I inhibitor that leads to S-phase specific double strand breaks (Li and Liu, 2001); and bleo-or phleomycin, radiomimetic drugs that causes single-and DSBs (Povirk, 1996).
The current study seeks to expand the library of tagged proteins, part of our strategy to develop a fingerprint for the response to different forms of genotoxic stress. We investigated fluorescently tagged Mre11 and Pku70 as markers for DNA breaks.
The MRN complex is one of the earliest responders to DSBs (Shibata et al., 2018;Syed and Tainer, 2018) and is essential to drive resection (Wang et al., 2014;Shibata et al., 2018;Langerak et al., 2011;Teixeira-Silva et al., 2017). Our Mre11-mCherry construct showed a diffuse pan-nuclear signal in untreated cells. We did not see obvious focus formation of Mre11-mCherry following treatment with genotoxins. Rather, it maintained a diffuse signal with regions of brightness. In other systems, MRN has been shown to be an immediate responder to DSBs induced by ionizing radiation (Maser et al., 1997). Our failure to see this form of localization may be related to the timing of our analysis, and/or the diffuse distribution of lesions in drug-treated cells, compared to concentrated sites of damage from of ionizing radiation.
Previous whole-cell localization of Pku70 in S. pombe was carried out using C terminal epitope-tagged Pku70 and immunofluorescence on fixed cells (Manolis et al., 2001). In unperturbed cells, a diffuse pan-nuclear localization was observed. Association of Ku with DNA ends has been investigated using chromatin immunoprecipitation; in wild-type cells, it is not enriched unless the MRN complex is missing (Langerak et al., 2011;Teixeira-Silva et al., 2017). Visualization of Pku70 in live fission yeast cells has not previously been performed.
We saw few Ku foci in wild-type cells, consistent with the previous immunofluorescence studies. Treatment for 4 h with our panel of genotoxins showed that HU has little to no accumulation of Ku foci. Treatment with CPT causes a modest increase in the fraction of cells with foci at 10 µM and a more dramatic increase at 20 µM. Similarly, phleomycin, a radiomimetic that causes DNA breaks throughout the cell cycle, has a modest but limited increase in foci relative to untreated cells.
We found that the most dramatic increase of cells with Pku70 foci was obtained by treating with MMS at 0.9 mM. MMS is an alkylating agent that results in error-free and error prone base excision repair during S phase, and thus leading to trans lesion synthesis (Memisoglu and Samson, 2000). This induction in MMS is consistent with prior observations suggesting that Ku is recruited to regressed or broken replication forks in order to stabilize the free end (Langerak et al., 2011;Teixeira-Silva et al., 2017). This suggests that even in MRN + cells, there are situations where Ku remains associated with sites of genome stress.
We observed a substantial colocalization between RPA or Rad52 and Ku, in cells treated with MMS, CPT, or Phleo. This result was a surprise as many models suggest Pku should be removed by the time resection and recombination proteins are recruited. One possibility for the S phase specific toxins is that Pku could be binding to reversed forks at repair centers. Previous studies suggest that Pku plays a role at reversed forks in order to maintain genome stability, particularly in cells with defective HR repair such as brc1Δ (Sánchez and Russell, 2015;Teixeira-Silva et al., 2017). This may reflect that other mechanisms than exonuclease activity can generate ssDNA, including helicase unwinding and strand invasion.
To address this finding in dynamic conditions, we examined MMS-treated cells as a model for stalled replication forks. Previously, we showed that MMS induces a dramatic increase in RPA and Rad52 foci relative to other genotoxins (Ranatunga and Forsburg, 2016). We observe substantial recruitment of Rad52-mCherry and brief, partial co-localization of Pku70. The Pku70 signal, largely in 1-2 foci, appears after Rad52 and disappears before Rad52 is resolved. Further molecular work will be required to determine what this signal represents.
It is likely that Ku foci will define distinct structures associated with particular forms of replication stress. For example, in a recent study, our lab showed that a mutant mcm4-dg with a defect in the MCM helicase accumulates Ku foci (Kim and Forsburg, 2020). This accumulation can be reversed by activation of the Mus81 resolvase. Mus81 is essential for viability in pku80Δ brc1Δ mutants (Sánchez and Russell, 2015), indicating a collaboration between Ku and Mus81 in response to replication stress. Our Pku70-citrine fusion will be a key reagent in dissecting this and other activities.

Cell growth and physiology
Fission yeast strains are described in Table 1, and were grown as in (Sabatinos et al., 2012).

Construction of tagged strains
All fragments were lengthened using the Expand Long Template PCR System (Roche Diagnostics, Mannheim Germany). Primers were designed using the NCBI Primer design tool and optimized to an annealing temperature of 52-54°C (Ye et al., 2012). Full length fragments were transformed using electroporation and selected using the appropriate marker (Sabatinos and Forsburg, 2010). Upon transformation, instead of plating directly onto selective minimal media, the cells were first plated on YES for 24 h then replica plated onto YES-Hph. Candidate colonies growing on Hph after 4-5 days were then restreaked onto Hph twice and visually screened for nuclear localizing foci. Pku-Citrine::Hph The Pku C-terminal Citrine fragment was formed from five fragments, Pku 5'overhang (FY2710+FY2711), Citrine (FY2561+FY2562), Hph (FY2563+ FY2564), Citrine UTR (FY2565+FY2566), and Pku 3′ UTR overhang (FY2712+FY2713). The Citrine and Citrine UTR fragments were lengthened from Addgene plasmid pKT0139 (Sheff and Thorn, 2004). The Hph fragment was lengthened from pFA6a-hphMX6 (Hentges et al., 2005). The 5′ and 3′ UTR overhang fragments were lengthened from phenol:chloroform extracted WT (FY527) DNA (Forsburg and Rhind, 2006). The Citrine, Hph, and Citrine UTR fragments were first lengthened to form a full Citrine::Hph fragment. A single PCR reaction was then done with Pku 5′ overhang, Citrine::Hph, and Pku 3′ UTR overhang fragments forming the full fragment. This fragment was then used for electroporation transformation. All primers used to make these fragments can be found in Table 2.

Plasmid religation assay
Plasmid pAL19 (Barbet et al., 1992) was previously modified by insertion of an mst1 derivative in the HindIII sites of the multiple cloning sequence, creating pRCP44. pRCP44 was digested for one hour at 37°C with HindIII. The digested fragment containing only the linearized backbone pAL19 was gel-extracted. This ensured that any nonlinearized plasmid was not contaminating the sample. All strains contained the leu1-32 mutation and were transformed by lithium acetate transformation with either the circularized backbone pAL19 or the linearized backbone pAL19 from the digested/extracted fragments of pRCP44 containing the S. cerevisiae LEU2+ gene. Colonies were grown on -Leu plates for 5 days. Colony counts were normalized to wild-type plasmid religation versus circular plasmid transformation rates.

Live cell imaging
Cells were prepared as in (Green et al., 2015). Medium for all live cell imaging was PMG-HULALA (PMG+Histidine, Uracil, Leucine, Adenine, Lysine, Arginine) (225 mg/l each) (Sabatinos and Forsburg, 2010 Three-dimensional stacks were deconvolved with manufacturer provided OTFs using a constrained iterative algorithm, images were maximum intensity projected for presentation. Images were contrast adjusted using a histogram stretch with an equivalent scale and gamma for comparability. Brightfield images were also acquired.

Image processing and analysis
Images were contrast adjusted using an equivalent histogram stretch on all samples. Significance was assessed with Mann-Whitney tests. Long-term time lapse videos were stabilized in ImageJ-Fiji (Schindelin et at. 2012) using the package 'StackReg' by Philippe Thevanaz from the Biomedical Imaging Group at the Swiss Federal Institute of Technology Lausanne (Thevenaz et al., 1998). Foci were automatically quantified using a computational algorithm based on uniform threshold per fluorescence channel as described by the light microscopy core facility at Duke University (https://microscopy.duke.edu/guides/count-nuclear-foci-ImageJ). Object based colocalization analysis was performed using the ImageJ plug-in JACoP on the same images used for the focus quantification. However this object based colocalization analysis method still requires observer-based thresholding before analysis. In order to mitigate observerbased thresholding bias, the number of observed objects after thresholding per fluorescence channel was calculated to be within 10 foci of the automatically counted foci during the previous computer-based foci quantification analysis described above.