The thymine analogue uracil is one of the most frequent non-canonical bases in DNA appearing either by thymine replacing misincorporation or as a product of spontaneous or enzymatic cytosine deamination reaction (Krokan et al., 2002). Consequently, uracil in DNA is usually recognized as an error that is efficiently repaired by the multistep base excision repair (BER) pathway initiated by uracil-DNA glycosylases (UDGs) (Krokan and Bjørås, 2013; Wallace, 2014). In other respects, uracil in DNA is known to be involved in several physiological processes (e.g. antibody maturation [Liu and Schatz, 2009; Maul and Gearhart, 2010; Maul and Gearhart, 2014; Xu et al., 2012], antiviral response [Burns et al., 2015; Stenglein et al., 2010], insect development [Horváth et al., 2013; Muha et al., 2012]), however, the exact mechanism and regulation of uracil-DNA metabolism including the roles of UDGs need to be elucidated. There are four known members of the UDG family in humans: (i) the most active uracil-DNA glycosylase encoded by the ung gene (UNG1 mitochondrial and UNG2 nuclear isoform), (ii) the single-strand selective monofunctional uracil-DNA glycosylase 1 (SMUG1), (iii) thymine DNA glycosylase (TDG specialized for repair of T:G and U:G) and (iv) methyl CpG binding domain protein 4 (MBD4 repairs U:G) (Visnes et al., 2009). UNG2 removes most of the genomic uracil from both single- and double-stranded DNA regardless of the uracil originating from mutagenic cytosine deamination or thymine replacing misincorporation (Kavli et al., 2002).

Thymine replacing uracil misincorporation is normally prevented by the tight regulation of the cellular dUTP/dTTP ratio maintained by two enzymes, the dUTPase and the thymidylate synthase. The dUTPase enzyme removes dUTP from the cellular pool by catalyzing dUTP hydrolysis into dUMP and PP_i (Vértessy and Tóth, 2009). Lack or inhibition of dUTPase leads to increased dUTP levels and under such conditions, DNA polymerases readily incorporate uracil opposite to adenine. Similarly, several anticancer drugs (such as 5-fluorouracil (5-FU), 5-fluoro-2’-deoxyuridine (5FdUR), capecitabine, methotrexate, raltitrexed (RTX), pemetrexed) target the de novo thymidylate synthesis pathway via thymidylate synthase inhibition to perturb the tightly regulated dUTP/dTTP ratio, eventually triggering thymineless cell death (Blackledge, 1998; Requena et al., 2016; Wilson et al., 2014). Although the exact molecular mechanism is not yet fully understood, massive uracil misincorporation, hyperactivity of the repair process and/or stalling of the replication fork are all suggested to be involved in the process (Khodursky et al., 2015; Ostrer et al., 2015). UNG has been suggested to play a key role in this mechanism, as being responsible for the initiating step in uracil removal that may lead to futile cycles if the cellular dUTP/dTTP ratio is elevated. A quantitative insight into the magnitude and the pattern of uracil incorporation into genomic DNA as induced by these chemotherapeutic treatments is expected to contribute to a better understanding of the cell death mechanism induced by the respective drugs.

Direct observation of the uracil moieties incorporated upon drug treatments have been hampered by the efficient and fast action of UNG. To overcome this problem, we wished to counteract the action of UNG in human cells via introduction of the well characterized, specific UNG inhibitor, UGI (Luo et al., 2008; Mol et al., 1995) into the cellular milieu. It has already been shown that UGI expression does not affect either the cytotoxicity, or the DNA damage and cell cycle response upon RTX and 5FdUR treatment (Luo et al., 2008). Using UGI expressing cell lines, we aimed to reveal the nascent pattern of uracil moieties in DNA induced by perturbation of thymidylate metabolism both using genome-wide uracil-specific sequencing and in situ cellular imaging of uracils within human genomic DNA. Previously, we designed a uracil-DNA (U-DNA) sensor tailored from an inactive mutant of human UNG2 that was successfully applied in semi-quantitative dot blot analysis and direct immunocytochemistry (Róna et al., 2016). Some additional approaches have also been published to detect uracil-DNA within its genomic context such as (i) techniques focusing on specific, well-defined regions of the genome (qPCR [Horváth and Vértessy, 2010] and 3D-PCR [Suspène et al., 2005]), (ii) techniques that have been applied only to smaller sized genomes (Excision-seq [Bryan et al., 2014] and UPD-seq [Sakhtemani et al., 2019]), and (iii) techniques requiring labor-intensive isolation and multistep processing of genomic DNA samples (dU-seq [Shu et al., 2018]).

Here, we employ the U-DNA sensor in a DNA-IP-seq-like (DIP-seq-like) approach (termed as U-DNA-Seq) and develop a robust bioinformatic pipeline specifically designed for reliable interpretation of next generation sequencing (NGS) data for genome-wide distribution of uracil. We selected two drugs, RTX (raltitrexed, or tomudex) and 5FdUR that perturb thymidylate biosynthesis with different modes of action and analyzed their effects on genomic uracil distribution. These two drugs are frequently applied in treatment of colon cancers, therefore we chose a human colon carcinoma cell line, HCT116 and its mismatch repair (MMR) proficient variant as well-established and relevant cellular models (Koi et al., 1994; Meyers et al., 2001; Rashid et al., 2019). We show that drug treatment led to increased probability of uracil incorporation into more active chromatin regions in HCT116 cells expressing the UNG inhibitor protein UGI. In contrast, uracil was rather restricted to constitutive heterochromatic regions both in wild type cells and in non-treated UGI-expressing cells. Moreover, we further developed the U-DNA sensor-based staining method (Róna et al., 2016) that now uniquely allows in situ microscopic visualization of uracil in human genomic DNA. Confocal and super-resolution microscopy images and colocalization measurements strengthened the sequencing-based distribution patterns.

Here we focus on the alteration of U-DNA distribution pattern upon treatment with drugs perturbing thymidylate biosynthesis. Towards this end, we combined two new applications of further developed U-DNA sensor that was originally described in Róna et al., 2016. On one hand, using a DNA-IP-seq like application, termed U-DNA-Seq, we provided genome-wide uracil distribution data that was compared to the patterns of different genomic features. On the other hand, in immunocytochemistry, the sensor was applied to detect colocalization of U-DNA and selected histone markers.

Using U-DNA-Seq, here we demonstrate that the distribution of uracil-containing regions is altered in the drug-treated (5FdUR or RTX, in combination with UGI) as compared to the non-treated (wild type and UGI-expressing) samples. We demonstrated that UGI expression did not cause any observable change either on cell cycle progression (Figure 5, and Luo et al., 2008) or uracil distribution pattern (Figure 3). We chose HCT116 cancer cell line that is deficient in mismatch repair (MLH1^-/-), similarly to many types of cancer especially in colon cancer (Germano et al., 2018; Gupta and Heinen, 2019; Sekine et al., 2017). As its mismatch repair proficient counterpart is also available (Koi et al., 1994), we took the opportunity to address the impact of the MMR status on genomic uracil distribution. We found that the genomic uracil pattern is much more influenced by MMR proficiency in case of the 5FdUR treatment than in case of the RTX treatment (Figure 3). The genomic uracil distribution patterns either in non-treated or in drug-treated cells are found to be non-random: broad regions of uracil enriched genomic segments were detected. Within the third part of our pipeline (Figure 2—figure supplement 1 and Supplementary file 3–5), we also analyzed the distribution pattern of these broad peaks comparing them to a set of relevant and cell type-specific data of ChIP-seq experiments and other genomic features. In drug-treated cells, these broad segments showed the highest correlation with ChIP-seq-based patterns published for predominantly euchromatin and facultative heterochromatin markers (Figure 4A–B). Increasing evidence suggests that active and repressed chromatin states can be determined in a combinatorial fashion where simultaneous histone marks can efficiently shift gene expression from inactive to active states or vice versa (Gates et al., 2017; Hyun et al., 2017). Hence, it is of special interest to note that our colocalization data show similarity scores not just for one but for a variety of factors. Such combinatorial behavior was further demonstrated by the genome segmentation analysis using the Segway package (Figure 4B) that also pointed to the fact that the distribution of histone markers are not fully matched with the detected U-DNA pattern. Hence further genomic features were also studied (Figure 4C). Importantly, regarding these factors and additional features, our results are highly coherent. Namely, the outstanding correlation of uracil-DNA patterns in drug-treated samples with active promoters, CpG islands, early replicating segments and DNase hypersensitive sites, all of which are published for normally cycling cells, highly supports the above conclusion. Euchromatin was shown to imply early replicating genomic regions, whereas heterochromatin replicates in late S-phase (Black et al., 2012). Accordingly, we report that the drug treatment induced U-DNA pattern is more similar to the early replicating segments, whereas U-DNA distribution in non-treated (wild type and UGI-expressing) cells shows simultaneous association with both heterochromatin markers and late replicating regions (Figure 4C–D, also supported by Figure 4—figure supplement 3). It has to be noted that MMR proficiency leads to a major decrease in the correlation with early replication timing in case of the 5FdUR treated sample (Figure 4C–D), and a smaller decrease in the correlation with transcriptionally active regions in case of both treatments (Figure 4A–C). Still, difference between the uracil-DNA patterns of drug-treated and non-treated samples remains unambiguous, regardless the MMR status (Figures 3 and 4).

Taken together, in the non-treated cells, where the level of genomic uracil is low, we show that uracil is preferentially located in the constitutive heterochromatin, which can be explained by the fact that heterochromatin is generally highly condensed and thus less accessible for DNA repair and replicative DNA synthesis. In contrast, in the open, more frequently transcribed euchromatin, DNA repair can efficiently correct uracils in the presence of a balanced dNTP pool. The low amount of genomic uracil in non-treated cells might remain from either cytosine deamination or thymine replacing misincorporation that escaped DNA repair. However, drug (5FdUR or RTX) treatments perturb the cellular nucleotide pool, and consequently highly increase the rate of thymine replacing uracil misincorporation events overwriting the background uracil pattern of non-treated cells’ genome. Uracil appearance via thymine replacing misincorporation implies prior DNA synthesis involved in either replication, or transcription-coupled DNA repair, or epigenetic reprogramming (e. g. erasing the methyl-cytosine epigenetic mark). Importantly, we found that uracil pattern showed the highest correlation with the features (early replication, active promoters and DNase hypersensitive sites, and CpG islands) linked exactly to these processes (Figure 4C). This is further supported by the fact that in MMR proficient drug-treated samples higher U-DNA content was measured as compared to the MMR deficient ones (Figure 1—figure supplement 1C–D). This observation in MMR proficient cells might be explained by either longer segments synthesized during the MMR process (Bowen et al., 2013), or the less tight control on cell cycle arrest (Figure 5) allowing more extended replicative synthesis.

Our data showing that under normal conditions, that is in lack of drug treatment, localization of human genomic uracils can be associated with the heterochromatic regions which is in agreement with the recent study by Shu et al., 2018. We propose that this pattern may reflect less efficient DNA repair in the heterochromatin. In accordance, it was shown that mutation rate within the later replicating heterochromatin is markedly increased (Stamatoyannopoulos et al., 2009). Interestingly, uracil distribution in bacterial and yeast genomes was found to be mostly excluded from the earliest as well as from the latest replicating segments (Bryan et al., 2014), suggesting a partially different pattern as compared to what is observed in human cells. In yeast cells it was also shown that transcription coupled repair synthesis might result in elevated uracil incorporation into actively transcribed regions under normal conditions (Kim and Jinks-Robertson, 2009; Owiti et al., 2018). However, both yeast and bacteria show major differences in both mechanisms of dNTP pool regulation (Mathews, 2014) and the set of available UDGs, as they do not encode the SMUG1 enzyme which is an important backup of the UNG2 in human (Elateri et al., 2003; Kavli et al., 2002). These differences may account for the alterations found in the genomic uracil distribution patterns.The antifolate or nucleotide-based thymidylate synthase inhibitors, such as 5-FU, RTX or 5FdUR are known to lead to cell cycle arrest, as it is confirmed in our experimental system (Figure 5) and is also reflected in the detected uracil-DNA pattern that strongly correlates with the early replicating segments in case of both drug treatments. The two drugs caused similar, but not equivalent uracil-DNA pattern. On the one hand, the correlations with the H3K36me3 marker as well as with the early replicating segments are both markedly stronger with the RTX treated sample as compared to the 5FdUR treated sample (Figure 4). On the other hand, the correlation of uracil accumulation with the H3K27me3 marker and with the CpG islands is stronger in the 5FdUR treated sample. Moreover, the MMR status has markedly different influence on the resulting U-DNA pattern in case of the two drugs (Figures 3–4). Such differences might correspond to drug-specific mechanism of action, involving alterations in signaling processes, transcription regulation and the timing of cell cycle arrest (Van Triest et al., 2000). Details of these mechanisms remain obscure in the literature. Still, it is well-known that both drugs inhibit thymidylate synthase thereby facilitating dUMP incorporation into DNA, while the nucleotide analogue 5FdUR also leads to direct incorporation of 5-fluorodeoxyuridine monophosphate (FdUP) into the DNA (Longley et al., 2003; Pettersen et al., 2011). Genomic uracil and fluorouracil might have different effects on transcription and epigenetic regulation processes that could also contribute to the observed differences of the two U-DNA patterns. It should be noted that our method detects both uracil and also fluorouracil within the DNA, since the UNG enzyme binds to fluorouracil as well (Pettersen et al., 2011). Phenotypic differences in cell cycle progression upon the two drug treatments were also reported. The 5FdUR treatment was shown to cause an S-phase arrest in the second cycle (Huehls et al., 2016; Yan et al., 2016), while the actual time point of cell cycle arrest upon RTX treatment is still controversial (Blackledge, 1998; Ding et al., 2019; Zhao et al., 2016). Similarly, we also detected slightly altered cell cycle distribution patterns in case of the two drug treatments, which were differently influenced by the MMR status (Figure 5). In case of the 5FdUR treatment, MMR proficiency seems to lead to a weaker S-phase arrest. This might correspond to the observed decrease in the correlation of U-DNA pattern and early replication timing (Figure 4C). However, equally induced DNA damage response (reported by γH2AX) was detected upon both drug treatments (Figure 5—figure supplement 1). Consistently with our observations, Weeks et al recently showed that treatment with the antifolate pemetrexed in UNG -/- human colon cancer cells led to preferential enrichment of double-strand breaks (DSBs) within highly accessible euchromatic regions, like transcription factor binding sites, origins of replication, DNase hypersensitivity regions and CpG islands (Weeks et al., 2014). This study did not directly address the occurrence of uracil moieties but caught the process initiated by uracil incorporation at a later stage. Still, the distribution pattern of the resulting DSBs showed similarities to our U-DNA-Seq data.

As we demonstrated here, the genome-wide uracil distribution patterns have relevance for example in case of drug-treated cancer cells. Therefore, besides the global U-DNA quantification methods (MS based [Galashevskaya et al., 2013], and dot blot [Róna et al., 2016]), NGS-based techniques also have high impact.The presented new method, termed U-DNA-Seq is a direct, feasible alternative to the recently published UPD-Seq (Sakhtemani et al., 2019), Excision-seq (Bryan et al., 2014) or dU-seq (Shu et al., 2018) methods, all of which rely on indirect detection requiring one or more auxiliary chemical or enzymatic step(s). Only these three methods have the potential thus far to map genome-wide distribution of uracil within isolated genomic DNA based on NGS, and only dU-seq was used in the context of human genome. One advantage of our U-DNA-Seq is that it is a direct method employing U-DNA specific binding of catalytically inactive UNG-derived sensor constructs to pull-down uracil-containing genomic DNA-fragments. In terms of resolution, only the pre-digestion Excision-seq was shown to be able to provide single-base resolution in case of smaller size genomes with high uracil content (Bryan et al., 2014). The resolution of other methods including our new U-DNA-Seq is limited by the fragment size of the DNA library. Importantly, single-base resolution of uracil positions has decreased relevance in most cases, considering the basically stochastic nature of uracil appearance either by incorporation as a result of drug-treatment-induced dNTP pool perturbations during DNA synthesis due to insensitivity of the polymerases, or by spontaneous cytosine deamination. Due to the stochastic processes, the actual positions of uracils are expected to be variable in every single cell. Therefore, a statistical approach has higher descriptive value about the uracil distribution in these cases. Accordingly, we constructed a novel computational pipeline (Figure 2 and Figure 2—figure supplement 1) that is suitable for the description of this kind of uracil distribution patterns. We also demonstrated that the usual analysis methods designed for ChIP-seq experiment are suboptimal in this case (Figure 3—figure supplement 2). Moreover, re-analysis of the earlier published dU-seq data (Appendix 1—table 1) with the herein developed pipeline, showed very high correlation with our U-DNA-Seq data in case of comparable samples (non-treated K562 cells in both cases; and 5FdUR treated UGI expressing HCT116 vs 5FdUR treated UNG^-/- HEK293T cells, Appendix 1—figure 1–2) confirming robustness and reliability of our method. However, our interpretation is markedly different regarding the preferential centromeric location of uracils that has been suggested by Shu et al., 2018. We analyzed the underlying reasons of this discrepancy in Appendix 1.

The new U-DNA-Seq method was shown to be reliable, robust and potent enough to gain systematic information on uracil-DNA metabolism upon drug treatments. Such information could essentially contribute to the future understanding of the mechanistic details either of cytotoxic effect induced by anti-cancer drugs, or other biological processes involving genomic uracil appearance. To this end, it is also of key importance to establish new visualization methods allowing colocalization measurements between U-DNA and other factors in highly complex eukaryotic cells.

Therefore, we further developed the U-DNA sensor to visualize genomic uracil in situ in human cells. The FLAG-ΔUNG-SNAP sensor construct and the optimized staining method presented here were successfully applied in confocal and super-resolution (STED or dSTORM) microscopies (see Figures 6–8). To our knowledge, there is no alternative technique published so far for in situ microscopic detection of mammalian genomic uracil. A recent paper was published reporting a similar approach, where uracil-DNA glycosylase UdgX was coupled to a fluorescent tag and applied for staining of uracils in E. coli DNA (Datta et al., 2019), however, in our previous study ΔUNG had already been proved to be potent for in situ uracil detection in the same organism (Róna et al., 2016). Still, the UdgX-based tool was not further extended for detection of uracils within the highly complex chromatin of human cells. Moreover, our detection method also allows simultaneous staining for other factors in colocalization experiments, potentially providing mechanistic insight into several important biological phenomena that involve uracil-DNA. For colocalization studies, two histone markers were selected based on the U-DNA-Seq results, namely H3K36me3 and H3K27me3, which were the strongest correlating factors for RTX_UGI and 5FdUR_UGI U-DNA patterns, respectively (Figure 4A–B). Using dSTORM super-resolution microscopy we could confirm significant correlation of genomic uracil with both selected histone markers in drug-treated (5FdUR or RTX), UGI-expressing cells (Figure 8). H3K36me3 was shown to associate with actively transcribed genes (Becker et al., 2017; Hyun et al., 2017; Pfister et al., 2014), while H3K27me3 is the most cited marker for facultative heterochromatin (Becker et al., 2017; Saksouk et al., 2015). Strikingly, we found that H3K27me3 shows even stronger colocalization with the U-DNA pattern in case of the RTX treated sample as compared to the 5FdUR treated one, which might be indicative for RTX treatment induced chromatin remodeling at least regarding this histone modification. It is important to note that our U-DNA-Seq was compared to published data that corresponds to ChIP-seq experiments performed in non-treated cells. However, during in situ cellular colocalization studies, the drug treatment is obviously applied to both patterns (i.e. U-DNA and histone marker). With the ChIP-seq experiment for H3K36me3 performed on both RTX treated and non-treated cells, we demonstrated that such treatment-induced chromatin remodeling is not a general phenomenon, but may rather confine to certain factors (Figure 4A–B, and Figure 4—figure supplement 1). Based on these observations, we can confirm that such in silico correlation studies has a predictive potential allowing qualitative characterization, and further independent techniques are required for detailed studies. In summary, co-staining of the selected histone markers and the genomic uracil in drug-treated cells via dSTORM reinforced the association between uracil occurrence and transcriptionally active regions.

It has been argued that uracil accumulation may play a more decisive role in genomic instability than the induced uracil-excision repair (Huehls et al., 2016; Yan et al., 2016). Uracil in DNA may therefore be used as a key marker for estimating efficiency of chemotherapeutic drugs targeting thymidylate biosynthesis. Our presented new techniques, namely the U-DNA-Seq and the related in situ U-DNA detection methods provide key insights into the mechanism of chemotherapeutic drugs. The combination of these methods might become a highly potent approach in the future, that is to investigate the complex pattern of intra-tumor heterogeneity that is closely related to cancer progression and drug-resistance (Stanta and Bonin, 2018), therefore may contribute to improving clinical practice.

Sequencing data have been deposited into the Gene Expression Omnibus (GEO) under accession number GSE126822 and GSE153407, which have been unified under SuperSeries GSE153408. In the following Genome Browser session, we included all the log2 coverage ratio (bigwig) and the derived uracil enriched interval (bed) files corresponding to this manuscript. The color code and the names are the same as used here. https://genome.ucsc.edu/s/bekesiangi/GSE126822_UCSC_Genome_Browser_session. Source data have been provided for Figure 1-figure supplement 1, Figure 2-figure supplement 2, Figure 3, Figure 3-figure supplement 4, Figure 4, Figure 4-figure supplement 3, Figure 8, Appendix 1-figure 1, Appendix 1-figure 2.

1. 1. Pálinkás HL
   2. Békési A
   3. Pongor L
   4. Holub E
   5. Papp G
   6. Gemma C
   7. Ali S
   8. Győrffy B
   9. Vértessy BG

   (2020) NCBI Gene Expression Omnibus

   ID GSE126822. Genome-wide alterations of uracil distribution patterns in human DNA upon chemotherapeutic treatments.

   https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE126822
2. 1. Pálinkás HL
   2. Békési A
   3. Vértessy BG

   (2020) NCBI Gene Expression Omnibus

   ID GSE153407. H3K36me3 ChIP-seq in non-treated and raltitrexed treated UGI-expressing HCT116 cells.

   https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE153407

1. 1. Shu X
   2. Lu Z
   3. Yi C

   (2018) NCBI Gene Expression Omnibus

   ID GSE99011. Genome-wide mapping reveals that deoxyuridine is enriched in the human centromeric DNA.

   https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE99011

1. 1. Amemiya HM
   2. Kundaje A
   3. Boyle AP

   (2019) The ENCODE blacklist: identification of problematic regions of the genome

   Scientific Reports 9:9354.

   https://doi.org/10.1038/s41598-019-45839-z

   • PubMed
   • Google Scholar
2. 1. Becker JS
   2. McCarthy RL
   3. Sidoli S
   4. Donahue G
   5. Kaeding KE
   6. He Z
   7. Lin S
   8. Garcia BA
   9. Zaret KS

   (2017) Genomic and proteomic resolution of heterochromatin and its restriction of alternate fate genes

   Molecular Cell 68:1023–1037.

   https://doi.org/10.1016/j.molcel.2017.11.030

   • PubMed
   • Google Scholar
3. 1. Bermudez-Hernandez K
   2. Keegan S
   3. Whelan DR
   4. Reid DA
   5. Zagelbaum J
   6. Yin Y
   7. Ma S
   8. Rothenberg E
   9. Fenyö D

   (2017) A method for quantifying molecular interactions using stochastic modelling and Super-Resolution microscopy

   Scientific Reports 7:1–13.

   https://doi.org/10.1038/s41598-017-14922-8

   • PubMed
   • Google Scholar
4. 1. Black JC
   2. Van Rechem C
   3. Whetstine JR

   (2012) Histone lysine methylation dynamics: establishment, regulation, and biological impact

   Molecular Cell 48:491–507.

   https://doi.org/10.1016/j.molcel.2012.11.006

   • PubMed
   • Google Scholar
5. 1. Blackledge G

   (1998) New developments in Cancer treatment with the novel thymidylate synthase inhibitor raltitrexed ('Tomudex')

   British Journal of Cancer 77 Suppl 2:29–37.

   https://doi.org/10.1038/bjc.1998.423

   • PubMed
   • Google Scholar
6. 1. Boissinot S
   2. Furano AV

   (2005) The recent evolution of human L1 retrotransposons

   Cytogenetic and Genome Research 110:402–406.

   https://doi.org/10.1159/000084972

   • PubMed
   • Google Scholar
7. 1. Bowen N
   2. Smith CE
   3. Srivatsan A
   4. Willcox S
   5. Griffith JD
   6. Kolodner RD

   (2013) Reconstitution of long and short patch mismatch repair reactions using Saccharomyces cerevisiae proteins

   PNAS 110:18472–18477.

   https://doi.org/10.1073/pnas.1318971110

   • PubMed
   • Google Scholar
8. 1. Bryan DS
   2. Ransom M
   3. Adane B
   4. York K
   5. Hesselberth JR

   (2014) High resolution mapping of modified DNA nucleobases using excision repair enzymes

   Genome Research 24:1534–1542.

   https://doi.org/10.1101/gr.174052.114

   • PubMed
   • Google Scholar
9. 1. Burns MB
   2. Leonard B
   3. Harris RS

   (2015) APOBEC3B: pathological consequences of an innate immune DNA mutator

   Biomedical Journal 38:102–110.

   https://doi.org/10.4103/2319-4170.148904

   • PubMed
   • Google Scholar
10. 1. Buske OJ
    2. Hoffman MM
    3. Ponts N
    4. Le Roch KG
    5. Noble WS

    (2011) Exploratory analysis of genomic segmentations with segtools

    BMC Bioinformatics 12:415.

    https://doi.org/10.1186/1471-2105-12-415

    • PubMed
    • Google Scholar
11. 1. Chan RCW
    2. Libbrecht MW
    3. Roberts EG
    4. Bilmes JA
    5. Noble WS
    6. Hoffman MM

    (2018) Segway 2.0: gaussian mixture models and minibatch training

    Bioinformatics 34:669–671.

    https://doi.org/10.1093/bioinformatics/btx603

    • PubMed
    • Google Scholar
12. 1. Coltharp C
    2. Yang X
    3. Xiao J

    (2014) Quantitative analysis of single-molecule superresolution images

    Current Opinion in Structural Biology 28:112–121.

    https://doi.org/10.1016/j.sbi.2014.08.008

    • PubMed
    • Google Scholar
13. 1. Creyghton MP
    2. Cheng AW
    3. Welstead GG
    4. Kooistra T
    5. Carey BW
    6. Steine EJ
    7. Hanna J
    8. Lodato MA
    9. Frampton GM
    10. Sharp PA
    11. Boyer LA
    12. Young RA
    13. Jaenisch R

    (2010) Histone H3K27ac separates active from poised enhancers and predicts developmental state

    PNAS 107:21931–21936.

    https://doi.org/10.1073/pnas.1016071107

    • PubMed
    • Google Scholar
14. 1. Cutler PJ
    2. Malik MD
    3. Liu S
    4. Byars JM
    5. Lidke DS
    6. Lidke KA

    (2013) Multi-color quantum dot tracking using a high-speed hyperspectral line-scanning microscope

    PLOS ONE 8:e64320.

    https://doi.org/10.1371/journal.pone.0064320

    • PubMed
    • Google Scholar
15. 1. Datta M
    2. Aroli S
    3. Karmakar K
    4. Dutta S
    5. Chakravortty D
    6. Varshney U

    (2019) Development of mCherry tagged UdgX as a highly sensitive molecular probe for specific detection of uracils in DNA

    Biochemical and Biophysical Research Communications 518:38–43.

    https://doi.org/10.1016/j.bbrc.2019.08.005

    • PubMed
    • Google Scholar
16. 1. Ding WX
    2. Liu S
    3. Ma JX
    4. Pu J
    5. Wang HJ
    6. Zhang S
    7. Sun XC

    (2019) Raltitrexed increases radiation sensitivity of esophageal squamous carcinoma cells

    Cancer Cell International 19:36.

    https://doi.org/10.1186/s12935-019-0752-y

    • PubMed
    • Google Scholar
17. 1. Doetzlhofer A
    2. Rotheneder H
    3. Lagger G
    4. Koranda M
    5. Kurtev V
    6. Brosch G
    7. Wintersberger E
    8. Seiser C

    (1999) Histone deacetylase 1 can repress transcription by binding to Sp1

    Molecular and Cellular Biology 19:5504–5511.

    https://doi.org/10.1128/MCB.19.8.5504

    • PubMed
    • Google Scholar
18. 1. Elateri I
    2. Tinkelenberg BA
    3. Hansbury M
    4. Caradonna S
    5. Muller-Weeks S
    6. Ladner RD

    (2003) hSMUG1 can functionally compensate for Ung1 in the yeast Saccharomyces cerevisiae

    DNA Repair 2:315–323.

    https://doi.org/10.1016/S1568-7864(02)00221-5

    • PubMed
    • Google Scholar
19. 1. ENCODE Project Consortium

    (2012) An integrated encyclopedia of DNA elements in the human genome

    Nature 489:57–74.

    https://doi.org/10.1038/nature11247

    • PubMed
    • Google Scholar
20. 1. Feng J
    2. Liu T
    3. Zhang Y

    (2011) Using MACS to identify peaks from ChIP-Seq data

    Current Protocols in Bioinformatics Chapter 2:2.14.1–2.14.2.

    https://doi.org/10.1002/0471250953.bi0214s34

    • PubMed
    • Google Scholar
21. 1. Galashevskaya A
    2. Sarno A
    3. Vågbø CB
    4. Aas PA
    5. Hagen L
    6. Slupphaug G
    7. Krokan HE

    (2013) A robust, sensitive assay for genomic uracil determination by LC/MS/MS reveals lower levels than previously reported

    DNA Repair 12:699–706.

    https://doi.org/10.1016/j.dnarep.2013.05.002

    • PubMed
    • Google Scholar
22. 1. Gates LA
    2. Shi J
    3. Rohira AD
    4. Feng Q
    5. Zhu B
    6. Bedford MT
    7. Sagum CA
    8. Jung SY
    9. Qin J
    10. Tsai MJ
    11. Tsai SY
    12. Li W
    13. Foulds CE
    14. O'Malley BW

    (2017) Acetylation on histone H3 lysine 9 mediates a switch from transcription initiation to elongation

    Journal of Biological Chemistry 292:14456–14472.

    https://doi.org/10.1074/jbc.M117.802074

    • PubMed
    • Google Scholar
23. 1. Germano G
    2. Amirouchene-Angelozzi N
    3. Rospo G
    4. Bardelli A

    (2018) The clinical impact of the genomic landscape of mismatch Repair-Deficient cancers

    Cancer Discovery 8:1518–1528.

    https://doi.org/10.1158/2159-8290.CD-18-0150

    • PubMed
    • Google Scholar
24. 1. Giaimo BD
    2. Ferrante F
    3. Herchenröther A
    4. Hake SB
    5. Borggrefe T

    (2019) The histone variant H2A.Z in gene regulation

    Epigenetics & Chromatin 12:37.

    https://doi.org/10.1186/s13072-019-0274-9

    • PubMed
    • Google Scholar
25. 1. Gilbert DM

    (2002) Replication timing and transcriptional control: beyond cause and effect

    Current Opinion in Cell Biology 14:377–383.

    https://doi.org/10.1016/S0955-0674(02)00326-5

    • PubMed
    • Google Scholar
26. 1. Guo Y
    2. Dai Y
    3. Yu H
    4. Zhao S
    5. Samuels DC
    6. Shyr Y

    (2017) Improvements and impacts of GRCh38 human reference on high throughput sequencing data analysis

    Genomics 109:83–90.

    https://doi.org/10.1016/j.ygeno.2017.01.005

    • PubMed
    • Google Scholar
27. 1. Gupta D
    2. Heinen CD

    (2019) The mismatch repair-dependent DNA damage response: mechanisms and implications

    DNA Repair 78:60–69.

    https://doi.org/10.1016/j.dnarep.2019.03.009

    • PubMed
    • Google Scholar
28. 1. Hayden KE
    2. Strome ED
    3. Merrett SL
    4. Lee HR
    5. Rudd MK
    6. Willard HF

    (2013) Sequences associated with centromere competency in the human genome

    Molecular and Cellular Biology 33:763–772.

    https://doi.org/10.1128/MCB.01198-12

    • PubMed
    • Google Scholar
29. 1. Ho TH
    2. Park IY
    3. Zhao H
    4. Tong P
    5. Champion MD
    6. Yan H
    7. Monzon FA
    8. Hoang A
    9. Tamboli P
    10. Parker AS
    11. Joseph RW
    12. Qiao W
    13. Dykema K
    14. Tannir NM
    15. Castle EP
    16. Nunez-Nateras R
    17. Teh BT
    18. Wang J
    19. Walker CL
    20. Hung MC
    21. Jonasch E

    (2016) High-resolution profiling of histone h3 lysine 36 trimethylation in metastatic renal cell carcinoma

    Oncogene 35:1565–1574.

    https://doi.org/10.1038/onc.2015.221

    • PubMed
    • Google Scholar
30. 1. Hoffman MM
    2. Buske OJ
    3. Wang J
    4. Weng Z
    5. Bilmes JA
    6. Noble WS

    (2012) Unsupervised pattern discovery in human chromatin structure through genomic segmentation

    Nature Methods 9:473–476.

    https://doi.org/10.1038/nmeth.1937

    • PubMed
    • Google Scholar
31. 1. Holmquist G
    2. Gray M
    3. Porter T
    4. Jordan J

    (1982) Characterization of giemsa dark- and light-band DNA

    Cell 31:121–129.

    https://doi.org/10.1016/0092-8674(82)90411-1

    • PubMed
    • Google Scholar
32. 1. Horváth A
    2. Békési A
    3. Muha V
    4. Erdélyi M
    5. Vértessy BG

    (2013) Expanding the DNA alphabet in the fruit fly: uracil enrichment in genomic DNA

    Fly 7:23–27.

    https://doi.org/10.4161/fly.23192

    • PubMed
    • Google Scholar
33. 1. Horváth A
    2. Vértessy BG

    (2010) A one-step method for quantitative determination of uracil in DNA by real-time PCR

    Nucleic Acids Research 38:e196.

    https://doi.org/10.1093/nar/gkq815

    • PubMed
    • Google Scholar
34. 1. Huehls AM
    2. Huntoon CJ
    3. Joshi PM
    4. Baehr CA
    5. Wagner JM
    6. Wang X
    7. Lee MY
    8. Karnitz LM

    (2016) Genomically incorporated 5-Fluorouracil that escapes UNG-Initiated base excision repair blocks DNA replication and activates homologous recombination

    Molecular Pharmacology 89:53–62.

    https://doi.org/10.1124/mol.115.100164

    • PubMed
    • Google Scholar
35. 1. Hunter JD

    (2007) Matplotlib: a 2D graphics environment

    Computing in Science & Engineering 9:90–95.

    https://doi.org/10.1109/MCSE.2007.55

    • Google Scholar
36. 1. Hyun K
    2. Jeon J
    3. Park K
    4. Kim J

    (2017) Writing, erasing and reading histone lysine methylations

    Experimental & Molecular Medicine 49:e324.

    https://doi.org/10.1038/emm.2017.11

    • PubMed
    • Google Scholar
37. 1. Jensen MA
    2. Ferretti V
    3. Grossman RL
    4. Staudt LM

    (2017) The NCI genomic data commons as an engine for precision medicine

    Blood 130:453–459.

    https://doi.org/10.1182/blood-2017-03-735654

    • PubMed
    • Google Scholar
38. 1. Karolchik D
    2. Hinrichs AS
    3. Furey TS
    4. Roskin KM
    5. Sugnet CW
    6. Haussler D
    7. Kent WJ

    (2004) The UCSC table browser data retrieval tool

    Nucleic Acids Research 32:D493–D496.

    https://doi.org/10.1093/nar/gkh103

    • PubMed
    • Google Scholar
39. 1. Kavli B
    2. Sundheim O
    3. Akbari M
    4. Otterlei M
    5. Nilsen H
    6. Skorpen F
    7. Aas PA
    8. Hagen L
    9. Krokan HE
    10. Slupphaug G

    (2002) hUNG2 is the major repair enzyme for removal of uracil from U:a matches, U:g mismatches, and U in single-stranded DNA, with hSMUG1 as a broad specificity backup

    Journal of Biological Chemistry 277:39926–39936.

    https://doi.org/10.1074/jbc.M207107200

    • PubMed
    • Google Scholar
40. 1. Keppler A
    2. Gendreizig S
    3. Gronemeyer T
    4. Pick H
    5. Vogel H
    6. Johnsson K

    (2003) A general method for the covalent labeling of fusion proteins with small molecules in vivo

    Nature Biotechnology 21:86–89.

    https://doi.org/10.1038/nbt765

    • PubMed
    • Google Scholar
41. 1. Khodursky A
    2. Guzmán EC
    3. Hanawalt PC

    (2015) Thymineless death lives on: new insights into a classic phenomenon

    Annual Review of Microbiology 69:247–263.

    https://doi.org/10.1146/annurev-micro-092412-155749

    • PubMed
    • Google Scholar
42. 1. Kim N
    2. Jinks-Robertson S

    (2009) dUTP incorporation into genomic DNA is linked to transcription in yeast

    Nature 459:1150–1153.

    https://doi.org/10.1038/nature08033

    • PubMed
    • Google Scholar
43. 1. Koi M
    2. Umar A
    3. Chauhan DP
    4. Cherian SP
    5. Carethers JM
    6. Kunkel TA
    7. Boland CR

    (1994)

    Human chromosome 3 corrects mismatch repair deficiency and microsatellite instability and reduces N-Methyl-N’-nitro-N-nitrosoguanidine Tolerance in Colon Tumor Cells with Homozygous hMLHl Mutation

    Cancer Research 54:4308–4312.

    • PubMed
    • Google Scholar
44. 1. Kovalskaya E
    2. Buzdin A
    3. Gogvadze E
    4. Vinogradova T
    5. Sverdlov E

    (2006) Functional human endogenous retroviral LTR transcription start sites are located between the R and U5 regions

    Virology 346:373–378.

    https://doi.org/10.1016/j.virol.2005.11.007

    • PubMed
    • Google Scholar
45. 1. Kramerov DA
    2. Vassetzky NS

    (2005) Short retroposons in eukaryotic genomes

    International Review of Cytology 247:165–221.

    https://doi.org/10.1016/S0074-7696(05)47004-7

    • PubMed
    • Google Scholar
46. 1. Krokan HE
    2. Drabløs F
    3. Slupphaug G

    (2002) Uracil in DNA--occurrence, consequences and repair

    Oncogene 21:8935–8948.

    https://doi.org/10.1038/sj.onc.1205996

    • PubMed
    • Google Scholar
47. 1. Krokan HE
    2. Bjørås M

    (2013) Base excision repair

    Cold Spring Harbor Perspectives in Biology 5:a012583.

    https://doi.org/10.1101/cshperspect.a012583

    • PubMed
    • Google Scholar
48. 1. Kuhn RM
    2. Haussler D
    3. Kent WJ

    (2013) The UCSC genome browser and associated tools

    Briefings in Bioinformatics 14:144–161.

    https://doi.org/10.1093/bib/bbs038

    • PubMed
    • Google Scholar
49. 1. Layer RM
    2. Pedersen BS
    3. DiSera T
    4. Marth GT
    5. Gertz J
    6. Quinlan AR

    (2018) GIGGLE: a search engine for large-scale integrated genome analysis

    Nature Methods 15:123–126.

    https://doi.org/10.1038/nmeth.4556

    • PubMed
    • Google Scholar
50. 1. Li H
    2. Handsaker B
    3. Wysoker A
    4. Fennell T
    5. Ruan J
    6. Homer N
    7. Marth G
    8. Abecasis G
    9. Durbin R
    10. 1000 Genome Project Data Processing Subgroup

    (2009) The sequence alignment/Map format and SAMtools

    Bioinformatics 25:2078–2079.

    https://doi.org/10.1093/bioinformatics/btp352

    • PubMed
    • Google Scholar
51. 1. Li H
    2. Durbin R

    (2010) Fast and accurate long-read alignment with Burrows-Wheeler transform

    Bioinformatics 26:589–595.

    https://doi.org/10.1093/bioinformatics/btp698

    • PubMed
    • Google Scholar
52. 1. Liu M
    2. Schatz DG

    (2009) Balancing AID and DNA repair during somatic hypermutation

    Trends in Immunology 30:173–181.

    https://doi.org/10.1016/j.it.2009.01.007

    • PubMed
    • Google Scholar
53. 1. Longley DB
    2. Harkin DP
    3. Johnston PG

    (2003) 5-fluorouracil: mechanisms of action and clinical strategies

    Nature Reviews Cancer 3:330–338.

    https://doi.org/10.1038/nrc1074

    • PubMed
    • Google Scholar
54. 1. López-Flores I
    2. Garrido-Ramos MA

    (2012) The repetitive DNA content of eukaryotic genomes

    Genome Dynamics 7:1–28.

    https://doi.org/10.1159/000337118

    • PubMed
    • Google Scholar
55. 1. Luo Y
    2. Walla M
    3. Wyatt MD

    (2008) Uracil incorporation into genomic DNA does not predict toxicity caused by chemotherapeutic inhibition of thymidylate synthase

    DNA Repair 7:162–169.

    https://doi.org/10.1016/j.dnarep.2007.09.001

    • PubMed
    • Google Scholar
56. 1. Marchal C
    2. Sasaki T
    3. Vera D
    4. Wilson K
    5. Sima J
    6. Rivera-Mulia JC
    7. Trevilla-García C
    8. Nogues C
    9. Nafie E
    10. Gilbert DM

    (2018) Genome-wide analysis of replication timing by next-generation sequencing with E/L Repli-seq

    Nature Protocols 13:819–839.

    https://doi.org/10.1038/nprot.2017.148

    • PubMed
    • Google Scholar
57. 1. Mathews CK

    (2014) Deoxyribonucleotides as genetic and metabolic regulators

    The FASEB Journal 28:3832–3840.

    https://doi.org/10.1096/fj.14-251249

    • PubMed
    • Google Scholar
58. 1. Maul RW
    2. Gearhart PJ

    (2010) AID and somatic hypermutation

    Advances in Immunology 105:159–191.

    https://doi.org/10.1016/S0065-2776(10)05006-6

    • PubMed
    • Google Scholar
59. 1. Maul RW
    2. Gearhart PJ

    (2014) Refining the neuberger model: uracil processing by activated B cells

    European Journal of Immunology 44:1913–1916.

    https://doi.org/10.1002/eji.201444813

    • PubMed
    • Google Scholar
60. 1. Mei S
    2. Qin Q
    3. Wu Q
    4. Sun H
    5. Zheng R
    6. Zang C
    7. Zhu M
    8. Wu J
    9. Shi X
    10. Taing L
    11. Liu T
    12. Brown M
    13. Meyer CA
    14. Liu XS

    (2017) Cistrome data browser: a data portal for ChIP-Seq and chromatin accessibility data in human and mouse

    Nucleic Acids Research 45:D658–D662.

    https://doi.org/10.1093/nar/gkw983

    • PubMed
    • Google Scholar
61. 1. Meyers M
    2. Wagner MW
    3. Hwang HS
    4. Kinsella TJ
    5. Boothman DA

    (2001)

    Role of the hMLH1 DNA mismatch repair protein in fluoropyrimidine-mediated cell death and cell cycle responses

    Cancer Research 61:5193–5201.

    • PubMed
    • Google Scholar
62. 1. Miga KH
    2. Newton Y
    3. Jain M
    4. Altemose N
    5. Willard HF
    6. Kent WJ

    (2014) Centromere reference models for human chromosomes X and Y satellite arrays

    Genome Research 24:697–707.

    https://doi.org/10.1101/gr.159624.113

    • PubMed
    • Google Scholar
63. 1. Mol CD
    2. Arvai AS
    3. Sanderson RJ
    4. Slupphaug G
    5. Kavli B
    6. Krokan HE
    7. Mosbaugh DW
    8. Tainer JA

    (1995) Crystal structure of human uracil-DNA glycosylase in complex with a protein inhibitor: protein mimicry of DNA

    Cell 82:701–708.

    https://doi.org/10.1016/0092-8674(95)90467-0

    • PubMed
    • Google Scholar
64. 1. Molnár P
    2. Marton L
    3. Izrael R
    4. Pálinkás HL
    5. Vértessy BG

    (2018) Uracil moieties in Plasmodium falciparum genomic DNA

    FEBS Open Bio 8:1763–1772.

    https://doi.org/10.1002/2211-5463.12458

    • PubMed
    • Google Scholar
65. 1. Muha V
    2. Horváth A
    3. Békési A
    4. Pukáncsik M
    5. Hodoscsek B
    6. Merényi G
    7. Róna G
    8. Batki J
    9. Kiss I
    10. Jankovics F
    11. Vilmos P
    12. Erdélyi M
    13. Vértessy BG

    (2012) Uracil-containing DNA in Drosophila: stability, stage-specific accumulation, and developmental involvement

    PLOS Genetics 8:e1002738.

    https://doi.org/10.1371/journal.pgen.1002738

    • PubMed
    • Google Scholar
66. 1. Ostrer L
    2. Hamann BL
    3. Khodursky A

    (2015) Perturbed states of the bacterial chromosome: a thymineless death case study

    Frontiers in Microbiology 6:363.

    https://doi.org/10.3389/fmicb.2015.00363

    • PubMed
    • Google Scholar
67. 1. Owiti N
    2. Wei S
    3. Bhagwat AS
    4. Kim N

    (2018) Unscheduled DNA synthesis leads to elevated uracil residues at highly transcribed genomic loci in Saccharomyces cerevisiae

    PLOS Genetics 14:e1007516.

    https://doi.org/10.1371/journal.pgen.1007516

    • PubMed
    • Google Scholar
68. 1. Pettersen HS
    2. Visnes T
    3. Vågbø CB
    4. Svaasand EK
    5. Doseth B
    6. Slupphaug G
    7. Kavli B
    8. Krokan HE

    (2011) UNG-initiated base excision repair is the major repair route for 5-fluorouracil in DNA, but 5-fluorouracil cytotoxicity depends mainly on RNA incorporation

    Nucleic Acids Research 39:8430–8444.

    https://doi.org/10.1093/nar/gkr563

    • PubMed
    • Google Scholar
69. 1. Pfister SX
    2. Ahrabi S
    3. Zalmas LP
    4. Sarkar S
    5. Aymard F
    6. Bachrati CZ
    7. Helleday T
    8. Legube G
    9. La Thangue NB
    10. Porter AC
    11. Humphrey TC

    (2014) SETD2-dependent histone H3K36 trimethylation is required for homologous recombination repair and genome stability

    Cell Reports 7:2006–2018.

    https://doi.org/10.1016/j.celrep.2014.05.026

    • PubMed
    • Google Scholar
70. 1. Qin Q
    2. Mei S
    3. Wu Q
    4. Sun H
    5. Li L
    6. Taing L
    7. Chen S
    8. Li F
    9. Liu T
    10. Zang C
    11. Xu H
    12. Chen Y
    13. Meyer CA
    14. Zhang Y
    15. Brown M
    16. Long HW
    17. Liu XS

    (2016) ChiLin: a comprehensive ChIP-seq and DNase-seq quality control and analysis pipeline

    BMC Bioinformatics 17:404.

    https://doi.org/10.1186/s12859-016-1274-4

    • PubMed
    • Google Scholar
71. 1. Quinlan AR
    2. Hall IM

    (2010) BEDTools: a flexible suite of utilities for comparing genomic features

    Bioinformatics 26:841–842.

    https://doi.org/10.1093/bioinformatics/btq033

    • PubMed
    • Google Scholar
72. 1. Ramírez F
    2. Ryan DP
    3. Grüning B
    4. Bhardwaj V
    5. Kilpert F
    6. Richter AS
    7. Heyne S
    8. Dündar F
    9. Manke T

    (2016) deepTools2: a next generation web server for deep-sequencing data analysis

    Nucleic Acids Research 44:W160–W165.

    https://doi.org/10.1093/nar/gkw257

    • PubMed
    • Google Scholar
73. 1. Rashid S
    2. Freitas MO
    3. Cucchi D
    4. Bridge G
    5. Yao Z
    6. Gay L
    7. Williams M
    8. Wang J
    9. Suraweera N
    10. Silver A
    11. McDonald SAC
    12. Chelala C
    13. Szabadkai G
    14. Martin SA

    (2019) MLH1 deficiency leads to deregulated mitochondrial metabolism

    Cell Death & Disease 10:795.

    https://doi.org/10.1038/s41419-019-2018-y

    • PubMed
    • Google Scholar
74. 1. Requena CE
    2. Pérez-Moreno G
    3. Horváth A
    4. Vértessy BG
    5. Ruiz-Pérez LM
    6. González-Pacanowska D
    7. Vidal AE

    (2016) The nucleotidohydrolases DCTPP1 and dUTPase are involved in the cellular response to decitabine

    The Biochemical Journal 473:2635–2643.

    https://doi.org/10.1042/BCJ20160302

    • PubMed
    • Google Scholar
75. 1. Róna G
    2. Scheer I
    3. Nagy K
    4. Pálinkás HL
    5. Tihanyi G
    6. Borsos M
    7. Békési A
    8. Vértessy BG

    (2016) Detection of uracil within DNA using a sensitive labeling method for in vitro and cellular applications

    Nucleic Acids Research 44:e28.

    https://doi.org/10.1093/nar/gkv977

    • PubMed
    • Google Scholar
76. 1. Rona G
    2. Roberti D
    3. Yin Y
    4. Pagan JK
    5. Homer H
    6. Sassani E
    7. Zeke A
    8. Busino L
    9. Rothenberg E
    10. Pagano M

    (2018) PARP1-dependent recruitment of the FBXL10-RNF68-RNF2 ubiquitin ligase to sites of DNA damage controls H2A.Z loading

    eLife 7:e38771.

    https://doi.org/10.7554/eLife.38771

    • PubMed
    • Google Scholar
77. 1. Sakhtemani R
    2. Senevirathne V
    3. Stewart J
    4. Perera MLW
    5. Pique-Regi R
    6. Lawrence MS
    7. Bhagwat AS

    (2019) Genome-wide mapping of regions preferentially targeted by the human DNA-cytosine deaminase APOBEC3A using uracil-DNA pulldown and sequencing

    Journal of Biological Chemistry 294:15037–15051.

    https://doi.org/10.1074/jbc.RA119.008053

    • PubMed
    • Google Scholar
78. 1. Saksouk N
    2. Simboeck E
    3. Déjardin J

    (2015) Constitutive heterochromatin formation and transcription in mammals

    Epigenetics & Chromatin 8:3.

    https://doi.org/10.1186/1756-8935-8-3

    • PubMed
    • Google Scholar
79. 1. Sekine S
    2. Mori T
    3. Ogawa R
    4. Tanaka M
    5. Yoshida H
    6. Taniguchi H
    7. Nakajima T
    8. Sugano K
    9. Yoshida T
    10. Kato M
    11. Furukawa E
    12. Ochiai A
    13. Hiraoka N

    (2017) Mismatch repair deficiency commonly precedes adenoma formation in lynch Syndrome-Associated colorectal tumorigenesis

    Modern Pathology 30:1144–1151.

    https://doi.org/10.1038/modpathol.2017.39

    • PubMed
    • Google Scholar
80. 1. Shu X
    2. Liu M
    3. Lu Z
    4. Zhu C
    5. Meng H
    6. Huang S
    7. Zhang X
    8. Yi C

    (2018) Genome-wide mapping reveals that deoxyuridine is enriched in the human centromeric DNA

    Nature Chemical Biology 14:680–687.

    https://doi.org/10.1038/s41589-018-0065-9

    • PubMed
    • Google Scholar
81. 1. Stamatoyannopoulos JA
    2. Adzhubei I
    3. Thurman RE
    4. Kryukov GV
    5. Mirkin SM
    6. Sunyaev SR

    (2009) Human mutation rate associated with DNA replication timing

    Nature Genetics 41:393–395.

    https://doi.org/10.1038/ng.363

    • PubMed
    • Google Scholar
82. 1. Stanta G
    2. Bonin S

    (2018) Overview on clinical relevance of Intra-Tumor heterogeneity

    Frontiers in Medicine 5:85.

    https://doi.org/10.3389/fmed.2018.00085

    • PubMed
    • Google Scholar
83. 1. Stenglein MD
    2. Burns MB
    3. Li M
    4. Lengyel J
    5. Harris RS

    (2010) APOBEC3 proteins mediate the clearance of foreign DNA from human cells

    Nature Structural & Molecular Biology 17:222–229.

    https://doi.org/10.1038/nsmb.1744

    • PubMed
    • Google Scholar
84. 1. Suspène R
    2. Henry M
    3. Guillot S
    4. Wain-Hobson S
    5. Vartanian JP

    (2005) Recovery of APOBEC3-edited human immunodeficiency virus G->A hypermutants by differential DNA denaturation PCR

    Journal of General Virology 86:125–129.

    https://doi.org/10.1099/vir.0.80426-0

    • PubMed
    • Google Scholar
85. 1. Thorvaldsdóttir H
    2. Robinson JT
    3. Mesirov JP

    (2013) Integrative genomics viewer (IGV): high-performance genomics data visualization and exploration

    Briefings in Bioinformatics 14:178–192.

    https://doi.org/10.1093/bib/bbs017

    • PubMed
    • Google Scholar
86. 1. Van Triest B
    2. Pinedo HM
    3. Giaccone G
    4. Peters GJ

    (2000) Downstream molecular determinants of response to 5-fluorouracil and antifolate thymidylate synthase inhibitors

    Annals of Oncology 11:385–391.

    https://doi.org/10.1023/A:1008351221345

    • PubMed
    • Google Scholar
87. 1. Veatch SL
    2. Machta BB
    3. Shelby SA
    4. Chiang EN
    5. Holowka DA
    6. Baird BA

    (2012) Correlation functions quantify super-resolution images and estimate apparent clustering due to over-counting

    PLOS ONE 7:e31457.

    https://doi.org/10.1371/journal.pone.0031457

    • PubMed
    • Google Scholar
88. 1. Vértessy BG
    2. Tóth J

    (2009) Keeping uracil out of DNA: physiological role, structure and catalytic mechanism of dUTPases

    Accounts of Chemical Research 42:97–106.

    https://doi.org/10.1021/ar800114w

    • PubMed
    • Google Scholar
89. 1. Visnes T
    2. Doseth B
    3. Pettersen HS
    4. Hagen L
    5. Sousa MM
    6. Akbari M
    7. Otterlei M
    8. Kavli B
    9. Slupphaug G
    10. Krokan HE

    (2009) Uracil in DNA and its processing by different DNA glycosylases

    Philosophical Transactions of the Royal Society B: Biological Sciences 364:563–568.

    https://doi.org/10.1098/rstb.2008.0186

    • PubMed
    • Google Scholar
90. 1. Wallace SS

    (2014) Base excision repair: a critical player in many games

    DNA Repair 19:14–26.

    https://doi.org/10.1016/j.dnarep.2014.03.030

    • PubMed
    • Google Scholar
91. 1. Weddington N
    2. Stuy A
    3. Hiratani I
    4. Ryba T
    5. Yokochi T
    6. Gilbert DM

    (2008) ReplicationDomain: a visualization tool and comparative database for genome-wide replication timing data

    BMC Bioinformatics 9:530.

    https://doi.org/10.1186/1471-2105-9-530

    • PubMed
    • Google Scholar
92. 1. Weeks LD
    2. Zentner GE
    3. Scacheri PC
    4. Gerson SL

    (2014) Uracil DNA glycosylase (UNG) loss enhances DNA double strand break formation in human Cancer cells exposed to pemetrexed

    Cell Death & Disease 5:e1045.

    https://doi.org/10.1038/cddis.2013.477

    • PubMed
    • Google Scholar
93. 1. Whelan DR
    2. Lee WTC
    3. Yin Y
    4. Ofri DM
    5. Bermudez-Hernandez K
    6. Keegan S
    7. Fenyo D
    8. Rothenberg E

    (2018) Spatiotemporal dynamics of homologous recombination repair at single collapsed replication forks

    Nature Communications 9:3882.

    https://doi.org/10.1038/s41467-018-06435-3

    • PubMed
    • Google Scholar
94. 1. Wilson PM
    2. Danenberg PV
    3. Johnston PG
    4. Lenz HJ
    5. Ladner RD

    (2014) Standing the test of time: targeting thymidylate biosynthesis in Cancer therapy

    Nature Reviews Clinical Oncology 11:282–298.

    https://doi.org/10.1038/nrclinonc.2014.51

    • PubMed
    • Google Scholar
95. 1. Xu Z
    2. Zan H
    3. Pone EJ
    4. Mai T
    5. Casali P

    (2012) Immunoglobulin class-switch DNA recombination: induction, targeting and beyond

    Nature Reviews Immunology 12:517–531.

    https://doi.org/10.1038/nri3216

    • PubMed
    • Google Scholar
96. 1. Yan Y
    2. Han X
    3. Qing Y
    4. Condie AG
    5. Gorityala S
    6. Yang S
    7. Xu Y
    8. Zhang Y
    9. Gerson SL

    (2016) Inhibition of uracil DNA glycosylase sensitizes Cancer cells to 5-fluorodeoxyuridine through replication fork collapse-induced DNA damage

    Oncotarget 7:59299–59313.

    https://doi.org/10.18632/oncotarget.11151

    • PubMed
    • Google Scholar
97. 1. Yin Y
    2. Rothenberg E

    (2016) Probing the spatial organization of molecular complexes using Triple-Pair-Correlation

    Scientific Reports 6:30819.

    https://doi.org/10.1038/srep30819

    • PubMed
    • Google Scholar
98. 1. Zerbino DR
    2. Achuthan P
    3. Akanni W
    4. Amode MR
    5. Barrell D
    6. Bhai J
    7. Billis K
    8. Cummins C
    9. Gall A
    10. Girón CG
    11. Gil L
    12. Gordon L
    13. Haggerty L
    14. Haskell E
    15. Hourlier T
    16. Izuogu OG
    17. Janacek SH
    18. Juettemann T
    19. To JK
    20. Laird MR
    21. Lavidas I
    22. Liu Z
    23. Loveland JE
    24. Maurel T
    25. McLaren W
    26. Moore B
    27. Mudge J
    28. Murphy DN
    29. Newman V
    30. Nuhn M
    31. Ogeh D
    32. Ong CK
    33. Parker A
    34. Patricio M
    35. Riat HS
    36. Schuilenburg H
    37. Sheppard D
    38. Sparrow H
    39. Taylor K
    40. Thormann A
    41. Vullo A
    42. Walts B
    43. Zadissa A
    44. Frankish A
    45. Hunt SE
    46. Kostadima M
    47. Langridge N
    48. Martin FJ
    49. Muffato M
    50. Perry E
    51. Ruffier M
    52. Staines DM
    53. Trevanion SJ
    54. Aken BL
    55. Cunningham F
    56. Yates A
    57. Flicek P

    (2018) Ensembl 2018

    Nucleic Acids Research 46:D754–D761.

    https://doi.org/10.1093/nar/gkx1098

    • PubMed
    • Google Scholar
99. 1. Zhang Y
    2. Liu T
    3. Meyer CA
    4. Eeckhoute J
    5. Johnson DS
    6. Bernstein BE
    7. Nusbaum C
    8. Myers RM
    9. Brown M
    10. Li W
    11. Liu XS

    (2008) Model-based analysis of ChIP-Seq (MACS)

    Genome Biology 9:R137.

    https://doi.org/10.1186/gb-2008-9-9-r137

    • PubMed
    • Google Scholar
100. 1. Zhao H
     2. Zhang Y
     3. Sun J
     4. Zhan C
     5. Zhao L

     (2016) Raltitrexed inhibits HepG2 cell proliferation via G0/G1 cell cycle arrest

     Oncology Research 23:237–248.

     https://doi.org/10.3727/096504016X14562725373671

     • PubMed
     • Google Scholar

Acceptance summary:

Your manuscript describes the application of combination of novel approaches – one molecular and the other microscopic, to map the location of uracil in the human genome in cells treated with the drugs 5-FdUR and RTX. The significant finding showing the correlation between a specific set of epigenetic markers and the distribution of uracil should be of great interest to the field of DNA repair and genome instability. Additionally, these findings provide significant leads for the mechanism of how the nucleotide pool imbalance caused by chemotherapeutics such 5-FdUR or RTX leads to genotoxic effects.

Decision letter after peer review:

[Editors’ note: the authors submitted for reconsideration following the decision after peer review. What follows is the decision letter after the first round of review.]

Thank you for choosing to send your work, "Genome-wide alterations of uracil distribution patterns in human DNA upon chemotherapeutic treatments", for consideration at eLife. Your article has been reviewed by three peer reviewers, one of whom is a member of our Board of Reviewing Editors, and the evaluation has been overseen by Jessica Tyler as the Senior Editor. The reviewers have opted to remain anonymous.

Our decision has been reached after consultation between the reviewers. Based on these discussions and the individual reviews below, we regret to inform you that your work will not be considered at this moment for publication in eLife. However, we believe that this is a very interesting study that after further work would be appropriate for eLife. It reveals that uracil distribution in DNA after the treatment with inhibitors of thymidine synthesis pathway is non-random. The pattern of enrichment of uracils at actively transcribed regions is very intriguing and anticipates many interesting questions regarding the mechanism. The correlation between uracil distribution and the histone markers associated with active chromatin is also very interesting, but the epigenetic marks have to be determined again under the conditions of this experimental system. The technique used to confirm such correlation, high-resolution microscopy, is very novel and promising. However, as detailed in the reviews, many of the controls that would add rigor to such techniques and findings are lacking. With the sufficient controls to support the main findings of the paper, this work would be highly significant reporting of novel findings and approaches. Please note that our policy is to aim to publish articles with a single round of revision that would typically be accomplished within two months. In our judgment, this manuscript would need extensive additional work beyond such time limit. So we would invite to resubmit a properly revised version attending the comments of the 3 reviewers accompanying your submission with a rebuttal explaining point-by-point the comments of the reviewers.

Reviewer #1:

In this manuscript, a catalytically inactive form of uracil DNA glycosylase is used to probe the genome-wide distribution patter of uracil in human cells. The expression of mutant UNG construct to detect uracil in DNA has been previously reported by the same group in Rona et al., 2016. In the same previous paper, IF was carried out to show that this construct is capable of detecting uracil-containing DNA in E. coli cells or on plasmids in MEF. This manuscript goes further from the previous report by analyzing the uracil distribution pattern in human cells following treatment with drugs (5FdUR and RTX) that inhibit thymidylate synthase enzyme and thus disrupt the dUTP-dTTP balance. Several different variations of the mutant UNG construct are used in two distinct major approaches in this manuscript. The first is to carry out ChIP-seq with antibody against the mutant UNG protein that will bind to uracil in DNA. The second is to modify the construct for the super-resolution immunofluorescent experiments to visualize the regions of uracil in DNA. The first approach was used to show that U-DNA are enriched at transcribed regions and early replicating regions, among others. The pattern of U-DNA in the genome after treatment with 5FdUR or RTX was also compared to the distribution pattern of multiple transcription factors and histone modification markers. Similar conclusion was made that histone markers unique to actively transcribed regions showed most similar pattern of distribution compared to U-DNA after drug treatment. The latter approach with IF was used to confirm that uracil in genomic DNA co-localize with two histone markers H3K36me3 and H3K27me3. Also, the potential mechanism of such distribution pattern is discussed. These are important findings that can also add to the understanding of the mechanism of the cytotoxicity of TS-inhibiting drugs. Overall, since the very interesting technical approach largely overlaps with the previously published report, the significance of the current manuscript should be in the substantive support for the new and significant finding regarding the distribution pattern of uracil after the drug treatment. There are however a few major flaws that detracts from the solidness of the significance of those findings. Two major weak points that call for further experiments are detailed below.

1) Figure 4. For the GIGGLE analysis, the ChIP-seq data for the histone markers and transcription factors were assembled from previously published data. The data set were matched for the same cell type used by the authors for U-DNA sequencing (HCT116). However, there was no consideration given to the fact that the treatment with 5FdUR or RTX could alter the pattern of transcription factors/histone markers significantly, obscuring the GIGGLE analysis between these data sets. Although generating a complete set of ChIP-seq data for all the different factors/markers for the 5FdUR or RTX-treated HCT116 cells would be beyond the scope of this manuscript and is not expected, a confirmation of the analysis for a number of the top hits would make for a much stronger argument. Generation of ChIP-seq results for H3K36me3 in RTX treated cells would complement the IF co-localization data shown in Figure 7 and more importantly would add to convince that such IF co-localization could reach the resolution to state confidently that U-DNA and the region of H3K36me3-modification overlap.

2) Figure 7: For the co-localization assay, H3K36me3 and H3K27me3 were chosen for further study. The rationale given by the authors is "As shown in Figure 4A, the highest similarity (GIGGLE) score corresponded to H3K36me3 and H3K27me3 for the RTX and 5FdUR treated samples." But according to Figure 4A, the correlation appears to be clearer for the H3K27ac. Why H3K27ac was not chosen instead for further study by IF should be discussed. Most notably, the negative correlation with WT or untreated sample was not very distinctive for the H3K27me3, making the difference in GIGGlE scores between NT_UGI and RTX_UGI very small. This is somewhat conflicting with the significant difference noted in the "Overlap area IF" between NT_UGI and RTX_UGI in Figure 7D and requires explanation. In general, αU-DNA signal for the 5FdUR or RTX-treated samples are much stronger than non-treated samples and whether difference in the signal strength interferes with the calling of the co-localization (overlap) is not fully discussed or controlled for. For convincing argument that super-resolution imaging can be used to analyze the genomic distribution pattern of uracil, similar experiments with at least one negative marker (such as H3K9me3 in Figure 4A) should be carried out.

Reviewer #2:

This manuscript describes a promising new low-resolution method for the visualization and mapping of uracils in DNA. It uses UNG inhibitor UGI in combination with a TYMS inhibitor Raltitrexed (RTX) or 5FdUR to increase uracils in the genome and then uses a mutant UNG to pull-down and sequence the DNA fragments. The authors claim that their data suggest that the increased uracils are predominantly in euchromatin/early replicating regions based on epigenetic markers. While the method has several attractive features including an ability to visualize uracil foci (especially Figure 7), there are several serious shortcomings of the manuscript.

1) The use of HCT116 cell line and UGI are troubling. The lack of mismatch repair (MMR) in HCT116 may affect their results. It is possible that the lack of active MMR during early replication affects the removal of uracils and thymines misincorporated across guanines. While UGI binds UNG and inhibits its action, it is unknown whether it has other physiological effects in the cells. In the very least, the results of this study should be compared with those in which MMR+ derivatives of HCT116 cells are used and UNG KO cells lines are compared with their WT parents.

2) Their uracil mapping software lacks a proper negative control. They calculate log2 ratio and plot this ratio across the genome for treated and untreated cells. The input DNA has not undergone the pull-down and hence the DNA pulled down using anti-FLAG antibodies will always show a different genomic distribution than the input DNA. The U-DNA sequencing results from treated cells should be normalized with respect to U-DNA sequencing reads from the pull-down of untreated samples, not the input DNA.

3) This method is unable to pinpoint uracils in the DNA that is pulled down. This is because, the uracils are in U:A pairs and hence they cannot be distinguished from normal T:A pairs after pull-down and sequencing. This is in contrast with most of the other available methods that can map uracils to a specific base pair. For this reason, this is a low resolution alternative to existing uracil mapping methods.

4) The manuscript lacks a proper positive control. It may be useful for the authors to introduce uracils by other means such introduction of AID gene to introduce uracils in the Ig locus and then demonstrate that their method can detect such localization of uracils. In the absence of a positive control, it is hard to know whether their method works correctly. It is possible- for example- that the treatment of cells with RTX or 5FdUR arrests the cells in early S phase and that is why the uracil incorporation occurs in early replicating regions. In which case, the lack of uracil-enriched fragments from the late replicating heterochromatin regions is a somewhat trivial result.

5) The authors dismiss the dU-seq (Shu et al., 2018) rather lightly. Shu and colleagues verified their findings by enrichment of centromeric regions showed by LC-MS/MS that centromeric uracils increased in UNG-knockout cells. Finally, they reintroduced UNG in the cells and showed that this reversed the uracil localization to centromeres. As noted above, the authors of the current manuscript have not done such careful controls in their experiments.

In summary, this is a moderately useful new technique to map uracils that does not distinguish itself from the other existing techniques except in the area of direct visualization of uracils in cells. If the authors were to use this unique feature of their method to answer an existing question, it would be attractive.

Reviewer #3:

The authors develop a new approach to measure the location and density of incorporated uracils in the human genome. They identify an interesting pattern of uracil incorporation in the human genome under normal and thymidine-depleted (using TS inhibitors) conditions, and associate these patterns with specific chromatin features. They also generalize the tool to visualize uracil incorporation with fluorescence microscopy. Both of these are significant advances of use to the DNA repair community. Moreover, these data may eventually be used to probe molecular and cellular consequences caused by common chemotherapies.

1) This study would be improved by employing computational methods designed to identify patterns across multiple independent genomic signals. The one-by-one correlations the authors present provide some insight into general correlations between uracil density, replication timing, and histone modifications, but this ad hoc approach could miss important correlations across all these signals. The segway software (Hoffman, Noble, et al) was developed specifically for this purpose and could be applied to data they have in hand to understand broader genome-wide correlations.

2) More characterization of any DNA damage response in these cells would be helpful to understand whether human uracil incorporation patterns are a "neutral" pattern in the absence of DDR (e.g., similar to high levels of seemingly innocuous uracil incorporation found in dut ung E. coli and yeast) , or represent a stress-induced condition caused by uracil incorporation followed by some checkpoint activation. Some flow cytometry would be useful to see the effect of RTX on cycling of these cells, and H2AX phosphorylation could be measured in bulk by western and in the microscopy experiments upon treatment.

3) Given previous indications that replication timing is somehow linked to uracil incorporation, more discussion is warranted to reconcile similarities and differences in this relationship in E. coli, yeast, and human uracil incorporation patterns relative to replication timing and associated changes in nucleotide pools.

[Editors’ note: further revisions were suggested prior to acceptance, as described below.]

Thank you for submitting your article "Genome-wide alterations of uracil distribution patterns in human DNA upon chemotherapeutic treatments" for consideration by eLife. Your article has been reviewed by three peer reviewers, one of whom is a member of our Board of Reviewing Editors, and the evaluation has been overseen by Jessica Tyler as the Senior Editor. The reviewers have opted to remain anonymous.

The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission.

We would like to draw your attention to changes in our revision policy that we have made in response to COVID-19 (https://elifesciences.org/articles/57162). Specifically, we are asking editors to accept without delay manuscripts, like yours, that they judge can stand as eLife papers without additional data, even if they feel that they would make the manuscript stronger. Thus the revisions requested below only address clarity and presentation.

Summary:

In this paper, the authors probe the genome-wide distribution patter of uracil in human cells using two innovative approaches. First, a catalytically inactive form of uracil DNA glycosylase is expressed and subjected to ChIP-seq to generate the genomic map of uracil distribution. The mutant UDG is also used in a super-resolution immuno-fluorescence experiments for higher throughput confirmation of uracil distribution pattern. These two complimentary approaches represent significant and useful advances in understanding the genome instability induced by uracil in DNA. Here, the authors' specific interest is the uracil distribution pattern in human cells following treatment with drugs (5FdUR and RTX) that inhibit thymidylate synthase enzyme and thus disrupt the dUTP-dTTP balance. The major findings are that U-DNA are mostly enriched at transcribed regions and early replicating regions. The correlation between the U-DNA in 5FdUR- or RTX-treated cells with epigenetic markers unique to actively transcribed regions, first determined using the ChIP-seq data, was further confirmed through the super resolution IF experiments. Such correlation is significant because it can inform the potential mechanism of uracil incorporation into the genome in thymine-depleted cells and thereby add to the understanding of the mechanism of the cytotoxicity of TS-inhibiting drugs.

Revisions:

The following minor revisions are strongly recommended for acceptance to publish in eLife.

1) The criticism of other methods are given the disproportionate significance in the discussion. Although the effort to highlight the advance and innovation of the technique used in the current manuscript over other previously published methods is completely valid and helpful, overly long and highly critical discussion of other methods diminishes the meaningful discussion of the findings in this paper and becomes unnecessary distraction. There is also a risk of misstating what the other methods do or point out deficiencies that may have simple explanations. A separate review paper comparing various methods would better suit such discussion. I recommend that the authors reorganize the Discussion section so that it starts by describing what this manuscript has accomplished and briefly summarize the pros and cons of all the methods towards the end.

2) Clarification is required regarding the replication scores derived from Weddington et al. Is the distribution of scores correlated with timing (negative = early, positive = late?). It's unfortunate that ENCODE Repli-seq data is not available for the HCT116 cell line; this would have been a better comparison. Please add discussion or speculate on the possible mechanism regarding why uracil enrichment upon treatment both appear at similar replication timing, relative to the control samples (Figure 4—figure supplement 3)?

[Editors’ note: the authors resubmitted a revised version of the paper for consideration. What follows is the authors’ response to the first round of review.]

It reveals that uracil distribution in DNA after the treatment with inhibitors of thymidine synthesis pathway is non-random. The pattern of enrichment of uracils at actively transcribed regions is very intriguing and anticipates many interesting questions regarding the mechanism. The correlation between uracil distribution and the histone markers associated with active chromatin is also very interesting, but the epigenetic marks have to be determined again under the conditions of this experimental system.

We performed a pair of ChIP-seq experiments for H3K36me3 epigenetic mark (as reviewer #1 suggested) in non-treated and RTX-treated UGI-expressing HCT116 cells. We found that there are no major differences in the genome-wide distribution of this factor, or in its correlation with the U-DNA-Seq patterns upon the applied treatment conditions. Results are presented in Figure 4—figure supplement 1, and Figure 4A-B. Corresponding texts are inserted to the respective figure legends (4A: “GIGGLE scores are also indicated for our own H3K36me3 ChIP-seq…”, 4B: “…our own ChIP-seq data for H3K36me3…”); to the Results (a paragraph started with “In order to decide whether drug treatments may cause any notable changes…”); to the Materials and methods (a paragraph with the title: “Chromatin immunoprecipitation and sequencing (ChIP-seq)”), to the Discussion (“With the ChIP-seq experiment for H3K36me3…”); and to the Supplementary file 3 (a section with the title: “ChIP-seq for H3K36me3 histone marker in non-treated and RTX treated UGI-expressing HCT116 cells.” including Supplementary file 3—table 2). ChIP-seq peak results are also uploaded to the GEO GSE126822.

The technique used to confirm such correlation, high-resolution microscopy, is very novel and promising. However, as detailed in the reviews, many of the controls that would add rigor to such techniques and findings are lacking. With the sufficient controls to support the main findings of the paper, this work would be highly significant reporting of novel findings and approaches.

In the revised manuscript, we provide a positive control experiment for the U-DNA-Seq, namely, a uracil-containing spike-in sequence was added to genomic DNA fragments, then the U-DNA-IP was performed and the enrichment of the spike-in was followed by qPCR. Corresponding sentences are inserted to the Results (“To further confirm the capability of U-DNA-IP, uracil-containing spike-in DNA was combined with…”), and to the Materials and methods (a paragraph with the title: “Controls of U-DNA-IP method”).

In the submitted manuscript, we have already mentioned a negative control experiment for the U-DNA-Seq: “Specificity of U-DNA immunoprecipitation is also underlined by the fact that pull-down with empty anti-FLAG beads not containing the U-DNA sensor resulted in negligible amount of DNA (less than 5%).” During the revision, we have obtained the full genome sequencing data for this negative (empty bead) control, and we provide the corresponding data in Figure 1—figure supplement 2, and Supplementary file 1—table 1. The main text was also modified accordingly, see text segment: “Still, genome-wide sequencing data could be obtained from these negative control samples as well. We demonstrated that subtracting such control signals (for details see Supplementary file 1) will not affect the detected uracil distribution pattern regardless if the sample was drug-treated or not (Figure 1—figure supplement 2B-C). These control experiments provided confidence about the applicability and specificity of our U-DNA-IP method.”. Corresponding description was added to the Materials and methods (a paragraph with the title: “Controls of U-DNA-IP method” and “High-throughput DNA sequencing and data analysis”).

Reviewer #1:

In this manuscript, a catalytically inactive form of uracil DNA glycosylase is used to probe the genome-wide distribution patter of uracil in human cells. The expression of mutant UNG construct to detect uracil in DNA has been previously reported by the same group in Rona et al., 2016. In the same previous paper, IF was carried out to show that this construct is capable of detecting uracil-containing DNA in E. coli cells or on plasmids in MEF. This manuscript goes further from the previous report by analyzing the uracil distribution pattern in human cells following treatment with drugs (5FdUR and RTX) that inhibit thymidylate synthase enzyme and thus disrupt the dUTP-dTTP balance. Several different variations of the mutant UNG construct are used in two distinct major approaches in this manuscript. The first is to carry out ChIP-seq with antibody against the mutant UNG protein that will bind to uracil in DNA. The second is to modify the construct for the super-resolution immunofluorescent experiments to visualize the regions of uracil in DNA. The first approach was used to show that U-DNA are enriched at transcribed regions and early replicating regions, among others. The pattern of U-DNA in the genome after treatment with 5FdUR or RTX was also compared to the distribution pattern of multiple transcription factors and histone modification markers. Similar conclusion was made that histone markers unique to actively transcribed regions showed most similar pattern of distribution compared to U-DNA after drug treatment. The latter approach with IF was used to confirm that uracil in genomic DNA co-localize with two histone markers H3K36me3 and H3K27me3. Also, the potential mechanism of such distribution pattern is discussed. These are important findings that can also add to the understanding of the mechanism of the cytotoxicity of TS-inhibiting drugs. Overall, since the very interesting technical approach largely overlaps with the previously published report, the significance of the current manuscript should be in the substantive support for the new and significant finding regarding the distribution pattern of uracil after the drug treatment. There are however a few major flaws that detracts from the solidness of the significance of those findings. Two major weak points that call for further experiments are detailed below.

1) Figure 4. For the GIGGLE analysis, the ChIP-seq data for the histone markers and transcription factors were assembled from previously published data. The data set were matched for the same cell type used by the authors for U-DNA sequencing (HCT116). However, there was no consideration given to the fact that the treatment with 5FdUR or RTX could alter the pattern of transcription factors/histone markers significantly, obscuring the GIGGLE analysis between these data sets. Although generating a complete set of ChIP-seq data for all the different factors/markers for the 5FdUR or RTX-treated HCT116 cells would be beyond the scope of this manuscript and is not expected, a confirmation of the analysis for a number of the top hits would make for a much stronger argument. Generation of ChIP-seq results for H3K36me3 in RTX treated cells would complement the IF co-localization data shown in Figure 7 and more importantly would add to convince that such IF co-localization could reach the resolution to state confidently that U-DNA and the region of H3K36me3-modification overlap.

We agree to the possibility that treatment with 5FdUR or RTX could alter the pattern of transcription factors/histone markers. In our manuscript, we use the comparison between the uracil-DNA distribution pattern in normal and drug-treated cells and the pattern of the epigenetic markers reported in the databases for the same cell line (HCT116) under normal conditions as a potential indication that need to be investigated further by independent experiments. We use super-resolution microscopy (dSTORM) as a relevant and highly potent further independent technique. We found that dSTORM qualitatively confirmed the positive correlation between the epigenetic markers and the U-DNA patterns (reported in the GIGGLE scores). We also agree that determination of the epigenetic marker distribution under drug treatment would be very interesting and informative and this is one of the directions we plan to continue our further research.

We agree with the reviewer’s comment that “generating a complete set of ChIP-seq data for all the different factors/markers for the 5FdUR or RTX-treated HCT116 cells would be beyond the scope of this manuscript”. Instead, the reviewer suggested to generate ChIP-seq data for H3K36me3 in RTX treated cells, as an example to “complement the IF co-localization data”. We agree to this suggestion and in our revision, we have performed the required ChIP-seq experiments for non-treated UGI-expressing and RTX-treated UGI-expressing cells. We found that there are no major differences in the genome-wide distribution of this factor, or in its correlation with the U-DNA-Seq patterns upon the applied treatment conditions. Results are presented in Figure 4—figure supplement 1, and Figure 4A-B. Corresponding texts are inserted to the respective figure legends (4A: “GIGGLE scores are also indicated for our own H3K36me3 ChIP-seq…”, 4B: “…our own ChIP-seq data for H3K36me3…”); to the Results (a paragraph started with “In order to decide whether drug treatments may cause any notable changes…”); to the Materials and methods (a paragraph with the title: “Chromatin immunoprecipitation and sequencing (ChIP-seq)”), to the Discussion (“With the ChIP-seq experiment for H3K36me3…”); and to the Supplementary file 3 (a section with the title: “ChIP-seq for H3K36me3 histone marker in non-treated and RTX treated UGI-expressing HCT116 cells.” including Supplementary file 3—table 2). ChIP-seq peak results are also uploaded to the GEO GSE126822.

2) Figure 7: For the co-localization assay, H3K36me3 and H3K27me3 were chosen for further study. The rationale given by the authors is "As shown in Figure 4A, the highest similarity (GIGGLE) score corresponded to H3K36me3 and H3K27me3 for the RTX and 5FdUR treated samples." But according to Figure 4A, the correlation appears to be clearer for the H3K27ac. Why H3K27ac was not chosen instead for further study by IF should be discussed. Most notably, the negative correlation with WT or untreated sample was not very distinctive for the H3K27me3, making the difference in GIGGlE scores between NT_UGI and RTX_UGI very small. This is somewhat conflicting with the significant difference noted in the "Overlap area IF" between NT_UGI and RTX_UGI in Figure 7D and requires explanation. In general, αU-DNA signal for the 5FdUR or RTX-treated samples are much stronger than non-treated samples and whether difference in the signal strength interferes with the calling of the co-localization (overlap) is not fully discussed or controlled for. For convincing argument that super-resolution imaging can be used to analyze the genomic distribution pattern of uracil, similar experiments with at least one negative marker (such as H3K9me3 in Figure 4A) should be carried out.

The main reason for our selections of the H3K36me3 and the H3K27me3 markers for IF colocalization measurements was that these factors produced the highest scores for RTX and 5FdUR treatments, respectively, as reviewer #1 also wrote. We agree that we could have selected other factors like H3K27ac, H3K9ac, H3K4me3, or even the transcription factor SP1, where both treatments resulted in high GIGGLE scores and these are well separated from the non-treated cases. However, with the choice of H3K36me3 and the H3K27me3, we sought to cover better the spectra from the actively transcribing euchromatin to the facultative heterochromatin. Furthermore, with this choice, we could also address putative differences between the two drug treatments. A new genome segmentation analysis performed in our revision upon the request of the reviewer #3, also supports our choice (cf. Figure 4B in the revised manuscript, and inserted text to the Results (“Furthermore, Segway analysis strengthened…”)).

Regarding the reviewer’s comments that “the negative correlation with WT or untreated sample was not very distinctive for the H3K27me3…”, and that “this is somewhat conflicting with the significant difference noted in the "Overlap area IF" between NT_UGI and RTX_UGI in Figure 7D…”, we provide the following explanation. On the one hand, as the reviewer also pointed out (point #1), the two experimental setup has a clear difference, namely in GIGGLE analysis we compare our uracil patterns in non-treated and drug-treated cells with non-treated ChIP-seq data from databases, while in dSTORM microscopy the overlap of the two features can be calculated in the same treated cells. On the other hand, the available ChIP-seq data are from different experiments and experimental backgrounds, hence show quite big deviation among themselves. Therefore, such in silico correlation analysis can provide potential indications, but not strict rules for the choice of markers to be addressed in colocalization studies by dSTORM microscopy. Still, the in silico analysis gave meaningful prediction, as co-staining of the selected histone markers and the genomic uracil in drug-treated cells reinforced the association between uracil occurrence and transcriptionally active regions. Moreover, with the H3K36me3 ChIP-seq (cf. point #1), we also demonstrated that the treatment induced chromatin remodeling is not a general phenomenon, but may rather confines to certain histone marks. The apparent conflict between our results from these different techniques is further explained now in the Discussion of the revised manuscript (“Strikingly, we found that H3K27me3 shows even stronger colocalization with the U-DNA pattern in case of the RTX treated sample…”).

Regarding the possible influence of the overall signal strength on the calculation of IF (overlap), we have inserted a detailed explanation to the Results: “The cross-pair correlation method probes the probability distributions across all possible pair-wise distances between two species, taking in account the number of foci for each species (PMIDs: 25179006, 23717596, 22384026 and 27545293). This normalization of the number of foci ensures that any increase in IF is specifically due to an increase in their co-localization probability density, and not due to the increase in the amount of either species.”

Regarding the suggestion to stain for another marker (such as H3K9me3) that is supposed to negatively correlate with genomic uracils based on the GIGGLE scores and also on the genome segmentation results (cf. Figure 4A-B), we agree that it would be interesting to perform such an experiment. However, both Figure 4 and Figure 4—figure supplement 3 show that the distribution pattern of the genomic uracils is not fully represented by any of the histone markers. The best correlating features are the early replication timing and AT rich heterochromatin in case of the drug-treated and non-treated samples, respectively. As a negative control for the drug-treated samples, it would be required therefore to identify a true heterochromatin (or late replication) marker that might be challenging. We also wish to point out that due to the current pandemic situation of COVID-19, the laboratory at the New York University (School of Medicine) – where these studies are performed – is hardly available. We cannot define any clear schedule for performing additional dSTORM experiments, and this situation would lead to an unpredictable and disproportionate delay in the publication process.

Reviewer #2:

This manuscript describes a promising new low-resolution method for the visualization and mapping of uracils in DNA. It uses UNG inhibitor UGI in combination with a TYMS inhibitor Raltitrexed (RTX) or 5FdUR to increase uracils in the genome and then uses a mutant UNG to pull-down and sequence the DNA fragments. The authors claim that their data suggest that the increased uracils are predominantly in euchromatin/early replicating regions based on epigenetic markers. While the method has several attractive features including an ability to visualize uracil foci (especially Figure 7), there are several serious shortcomings of the manuscript.

1) The use of HCT116 cell line and UGI are troubling. The lack of mismatch repair (MMR) in HCT116 may affect their results. It is possible that the lack of active MMR during early replication affects the removal of uracils and thymines misincorporated across guanines. While UGI binds UNG and inhibits its action, it is unknown whether it has other physiological effects in the cells. In the very least, the results of this study should be compared with those in which MMR+ derivatives of HCT116 cells are used and UNG KO cells lines are compared with their WT parents.

The HCT116 cell line has been widely used as a model cell line in colon cancer research over the course of almost four decades (PMID: 7214343, 6956756, 22955616), for which numerous published results – including epigenomic data – are available e.g. in the ENCODE database. Colon cancer is frequently associated with deficiency in mismatch repair (MMR) (PMID: 30959407, 30442708, 28548127), and thymidylate synthase inhibitors represent a possible option as first-line treatment in the clinical practice (PMID: 24732946). Based on these data, we propose that the choice of HCT116 cell line was relevant in our study for revealing how TS inhibitors may affect uracil distribution patterns. We agree that the potential contribution of the mismatch repair to the mechanism of the effect of these drugs is a highly intriguing question that might have impact also on clinical prognosis and/or the choice of treatments. Initially, we felt that such an analysis is beyond the focus of this manuscript, but prompted by the request of the reviewer, we decided to include comparative studies on the MMR proficient version of HCT116 cells (created by the reintroduction of chromosome 3 (PMID: 8044777)), as well. For this, 12 new U-DNA-Seq experiments were performed on MMR proficient samples (shown in modified Figure 2—figure supplement 3, Figure 3, Figure 3—figure supplement 3, and the new Figure 3—figure supplement 4), we have carried out the corresponding correlation analysis (shown in modified Figure 4, Figure 4—figure supplement 2, and the new Figure 4—figure supplement 3), as well as measurement of uracil-content by dot blot (shown in new Figure 1—figure supplement 1 panel D) were also performed. Further experiments suggested by reviewer #3, namely cell cycle analysis (shown in new Figure 5) and γH2AX staining in flow cytometry (shown in new Figure 5—figure supplement 1) were also done on both the MMR deficient and MMR proficient samples. We found that MMR status does not have any observable impact on the U-DNA pattern of the non-treated samples, while in case of drug-treated ones, some drug-dependent differences were detected. We have described and analyzed these new data at different sections of the Results, as well as in the figure legends (all changes are tracked in the revised manuscript). To discuss these results, the following texts are inserted to the Discussion of the revised manuscript: “We chose the HCT116 cancer cell line that is deficient in mismatch repair…”; “It has to be noted that MMR proficiency leads to a major decrease in the correlation with early replication timing…”; “This is further supported by the fact that in MMR proficient drug-treated samples higher U-DNA content…”; “Moreover, the MMR status has markedly different influence on the resulting U-DNA pattern in case of the two drugs (cf. Figure 3-4).”; “Similarly, we also detected slightly altered cell cycle distribution patterns in case of the two drug treatments, which were differently influenced by the MMR status…”.

Regarding the use of UGI, we have to point out that UGI is a well characterized specific, small (10 kDa) proteinaceous UNG-inhibitor (PMID: 7671300). UGI expression in HEK293 cells led elevated genomic uracil content, while did not affect the cytotoxicity, the cell cycle arrest, the γH2AX signaling upon RTX or 5FdUR treatments (PMID: 17942376). We refer to these data in the Introduction of the revised manuscript (“It has already been shown that UGI expression does not affect…”). In agreement with the literature, we also detected (i) elevated U-DNA content upon drug treatments (cf. Figure 1—figure supplement 1B-D), (ii) no major physiological effect on the non-treated cells in respect to their cell cycling (cf. Figure 5A performed upon the request of the reviewer #3), and (iii) no difference in the genomic uracil distribution as compared to the non-treated wild type cells (cf. Figure 3, Figure 3—figure supplement 3, Figure 3—figure supplement 4, Figure 4, Figure 4—figure supplement 2, Figure 4—figure supplement 3). Based on these arguments, we feel that repeating all of the experiments (overall 28 sequencing, the corresponding dot blots, flow cytometry, super-resolution imaging) on UNG-KO cells would not add fundamental new insights to this present study, but may form the basis for further research.

2) Their uracil mapping software lacks a proper negative control. They calculate log2 ratio and plot this ratio across the genome for treated and untreated cells. The input DNA has not undergone the pull-down and hence the DNA pulled down using anti-FLAG antibodies will always show a different genomic distribution than the input DNA. The U-DNA sequencing results from treated cells should be normalized with respect to U-DNA sequencing reads from the pull-down of untreated samples, not the input DNA.

The reviewer suggested to use the non-treated IP samples as controls for the treated ones. In a way, we agree, and we basically did this, when we compared the log2 curves (cf. Figure 3, Figure 3—figure supplement 3), assessed their colocalization with other factors or genetic features (Figure 4, Figure 4—figure supplement 2, Figure 4—figure supplement 3), and then compared the results of the drug-treated and non-treated samples in many respects. However, for the calculation of log2 ratio and for the peak calling, we decided to use the corresponding sonicated input samples as controls. We claim that this choice was reasonable because of the following arguments.

First, sonicated genomic DNA samples provide a more balanced coverage of the reference genome, but it is still not without some fluctuation. Such fluctuation might be slightly different in case of non-treated and drug-treated cells, and therefore, it is the best to compare each IP sample (enriched in U-DNA) to their corresponding input (sonicated) ones. As expected, we did not experience highly enriched sharp peaks, therefore, such fluctuations might have a distorting impact and need to be avoided.

Second, using input as a background control for peak calling or log2 ratio enrichment calculation meets the current ENCODE standards for ChIP-seq. Currently, corresponding input is required, while previous standard suggested either input or Ig control experiment (https://www.encodeproject.org/chipseq/histone/#restrictions). Accordingly, we have inserted the following statement to the Supplementary file 1: “…as it is also recommended by the current ENCODE standard…”.

Third, given that the U-DNA patterns of the drug-treated and non-treated samples are almost complementary to each other (cf. Figure 3A and Supplementary file 2), applying non-treated samples as controls would result in a signal increase artifact rather than better description of the true enrichment.

Fourth, using the non-treated IP sample as control in the peak calling or the log2 enrichment analysis would evoke technical problems about the scaling and the handling of non-covered regions. (1) If a region is not covered by reads in case of the non-treated IP sample, then it is hard to calculate the log2 ratio. The two usual solutions for such problem may be skipping databins of zero that would result in loosing data or introducing a pseudocount that might influence the results especially at the regions of lower enrichment. (2) It is not trivial how to compare a relatively big amount of DNA immunoprecipitated from the drug-treated samples to a much smaller amount of DNA originating from the non-treated cells (5:1, cf. Figure 1—figure supplement 2A), given that the deepness of the sequencing is approximately the same for both samples.

In addition to the discussion above, we also performed blank IP experiments (using empty beads without ∆UNG sensor) and sequencing to make sure that the U-DNA patterns are valid, and not due to a specific pull-down, even in case of the non-treated samples. Data are presented in Figure 1—figure supplement 2, and Supplementary file 1—table 1. The main text was also modified accordingly, see text segment: “Still, genome-wide sequencing data could be obtained from these negative control samples as well. We demonstrated that subtracting such control signals (for details see Supplementary file 1) will not affect the detected uracil distribution pattern regardless if the sample was drug-treated or not (Figure 1—figure supplement 2B-C). These control experiments provided confidence about the applicability and specificity of our U-DNA-IP method.”. Corresponding description was added to the Materials and methods (a paragraph with the title: “Controls of U-DNA-IP method” and “High-throughput DNA sequencing and data analysis”).

3) This method is unable to pinpoint uracils in the DNA that is pulled down. This is because, the uracils are in U:A pairs and hence they cannot be distinguished from normal T:A pairs after pull-down and sequencing. This is in contrast with most of the other available methods that can map uracils to a specific base pair. For this reason, this is a low resolution alternative to existing uracil mapping methods.

We agree that our method is a low-resolution technique, but addressing the case at issue, it is relevant, adequate and robust. In the given model system, under thymidylate synthesis inhibition potentially leading to a perturbation of dUTP/dTTP ratios, uracil incorporation into the genome during either replicative or repair DNA synthesis is a basically stochastic process. Under such circumstances, characteristic U-DNA patterns might be associated with uneven genome-wide distribution of the DNA synthesis foci (e.g. at replication forks or different repair foci) and/or actual repair activities at certain regions. We claim that our approach can reliably describe these patterns. We also feel that to pinpoint the individual uracil residues would not provide additional value to this description, but could be rather misleading, if we consider that from about 10 millions of cells (=20 millions of genomes), we read on average one per million only. This issue is covered in the Discussion of the revised manuscript (see text starting with: “However, this aspect has lower impact, if we consider the basically stochastic nature of uracil…”).

We have reviewed the publications of other methods for determination of uracils in DNA and found that only the pre-digestion Excision-seq was explicitly stated to be capable to detect uracils with singlebase resolution. Excision-seq was used in the literature in samples with higher uracil content and smaller genome-size (PMID: 25015380). Although both UPD-seq (PMID: 31431505) and dU-seq (PMID: 29785056) use biotin-streptavidin system for the pull-down, and the biotin labeling could potentially allow site-specific detection, single-base resolution data have not been shown in either of these publications. In the revised manuscript, the first paragraph of the Discussion has been rephrased to provide a more elaborate comparison of the available methods.

The U-DNA-Seq method we present includes a refined analysis pipeline to ascertain its status as a useful alternative of the existing U-DNA sequencing methods. The use of the catalytically inactive UNG-based sensor renders it possible to complement the genome-wide distribution patterns with cellular localization studies. This advantage is also better underlined in the revised version, following the suggestion of the reviewer #3.

4) The manuscript lacks a proper positive control. It may be useful for the authors to introduce uracils by other means such introduction of AID gene to introduce uracils in the Ig locus and then demonstrate that their method can detect such localization of uracils. In the absence of a positive control, it is hard to know whether their method works correctly. It is possible- for example- that the treatment of cells with RTX or 5FdUR arrests the cells in early S phase and that is why the uracil incorporation occurs in early replicating regions. In which case, the lack of uracil-enriched fragments from the late replicating heterochromatin regions is a somewhat trivial result.

We agree that a relevant positive control is valuable. Reviewer #2 suggested expression of AID to introduce uracils specifically into the Ig locus in order to detect such localization of uracils by our method. However, overexpression of AID without its proper regulation and targeting is expected to lead to deamination events outside the Ig loci also, as it was reported in case of tumorigenesis (PMID: 26845615). We therefore designed another positive control experiment. Namely, a uracil-containing spike-in sequence was added to genomic DNA fragments, then the U-DNA-IP was performed and the enrichment of the spike-in was followed by qPCR, as it has been already mentioned above in the reply to the Editor’s main point #2. Corresponding sentences are inserted to the Results (“To further confirm the capability of U-DNA-IP, uracil-containing spike-in DNA was combined with non-treated genomic DNA samples (cf. Materials and methods). In these samples U-DNA-IP led to 4.5 fold enrichment of the uracil-containing spike-in DNA compared to the uracil-free spike-in as determined by qPCR.”), and to the Materials and methods (a paragraph with the Title: “Controls of U-DNA-IP method”).

We note that the robustness and the reproducibility over biological replicates reported in our study also support the applicability of our method. Moreover, comparison of U-DNA-Seq results to the independent dU-seq data also confirmed the reliability of our method (cf. Discussion of the submitted manuscript (“Re-analysis of the earlier published dU-seq data…”) and Appendix 1).

The reviewer also mentioned the possibility “that the treatment of cells with RTX or 5FdUR arrests the cells in early S phase and that is why the uracil incorporation occurs in early replicating regions. In which case, the lack of uracil-enriched fragments from the late replicating heterochromatin regions is a somewhat trivial result.”. We agree that the S-phase arrest will influence the pattern of uracil incorporation, and now in the revised manuscript, it is presented and discussed more extensively. Clear correlation has already been shown between the U-DNA pattern and early replicating segments in the submitted manuscript (cf. Figure 4C-D). Further genome segmentation analysis, presented in the revised manuscript, also confirmed this (Figure 4—figure supplement 3). In addition, upon the request of the reviewer #3, cell cycle analysis was also performed (Figure 5) supporting the presence of a cell cycle arrest upon the applied drug treatment. The connection of the cell cycle arrest and the correlation between U-DNA pattern and early replicating segments is now presented in the revised manuscript (cf. Figure 4C-D and Figure 5). We propose that these results provide additional insights. U-DNA-Seq might give a molecular approach to the replication arrest, while flow cytometry provides a phenotypic description of cell cycle phases. Interestingly, drug dependent and also MMR status dependent differences were detected by both approaches, which are discussed in the Discussion of the revised manuscript (“Similarly, we also detected slightly altered cell cycle distribution patterns…”). It also has to be noted that correlation with the replication timing is not exclusive, and DNA synthesis coupled to either transcriptional activity or epigenetic remodeling might also have an impact on the resulting genomic U-DNA pattern in the drug-treated samples.

5) The authors dismiss the dU-seq (Shu et al., 2018) rather lightly. Shu and colleagues verified their findings by enrichment of centromeric regions showed by LC-MS/MS that centromeric uracils increased in UNG-knockout cells. Finally, they reintroduced UNG in the cells and showed that this reversed the uracil localization to centromeres. As noted above, the authors of the current manuscript have not done such careful controls in their experiments.

It is our intention to provide a useful comparison among the relevant methods. Towards this end, we present a well detailed assessment of the dU-seq method in the Discussion section, and also in Appendix 1. We do not question the validity and reliability of the LC-MS/MS data of the Shu et al. publication and to emphasize this, we added the following sentence into the Discussion section: “It has to be noted that the authors also applied additional experimental approaches, like 3D-PCR for sensitive detection of U:G pairs, mass spectrometry on genomic DNA fraction enriched in centromeric DNA regions.”

However, there are serious concerns about the capability of dU-seq method to reveal centromeric localization (cf. Appendix 1). These are the following: (1) Short-reads were mapped to the highly repetitive centromeres using a routine mapping software without any extra care or quality control. (2) The data analysis pipeline is not published in sufficient clear details, major points are missing, e.g. whether only the uniquely mapped reads were used or not, whether blacklist and deduplication were applied or not, or which reference genome set was used. (3) Their peak calling was also not reported correctly, it is not fully described if they used controls, and which parameters they set in the MACS2. (4) Their probable approach was to subtract peaks of the control experiment from the peaks of pulled down samples, which is also questionable (cf. above). (5) Their results are shared on the GEO only partially, making a critical evaluation quite problematic.

Finally, we agree with the reviewer that our method has the novel potential for direct visualization of uracils in human genomic DNA. Here, we demonstrated that our technique allows (1) detection of genomic uracil in eukaryotic cells, (2) co-staining with other factors, and (3) measurement of co-localization for the predicted correlating factors from the U-DNA-Seq results. These are important steps towards its application in more sophisticated experimental setup also combined with time-resolved U-DNA-Seq and RNA-seq that might lead to deeper understanding of the molecular mechanisms, and especially the impact of uracil incorporation upon treatments with TS inhibitors. Our present results contribute novel information relevant to existing questions such as: (1) What kind of genomic uracil patterns can arise upon two widely used thymidylate synthase inhibitor drugs in a valid cancer cell line model? (2) Is there any correlation between the U-DNA patterns characteristic for these drug treatments and epigenetic factors or other genomic features? (3) Is there any connection between the known cell cycle arrest and the U-DNA pattern? (4) Is there any difference between the U-DNA patterns induced by treatments with the two drugs? (5) Is there any impact of the MMR status on the U-DNA pattern and does it correlate with mechanistic differences? Further studies based on our present results will reveal more mechanistic details.

Reviewer #3:

The authors develop a new approach to measure the location and density of incorporated uracils in the human genome. They identify an interesting pattern of uracil incorporation in the human genome under normal and thymidine-depleted (using TS inhibitors) conditions, and associate these patterns with specific chromatin features. They also generalize the tool to visualize uracil incorporation with fluorescence microscopy. Both of these are significant advances of use to the DNA repair community. Moreover, these data may eventually be used to probe molecular and cellular consequences caused by common chemotherapies.

1) This study would be improved by employing computational methods designed to identify patterns across multiple independent genomic signals. The one-by-one correlations the authors present provide some insight into general correlations between uracil density, replication timing, and histone modifications, but this ad hoc approach could miss important correlations across all these signals. The segway software (Hoffman, Noble, et al) was developed specifically for this purpose and could be applied to data they have in hand to understand broader genome-wide correlations.

We highly appreciate this suggestion and fully agree. In the revised version, we have performed the suggested Segway analysis on the U-DNA-Seq data also including those obtained from MMR proficient cells (latter experiments were performed according to the request of reviewer #2). We have used ChIP-seq data on HCT116 cells available in the ENCODE database, and also used data from our own H3K36me3 ChIP-seq experiment (performed according to the request of reviewer #1). The results shed light on the relations of the different correlating factors and features, suggesting that none of the histone markers alone can explain the detected U-DNA patterns, and that the correlation with the replication timing or the chromatin structure should be taken into account with higher impact. The resulted signal distribution is presented in Figure 4B and further calculations regarding the replication timing scores and AT content of these genomic segments are shown in Figure 4—figure supplement 3. Corresponding texts are inserted into the respective figure legends; the Materials and methods (“Genome segmentation analysis on our U-DNA-Seq data,…", and “Replication timing scores and AT content were calculated on the genomic segments…”); the Results (a complete section starting with “To understand broader genome-wide correlations, a genome segmentation approach was employed…“, and two sentences “The replication timing correlation and the AT content were also calculated…”, and “Furthermore, Segway analysis strengthened…”); and the Discussion (“Such combinatorial behavior was further demonstrated by the genome segmentation analysis…"). The applied commands/scripts are reported in the Supplementary file 3 with the subtitle “Genome segmentation analysis of U-DNA-Seq results and ChIP-seq data from the ENCODE using Segway genome segmentation algorithm.”. The title of Supplementary file 3 was changed accordingly. Similarly, Supplementary file 4 was completed with the calculation of replication timing and AT content.

2) More characterization of any DNA damage response in these cells would be helpful to understand whether human uracil incorporation patterns are a "neutral" pattern in the absence of DDR (e.g., similar to high levels of seemingly innocuous uracil incorporation found in dut ung E. coli and yeast) , or represent a stress-induced condition caused by uracil incorporation followed by some checkpoint activation. Some flow cytometry would be useful to see the effect of RTX on cycling of these cells, and H2AX phosphorylation could be measured in bulk by western and in the microscopy experiments upon treatment.

We agree that more characterization on the phenotype caused by the drug treatments could improve our understanding of the U-DNA pattern. Hence, based on the suggestion of the reviewer, we performed two types of flow cytometry experiments addressing the potential effect of drug treatments on the cell cycle progression as well as on the DNA-damage signaling under the same conditions that were used in the U-DNA-Seq and the colocalization dSTORM ICC experiments.

First, we addressed the cell cycle using propidium iodide and 5′-bromo-2′-deoxyuridine (BrdU) staining in all those HCT116 samples, for which we provided U-DNA-Seq data. Our results can be summarized as follows. (1) In non-treated cells, UGI expression did not cause any visible changes in the cell cycle progression. (2) Both RTX and 5FdUR treatment (in the presence of UGI) caused significant changes, resulting in arrest-like patterns, which are not identical in case of the two drugs, in good agreement with the detected differences of the corresponding U-DNA patterns. (3) MMR status has an influence on these patterns, which is in good agreement with the MMR status related changes in the correlations between U-DNA patterns and replication timing. These results are presented in Figure 5 of the revised manuscript. Corresponding texts are inserted into the figure legend; the Results (a paragraph that is started with “As the uracil distribution pattern in drug-treated cells shows correlation…”); the Discussion (the following sentences: “We demonstrated that UGI expression did not cause any observable…”, “… or the less tight control on cell cycle arrest (cf. Figure 5) allowing more extended replicative synthesis.”, “Similarly, we also detected slightly altered cell cycle distribution patterns in case of the two drug treatments…”), and the Materials and methods (a section with the title “Cell cycle analysis and γH2AX staining”).

Second, regarding the potentially induced DNA damage response (DDR), we also performed flow cytometry measurements using γH2AX immunostaining. The results revealed an increased γH2AX level indicative for DDR in case of both drug treatments. These results are shown in the Figure 5—figure supplement 1. Corresponding texts are inserted into the figure legend; the Results (“As expected (cf. (Meyers et al., 2001)), 5FdUR and RTX treatment…”); the Discussion (“However, equally induced DNA damage response (reported by γH2AX)…”); and the Materials and methods (“Occurrence of DSBs was investigated by immunofluorescent staining of γH2AX…”).

3) Given previous indications that replication timing is somehow linked to uracil incorporation, more discussion is warranted to reconcile similarities and differences in this relationship in E. coli, yeast, and human uracil incorporation patterns relative to replication timing and associated changes in nucleotide pools.

To discuss the correlation between uracil incorporation and replication timing, we have inserted a new paragraph into the Discussion starting with “Our data showing that under normal conditions…”. Here, we show differences between the timing of genomic uracil incorporation in our drug-treated cells and in other biological systems previously published in the literature, and the possible reasons are also discussed.

[Editors’ note: what follows is the authors’ response to the second round of review.]

Revisions:

The following minor revisions are strongly recommended for acceptance to publish in eLife.

1) The criticism of other methods are given the disproportionate significance in the discussion. Although the effort to highlight the advance and innovation of the technique used in the current manuscript over other previously published methods is completely valid and helpful, overly long and highly critical discussion of other methods diminishes the meaningful discussion of the findings in this paper and becomes unnecessary distraction. There is also a risk of misstating what the other methods do or point out deficiencies that may have simple explanations. A separate review paper comparing various methods would better suit such discussion. I recommend that the authors reorganize the Discussion section so that it starts by describing what this manuscript has accomplished and briefly summarize the pros and cons of all the methods towards the end.

We agree with the Editors’ recommendation, and accordingly we have restructured the Discussion. Comparison of the different methods was significantly shortened and moved from the beginning of the Discussion (see paragraph starting with "As we demonstrated here, the genome-wide uracil distribution patterns have relevance…"). For harmonization, a new first paragraph is inserted in the Discussion starting with "Here we focus on the alteration of U-DNA distribution pattern".

2) Clarification is required regarding the replication scores derived from Weddington et al. Is the distribution of scores correlated with timing (negative = early, positive = late?). It's unfortunate that ENCODE Repli-seq data is not available for the HCT116 cell line; this would have been a better comparison.

Thank you for pointing out the meaning of replication score was not defined. We have now inserted the following sentences with an extra reference into the figure legends of Figure 4—figure supplement 2: "Replication timing scores are derived from E/L Repli-seq experiments, where cycling cells are pulse-labeled with BrdU, then sorted to early and late S-phase fractions by flow cytometry, and BrdU labeled genomic DNA fragments are pulled down and subjected to NGS. Signal tracks are computed from the read coverages in early over late S-phase samples, therefore the higher score means earlier replication (Marchal et al., 2018). The scale goes from -2.5 to 5." Similar description is inserted to the figure legend of Figure 4—figure supplement 3: "Replication timing scores are derived from E/L Repli-seq experiments, the higher score means earlier replication (Marchal et al., 2018)."

Please add discussion or speculate on the possible mechanism regarding why uracil enrichment upon treatment both appear at similar replication timing, relative to the control samples (Figure 4—figure supplement 3)?

This is already discussed in details in the following sections:

"Uracil appearance via thymine replacing misincorporation implies prior DNA synthesis involved in either replication, or transcription-coupled DNA repair, or epigenetic reprogramming (e. g. erasing the methyl-cytosine epigenetic mark). Importantly, we found that uracil pattern showed the highest correlation with the features (early replication, active promoters and DNase hypersensitive sites, and CpG islands) linked exactly to these processes (cf. Figure 4C)."

With even more details starting with: "The antifolate or nucleotide-based thymidylate synthase inhibitors, such as 5-FU, RTX or 5FdUR are known to lead to cell cycle arrest…"

We also inserted a short reference: "also supported by Figure 4—figure supplement 3".

Author details

1. Hajnalka L Pálinkás

   1. Genome Metabolism Research Group, Institute of Enzymology, Research Centre for Natural Sciences, Budapest, Hungary
   2. Department of Applied Biotechnology and Food Sciences, Budapest University of Technology and Economics, Budapest, Hungary
   3. Doctoral School of Multidisciplinary Medical Science, University of Szeged, Szeged, Hungary

   Contribution

   Conceptualization, Formal analysis, Investigation, Visualization, Methodology, Writing - original draft, Project administration, Writing - review and editing

   For correspondence

   palinkas.hajnalka@ttk.hu

   Competing interests

   No competing interests declared

2. Angéla Békési

   1. Genome Metabolism Research Group, Institute of Enzymology, Research Centre for Natural Sciences, Budapest, Hungary
   2. Department of Applied Biotechnology and Food Sciences, Budapest University of Technology and Economics, Budapest, Hungary

   Contribution

   Conceptualization, Data curation, Software, Formal analysis, Validation, Investigation, Visualization, Writing - original draft, Writing - review and editing

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0003-2294-3002

3. Gergely Róna

   1. Department of Applied Biotechnology and Food Sciences, Budapest University of Technology and Economics, Budapest, Hungary
   2. Department of Biochemistry and Molecular Pharmacology, New York University School of Medicine, New York, United States
   3. Perlmutter Cancer Center, New York University School of Medicine, New York, United States
   4. Howard Hughes Medical Institute, New York University School of Medicine, New York, United States

   Contribution

   Conceptualization, Investigation, Visualization, Writing - original draft, Writing - review and editing

   Competing interests

   No competing interests declared

4. Lőrinc Pongor

   1. Cancer Biomarker Research Group, Institute of Enzymology, Research Centre for Natural Sciences, Budapest, Hungary
   2. Department of Bioinformatics and 2nd Department of Pediatrics, Semmelweis University, Budapest, Hungary

   Contribution

   Data curation, Software, Formal analysis

   Competing interests

   No competing interests declared

5. Gábor Papp

   Department of Applied Biotechnology and Food Sciences, Budapest University of Technology and Economics, Budapest, Hungary

   Contribution

   Data curation, Software, Formal analysis, Visualization

   Competing interests

   No competing interests declared

6. Gergely Tihanyi

   1. Genome Metabolism Research Group, Institute of Enzymology, Research Centre for Natural Sciences, Budapest, Hungary
   2. Department of Applied Biotechnology and Food Sciences, Budapest University of Technology and Economics, Budapest, Hungary

   Contribution

   Investigation

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0003-3729-6709

7. Eszter Holub

   Department of Applied Biotechnology and Food Sciences, Budapest University of Technology and Economics, Budapest, Hungary

   Contribution

   Formal analysis

   Competing interests

   No competing interests declared

8. Ádám Póti

   Genome Stability Research Group, Institute of Enzymology, Research Centre for Natural Sciences, Budapest, Hungary

   Contribution

   Formal analysis, Visualization

   Competing interests

   No competing interests declared

9. Carolina Gemma

   Department of Surgery and Cancer, Imperial College London, Hammersmith Hospital Campus, London, United Kingdom

   Contribution

   Resources

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0002-4890-9972

10. Simak Ali

    Department of Surgery and Cancer, Imperial College London, Hammersmith Hospital Campus, London, United Kingdom

    Contribution

    Resources

    Competing interests

    No competing interests declared

    "This ORCID iD identifies the author of this article:" 0000-0002-1320-0816

11. Michael J Morten

    Department of Biochemistry and Molecular Pharmacology, New York University School of Medicine, New York, United States

    Contribution

    Investigation

    Competing interests

    No competing interests declared

12. Eli Rothenberg

    Department of Biochemistry and Molecular Pharmacology, New York University School of Medicine, New York, United States

    Contribution

    Supervision

    Competing interests

    No competing interests declared

    "This ORCID iD identifies the author of this article:" 0000-0002-1382-1380

13. Michele Pagano

    1. Department of Biochemistry and Molecular Pharmacology, New York University School of Medicine, New York, United States
    2. Perlmutter Cancer Center, New York University School of Medicine, New York, United States
    3. Howard Hughes Medical Institute, New York University School of Medicine, New York, United States

    Contribution

    Supervision

    Competing interests

    No competing interests declared

    "This ORCID iD identifies the author of this article:" 0000-0003-3210-2442

14. Dávid Szűts

    Genome Stability Research Group, Institute of Enzymology, Research Centre for Natural Sciences, Budapest, Hungary

    Contribution

    Supervision, Writing - review and editing

    Competing interests

    No competing interests declared

15. Balázs Győrffy

    1. Cancer Biomarker Research Group, Institute of Enzymology, Research Centre for Natural Sciences, Budapest, Hungary
    2. Department of Bioinformatics and 2nd Department of Pediatrics, Semmelweis University, Budapest, Hungary

    Contribution

    Supervision, Writing - review and editing

    Competing interests

    No competing interests declared

16. Beáta G Vértessy

    1. Genome Metabolism Research Group, Institute of Enzymology, Research Centre for Natural Sciences, Budapest, Hungary
    2. Department of Applied Biotechnology and Food Sciences, Budapest University of Technology and Economics, Budapest, Hungary

    Contribution

    Conceptualization, Resources, Supervision, Funding acquisition, Writing - original draft, Project administration, Writing - review and editing

    For correspondence

    vertessy@mail.bme.hu

    Competing interests

    No competing interests declared

    "This ORCID iD identifies the author of this article:" 0000-0002-1288-2982

• 2,225

  Page views

• 227

  Downloads

• 9

  Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.