Enzymatic Removal of Ribonucleotides from DNA Is Essential for Mammalian Genome Integrity and Development

1DNA is believed to have evolved from an ancestral RNA world as a more stable store of genetic information. Ribonucleotides differ from deoxynucleotides by the presence of a single reactive hydroxyl group at the 2′ position of the ribose sugar, rendering RNA ∼100,000-fold more susceptible to spontaneous hydrolysis under physiological conditions. The presence of ribonucleotides in genomic DNA is therefore undesirable, as it renders DNA more sensitive to strand breakage. It has long been thought that such misincorporation is prevented by the stringent selectivity of replicative DNA polymerases, favoring deoxynucleoside triphosphate (dNTP) over ribonucleoside triphosphate (rNTP) substrates. However, recent in vitro experiments have demonstrated that, under physiologically relevant conditions in which rNTPs substantially exceed dNTPs, such DNA polymerases may incorporate a ribonucleotide base every few thousand base pairs. Budding yeast expressing a less selective replicative polymerase only displayed widespread ribonucleotide incorporation when ribonuclease (RNase) H2 activity was genetically abolished. This directly implicated RNase H2 in the removal of such ribonucleotides.

2RNase H enzymes hydrolyze the RNA strand of RNA/DNA hybrids. Such hybrids form during many cellular processes, including DNA replication, telomere elongation, and transcription. Eukaryotes have two types of RNase H with distinct biochemical properties and substrate specificity (reviewed in ). RNase H1 is a processive monomeric enzyme that requires interaction with 2′-OH groups from four consecutive ribonucleotides for efficient substrate cleavage. Mammalian RNase H1 has two isoforms: a nuclear isoform of undefined function and a mitochondrial isoform that is essential for mitochondrial DNA replication. However, the predominant source of RNase H activity in mammalian cells is RNase H2. Like RNase H1, it digests the RNA strand of RNA/DNA hybrids in a processive manner, but it also recognizes single ribonucleotides in a DNA duplex and cleaves the 5′-phosphodiester bond of the ribonucleotide. In eukaryotes, RNase H2 is a multimeric complex consisting of three subunits: RNASEH2A, RNASEH2B, and RNASEH2C. The RNASEH2A subunit contains the catalytic center, whereas the closely intertwined auxiliary RNASEH2B and C subunits are likely involved in interactions with other proteins. A PIP box motif at the C terminus of the RNASEH2B subunit guides the interaction between RNase H2 and PCNA and its localization to replication foci, consistent with a role for the RNase H2 enzyme in DNA replication and/or repair.

3Mutations in all three genes that encode the RNase H2 subunits cause the autosomal-recessive disorder Aicardi-Goutières syndrome (AGS). This early-onset neuroinflammatory condition mimics congenital viral infection and has immunological similarities to the autoimmune disease systemic lupus erythematosus. RNase H2 mutations that cause AGS result in partial rather than absolute loss of RNase H2 enzyme function. Two further enzymes have been implicated in AGS: the 3′→5′ DNA exonuclease TREX1 and a dNTP triphosphohydrolase, SAMHD1. Innate immune-mediated inflammation is thought to result from the accumulation of endogenous nucleic acid species that are usually degraded by these enzymes, e.g., during DNA replication/repair or suppression of endogenous retroelements.

4The nucleic acids that may accumulate as a consequence of impaired RNase H2 function are yet to be defined, and although RNase H2 enzyme activity has been studied for more than 40 years, its cellular functions are poorly understood. Initially, RNase H2 was proposed to act in removal of the RNA oligonucleotides that prime Okazaki fragment synthesis during lagging-strand replication. In vitro biochemical studies indicate that sequential action of RNase H2 and FEN1 are sufficient to complete this process. However, primer removal through flap processing by FEN1/DNA2 has become the predominant model for Okazaki fragment maturation. RNase H2 may also be important for the resolution of R loops that arise during transcription or for the repression of endogenous retroelements. Finally, the distinctive property that allows RNase H2 to recognize and cleave single ribonucleotides that are embedded in DNA duplexes suggests a role in the removal of such nucleotides from genomic DNA.

5Here, we performed targeted mutagenesis of the mouse Rnaseh2b gene to gain insight into the in vivo role of the mammalian RNase H2 enzyme. Ablation of Rnaseh2b in mice leads to early embryonic lethality due to elevated DNA damage and reduced cellular proliferation during gastrulation. We establish that the growth arrest is the consequence of a p53-dependent DNA damage response associated with the accumulation of single ribonucleotides in genomic DNA. Thus, we demonstrate that removal of ribonucleotides to preserve genome integrity is an essential cellular function of RNase H2 in mammals.

6Rnaseh2bE202X embryonic stem (ES) cells with a premature stop codon in exon 7 of Rnaseh2b were generated by targeted homologous recombination (Figure 1A). Correct recombination of both arms of the targeting cassette was confirmed by Southern blotting and long-range PCR (Figures 1A and 1B) and the presence of a nonsense mutation at codon 202 (E202X) established by sequencing (Figure 1C). ES cells were injected into C57BL/6J host blastocysts to generate germline chimeras and subsequently heterozygous Rnaseh2bE202X/+ mice. Intercrossing of Rnaseh2bE202X/+ animals failed to yield live-born homozygous null mice (p < 0.001; Figures 1D and 1E). Similarly, homozygotes were not present (p < 0.001) in litters of a second independent line, Rnaseh2btm1a, derived from EUCOMM “knockout-first” ES cells.

7As no viable homozygous animals were obtained from either of these alleles, we concluded that Rnaseh2b was likely to be an essential gene that is required for embryonic viability. At embryonic day 6.5 (E6.5), Rnaseh2bE202X/E202X embryos were present at Mendelian ratios and were almost indistinguishable in size from wild-type littermates, suggesting normal progression of early embryogenesis (Figures 2A and 2B). However, by E7.5, there was a 23% and 32% decrease in embryonic height and width, respectively (p < 0.005; Figure 2B), suggesting a failure to increase the rate of proliferation in the epiblast that normally occurs at the onset of gastrulation. Though all mutants proceeded through gastrulation, the embryos remained reduced in size and were developmentally retarded (Figure 2A). By E9.5 (n = 23), they were frequently truncated (57%) with few or very small postcervical somites (52%) and defects in allantois development (48%). At E10.5, histology demonstrated increased numbers of cells with condensed or fragmented nuclei, and at E11.5, a terminal phenotype was evident with loss of tissue morphology and integrity (data not shown).

8To examine the effect of the Rnaseh2bE202X mutation on the RNase H2 enzyme, immunoblotting was performed with affinity-purified antibodies. As the premature stop codon is not in the final Rnaseh2b exon, it is expected to cause nonsense-mediated decay of the transcript. Consistent with this, we did not detect a truncated RNASEH2BE202X protein (Figure S1B available online). Also, full-length RNASEH2A, B, and C subunits were all undetectable in E9.5 Rnaseh2bE202X/E202X embryo lysates (Figure 2C), indicating that RNASEH2B is required for in vivo stability of the heterotrimeric complex. Furthermore, type 2 RNase H activity against a DNA duplex oligonucleotide with an embedded ribonucleotide was undetectable in Rnaseh2bE202X/E202X embryos (Figure 2D). In contrast, general RNase H activity was retained, albeit at <10% of wild-type embryo levels (Figure 2D), consistent with retention of normal RNase H1 activity. Given that RNase H2 is absent from Rnaseh2bE202X/E202X embryos, we hereafter refer to these embryos and their derivative cells as RNaseH2null and their genotype as Rnaseh2b−/−. Significantly, such embryos entirely lack detectable enzyme activity against isolated ribonucleotides embedded in DNA.

9The anti-RNase H2 antibody specifically detected RNase H2 when used for immunofluorescence microscopy, as staining was absent in null embryos (Figures S2A and S2B). Endogenous RNase H2 exhibited nuclear localization in early embryos, consistent with proposed roles in DNA replication and repair. Expression was observed in blastocysts (Figure 3A) and in all three embryonic layers during gastrulation (Figure 3B), reflecting a ubiquitous presence at early stages of development. Later in embryogenesis and postnatally, expression became more restricted to highly proliferative tissues, such as the subventricular zone during neurogenesis and perinatal hair follicles (Figures S2C–S2F). In adults, RNase H2 was present in proliferative tissues, including intestinal crypt epithelium, and testes (Figures S2G and S2H). Expression correlated most closely with the proliferation marker Ki67, suggesting that RNase H2 is preferentially expressed in actively cycling cells at all stages of the cell cycle. In support of this, western analysis from synchronized HeLa cells demonstrated uniform expression levels of all three RNase H2 subunits throughout the cell cycle (Figure S2I).

10RNaseH2null embryonic growth failure could be the consequence of inefficient DNA replication or activated DNA damage signaling. Embryonic tissues were therefore immunostained for histone H2A.X phosphorylated at serine 139 (pH2AX), a marker of DNA-damage response to double-strand breaks and arrested replication. Though no difference in pH2AX levels was observed between mutant and control blastocysts, at E6.5, there was a substantial increase in nuclear pH2AX staining in epiblast cells of RNaseH2null embryos (Figure 4A), coinciding with a period of rapid cell cycles of less than 6 hr.

11At E6.5, there was no alteration in the percentage of epiblast cells undergoing active DNA synthesis; however, by E7.5, a significant reduction was evident (Figures 4B and 4C). Reduced embryo growth appeared to be the result of arrested cell proliferation, rather than cell death, as no widespread increase in apoptosis was observed at E7.5 or E9.5 by activated caspase 3 immunostaining (data not shown).

12To investigate the molecular basis of this growth arrest, whole-genome expression analysis was performed using Illumina microarrays. Transcript levels of 197 genes were significantly upregulated in E9.5 mutant embryos, whereas 115 genes were downregulated when compared to age-matched controls (p < 0.05; > 1.5-fold change). Of note, the four genes with the greatest fold increase in expression (Figure 4D), Ccng1/Cyclin G1, Cdkn1a/p21, Phlda3, and Trp53inp1, were all targets of the p53 transcriptional activator, a key transducer of ATM/ATR DNA damage signaling. Increased Cyclin G1 and p21 expression was confirmed by qPCR and immunoblotting (Figure 4E). In addition, Rnaseh2b was found to be <0.02% of wild-type by qPCR, in keeping with nonsense-mediated decay of this transcript (Figure 4E). Although RNase H2 could have a role in suppressing expression of endogenous retroelements, no changes in retroelement transcript levels were identified, indicating that there was no widespread dysregulation of retroelements at this stage of embryonic development. Similarly, there was no transcriptional evidence of an immune-mediated process (data not shown).

13We postulated that the DNA damage observed in the E6.5 embryos led to p53 activation and reduced cellular proliferation as a consequence of cyclin G1 and p21-mediated cell-cycle arrest. In keeping with this, partial rescue of the RNaseH2null embryonic phenotype was observed in a p53−/− background (Figure S3). Loss of p53 also fully rescued defective cell proliferation (Figure 4F) in primary cultures of E10.5/E11.5 mesodermal tissue, performed to generate mouse embryonic fibroblast (MEF) cell lines. Rnaseh2b−/−;p53+/+ cells completely failed to proliferate in culture (2/2), whereas those derived from Rnaseh2b−/−;p53−/− embryos grew well (p < 0.001; 5/5), though at 64% of the rate of Rnaseh2b+/+;p53−/− MEFs (Figures 4F and 4G). Rnaseh2b−/−;p53+/− cells showed very limited growth, and though MEF cell lines were eventually established from three out of eight embryos, all were found to have then lost the remaining wild-type p53 allele (data not shown). In conclusion, loss of RNase H2 enzyme activity results in p53-mediated arrest of cell growth.

14Rnaseh2b−/−;p53−/− MEFs were next used to investigate the molecular consequences of RNase H2 loss. As expected, type 2 RNase H activity and the RNase H2 protein complex were absent in these cell lines (Figures S4A and S4B). Significantly elevated levels of pH2AX foci were also present (Figures 5A and 5B), indicative of DNA double-strand breaks and/or replication fork arrest. In addition, we examined polyADPribosylation (PAR) as another early marker of DNA-damage activation, which occurs in response to DNA breaks. Levels of PAR were substantially raised in RNaseH2null MEFs, as shown by immunoblotting (Figure S4C), confirming the presence of cellular DNA-strand breaks.

15DNA damage in RNaseH2null cells might arise as the consequence of undegraded Okazaki RNA primers, misincorporation of ribonucleotides by DNA polymerases, or transcriptionally induced R loops. To discriminate between these possibilities, total nucleic acids from the MEFs were separated by gel electrophoresis under alkaline conditions. Substantially increased mobility of genomic DNA from both RNaseH2null cell lines after treatment with alkali was observed relative to control genomic DNA (Figures 5C–5E), whereas no significant difference was evident upon electrophoresis under neutral conditions (Figure S4D). Given that phosphodiester bonds 3′ of ribonucleotides, but not deoxynucleotides, are sensitive to alkali hydrolysis through nucleophilic attack by the 2′ hydroxyl group, such increased fragmentation (termed alkali sensitivity) likely indicates the incorporation of ribonucleotides into genomic DNA. Given that alkali treatment also denatures DNA, increased electrophoretic mobility could also be the consequence of increased nicking, or gaps, in genomic DNA. To address this possibility, gel electrophoresis of total cellular nucleic acids was performed after formamide denaturation, which denatures DNA without hydrolyzing ribonucleotide phosphodiester bonds (Figure 6A). Under these conditions, RNaseH2null DNA did not demonstrate increased mobility, in contrast to DNA treated with Nt.BspQI nicking endonuclease, which on average introduces nicks every ∼11 kb. Therefore, the observed alkali-sensitive sites were consistent with covalently incorporated ribonucleotides, rather than gaps or nicks in genomic DNA.

16Treatment with recombinant RNase H2 enzyme also led to widespread fragmentation of RNaseH2null genomic DNA, as shown by increased electrophoretic mobility after formamide denaturation. Significantly, the resulting fragmentation pattern was essentially identical to that seen after alkali treatment (Figure 6B), whereas no increased mobility was observed when inactive recombinant RNase H2 was used. In distinct contrast, recombinant E. coli type-1 RNase H (RNase HI) had no visible impact on the mobility of RNaseH2null genomic DNA (Figure 6C). RNase HI efficiently digests substrates with four or more ribonucleotides but can also hydrolyze double-stranded nucleic acids with three embedded ribonucleotides, albeit at substantially lower rates. Although the biological significance of this activity is unclear, oligonucleotide substrates with three ribonucleotides were cleaved efficiently under our assay conditions (Figures S5C and S5D). This activity was fully preserved in the presence of RNaseH2null genomic DNA, ruling out any inhibitory effects that may be present within the nucleic acid preparation (Figures S5A–S5C). Thus, the differential sensitivity of genomic DNA from RNaseH2null cells to RNase HI and RNase H2 activities established that the alkali and RNase H2 cleavable sites consist of one or, at most, two consecutive covalently incorporated ribonucleotides.

17Quantitative analysis of the alkali-induced fragmentation permitted us to estimate the frequency of embedded ribonucleotides. Determination of DNA fragment distributions (Figure 5E) from the densitometry data (Figure 5D) predicted a rate of ribonucleotide incorporation of ∼1 in 7,600 nucleotides (nt) (analytical method described in Figure S4F). Additionally, the fragmentation pattern of hydrolyzed RNaseH2null DNA lies between that generated by two nicking endonucleases, one that cuts, on average, once every 11 kb (7-cutter Nt.BspQI) and the other once every 3.7 kb (6-cutter Nb.BtsI) in the mouse reference genome (Figures 5F and 5G), supporting the computational estimate. Early and late passage MEFs exhibited similar levels of alkali sensitivity (data not shown). The incorporation of 1 ribonucleotide every 7,600 nucleotides during each round of replication would maintain such a steady-state level, so this is a minimum estimate for in vivo ribonucleotide misincorporation by polymerases.

18Taken together, these results demonstrate the widespread presence of incorporated ribonucleotides in genomic DNA of RNaseH2null MEFs. Ribonucleotide misincorporation was also evident in RNaseH2null embryos (Figures 5C and S4E), consistent with this molecular defect underlying the developmental phenotype. Finally, we employed a chemical genetic approach analogous to previous yeast genetic experiments, in which a mutated Pol ɛ with an enhanced propensity for ribonucleotide incorporation was used. Cells were treated with a low dose of hydroxyurea (HU) to reduce cellular dNTP:rNTP ratios through partial inhibition of ribonucleotide reductase activity. This would then increase ribonucleotide incorporation by DNA polymerases. RNaseH2null cells were observed to be hypersensitive to such a low dose of HU, accumulating in S phase (Figure S5G). Most significantly, such treatment resulted in additional alkali and RNase H2-sensitive sites in genomic DNA from RNaseH2null cells (Figures 6D–6F, S5E, and S5F), consistent with such sites resulting from ribonucleotide misincorporation by DNA polymerases.

19Substantial levels of micronuclei, indicative of chromosomal breakage, were observed in RNaseH2null MEFs (Figure 7A), suggesting that the excessive presence of ribonucleotides in DNA causes large-scale genome instability. Likewise, large-scale cytogenetic anomalies were evident in DAPI-stained metaphase chromosomes (Figure 7B). Using satellite FISH probes, chromosomal rearrangements were present in virtually all metaphases from Rnaseh2b−/−;p53−/− MEFs, whereas they were only occasionally seen in control MEFs. Both minutes (marker chromosomes) and interchromosomal translocations were frequently observed, with translocations confirmed by FISH and chromosomal painting (Figures 7, S6A, and S6B).

20Here, we report that substantial genome-wide incorporation of ribonucleotides occurs in mammalian genomic DNA and establish that RNase H2 is required for efficient removal of such nucleotides. Recent in vitro biochemical studies with yeast replicative polymerases have shown the misincorporation of ribonucleotides into DNA. Based on these findings, it was predicted that ribonucleotides may be incorporated into genomic DNA in vivo. Subsequent studies in both fission and budding yeast have established that this is indeed the case using alkali sensitivity assays of yeast genomic DNA. Here, we show that ribonucleotide incorporation also occurs in metazoans; demonstrate that such ribonucleotide lesions are harmful to mammalian cells; and establish that their removal is required for mouse embryonic development. Previous studies used an elegant genetic approach in which mutator replicative polymerases with increased propensity for ribonucleotide incorporation were used to infer the presence of incorporated ribonucleotides from enhanced alkali sensitivity of RNase H2 null yeast genomic DNA. Our findings provide further characterization of these alkali-sensitive sites by using enzymatic assays to directly substantiate that such lesions are single or diribonucleotides that are covalently incorporated into genomic DNA. Furthermore, we find that such lesions occur at a frequency of least 1,000,000 sites per cell, establishing them as the most common endogenous base lesions in the mammalian genome.

21The presence of such ribonucleotides is most readily explained by misincorporation by the major replicative polymerases, which are estimated to incorporate one ribonucleotide every few thousand nucleotides in vitro. Alternatively, embedded ribonucleotides could result from failure to remove RNA primers during Okazaki-lagging strand processing. Such primers are ∼10 nt in length, much longer than the single/diribonucleotides found. However, retention of single ribonucleotides during this process remains a possibility. Theoretically, oxidation of deoxynucleotides that are present in DNA could also result in embedded ribonucleotides, although this seems an unlikely explanation for their frequent occurrence.

22Genomic DNA from RNase H2 null S. cerevisiae exhibits differential alkali sensitivity that correlates with the propensity of a mutant Pol ɛ to incorporate ribonucleotides. Here, we performed an analogous experiment using hydroxyurea to alter dNTP:rNTP ratios, favoring ribonucleotide incorporation. This promoted increased alkali and RNase H2 sensitivity, leading us to conclude that ribonucleotides that are embedded in genomic DNA are most likely the consequence of misincorporation by DNA polymerases. Given their frequent occurrence in genomic DNA, the predominant sources of such ribonucleotides are likely to be Pol ɛ and Pol δ, the major replicative polymerases that are responsible for leading- and lagging-strand synthesis, respectively. However, other DNA polymerases, such as Pol β and lesion bypass polymerases, may also contribute.

23Ribonucleotide accumulation in genomic DNA of RNaseH2null mice (Figures 5 and 6) implicates the RNase H2 complex in the maintenance of genome integrity. This DNA repair function was originally suggested by Eder and colleagues. Ribonucleotides are likely to be harmful, as their ribose 2′-hydroxyl group increases susceptibility of the adjacent phosphodiester bond to hydrolysis. Specific patterns of mutations at nucleotide level have been observed in genomic DNA from RNase H2 null S. cerevisiae. Most frequently, these consist of 2–5 bp deletions, which are the result of topoisomerase-I-induced nicks at embedded ribonucleotides. However, the loss of yeast RNase H2 alone has a relatively small effect on mutation frequency, and the effect of embedded ribonucleotides on large-scale genome stability in yeast has not been reported. Therefore, the frequent occurrence of large-scale genome rearrangements in RNaseH2null MEFs is unanticipated and noteworthy (Figure 7).

24We estimate that ribonucleotides are incorporated at a rate of at least 1 every 7,600 nt in RNaseH2null cells, corresponding to ∼1,300,000 lesions per cell. This is within the same order of magnitude predicted from in vitro incorporation rates by eukaryotic replicative polymerases and is substantially higher than any other endogenous base lesions occurring in the mammalian genome. Even the previously most common lesions, such as abasic sites, 8-hydroxyguanine (8-oxoG), and 7-methylguanine, only occur up to 10,000 times per genome. As misincorporated ribonucleotides occur at at least 50-fold higher rates, without an efficient repair mechanism, they would be the most common noncanonical nucleotides present in mammalian DNA. Therefore, defining the processes that remove these ribonucleotides is of substantial interest. Of note, FEN1/Rad27 in conjunction with RNase H2 can excise ribonucleotides on an in vitro substrate, generating a single nucleotide gap on which a DNA polymerase and DNA ligase could act directly to repair the lesion.

25In itself, ribonucleotide incorporation does not prevent replication: cellular proliferation is seen in both RNase H2 null mouse cells (p53−/−; Figure 4) and RNase H2 null budding yeast, in which p53 signaling is not evolutionarily conserved. DNA polymerases can tolerate templates containing ribonucleotides, which may explain why early embryogenesis in RNaseH2null embryos proceeds normally. The absence of grossly perturbed transcriptional profiles later in development (Figure 4) suggests that mammalian RNA polymerases also tolerate ribonucleotide-containing templates.

26However, excessive numbers of ribonucleotides do appear to be problematic. Replication fork stalling may occur in regions that contain clustered ribonucleotides, as seen at the S. pombe mating switch locus. Incorporation of ribonucleotides in difficult to replicate regions or in close proximity to other lesions may be similarly detrimental. This is likely to explain the activation of DNA-damage response signaling observed in RNaseH2null MEFs and embryos (Figures 4 and 5). Chromosomal rearrangements and micronuclei indicate the occurrence of double-strand DNA breaks. Such breaks may result from subsequent replication fork collapse or may be caused directly by hydrolysis of ribonucleotides on opposing DNA strands (see model in Figures S6C and S6D). Alternatively, the observed increase in PAR (Figure S4C) could suggest the presence of frequent single-strand breaks that would be converted at low frequency to double-strand lesions during replication. The accumulation of ribonucleotides in conjunction with rapid cell cycles in the epiblast probably underlies the marked activation of DNA-damage signaling in the embryo. This then results in a p53-mediated inhibition of proliferation that is likely to substantially contribute to the lethality observed at E11.5 in RNaseH2null embryos.

27To our knowledge, stable incorporation of ribonucleotides has only been reported to date in two contexts. First, a diribonucleotide at the S. pombe mating switch locus is believed to be the signal initiating homologous recombination. Second, the presence of ribonucleotides in mature mitochondrial DNA has been previously established, and we now show these to be mono or diribonucleotides (Figure S5H). These sporadic ribonucleotides appear to be randomly distributed and thus are likely to result from replicative polymerase incorporation. The selectivity of the mitochondrial polymerase γ would be consistent with the presence of 10–30 ribonucleotides in mature mtDNA. These may be tolerated by the mitochondrial genome either because of its relatively slow replication rate or owing to different mechanism(s) of genome replication. Likewise, ribonucleotide incorporation is well tolerated in RNase H2-deficient S. cerevisiae, with normal viability and efficient cellular proliferation in unperturbed cells. As recently reported, template switch and translesion DNA synthesis postreplication repair pathways may permit such ribonucleotides to be tolerated. In contrast, in mice, we find that ribonucleotide removal is essential early in development. Similarly, mutation of other genes ensuring genome integrity, such as the catalytic subunit of pol ζ (Rev3), are viable in yeast but cause embryonic lethality at a similar stage in mutant mice. Such lethality may therefore be explained by the much larger size and complexity of the mammalian nuclear genome.

28Low levels of ribonucleotide incorporation in the nuclear genome may be tolerated, and this could well be relevant to the autoinflammatory disorder Aicardi-Goutières syndrome (AGS), in which aberrant nucleic acid substrates are thought to drive an innate immune response. Reduced RNase H2 activity in AGS may therefore result in a chronic low level of ribonucleotide incorporation that is then processed by alternative (non-RNase H2 dependent) repair pathways. The increased levels of polyADPribosylation (Figure S4C), as well as the enhanced sensitivity to hydroxyurea observed in Rnaseh2b+/− MEFs (Figure S5G), could be consistent with this possibility. Aberrant nucleic acid species generated by such repair could then trigger an innate immune response. Alternatively, embedded ribonucleotides might induce DNA-damage response signaling that itself stimulates interferon production.

29In summary, ribonucleotides are highly deleterious to the mammalian cell when left incorporated in the nuclear genome, causing substantial genome instability. RNase H2 is therefore a critical enzyme for ensuring the integrity of genomic DNA. Defining the pathway(s) that remove these ribonucleotides from genomic DNA, the site and nature of ribonucleotide-induced DNA damage, as well as the genome distribution of ribonucleotides will now be of substantial interest. This will help improve our understanding of the pathological and physiological roles of ribonucleotides in genomic DNA, of significance to both nucleic acid-driven autoimmunity and carcinogenesis.

30While performing targeted homologous recombination to generate 129/Ola ES cells with a c.520G > A (A174T) mutation in exon 7 of Rnaseh2b (that corresponds to the most common, hypomorphic AGS patient mutation, c.529G > A), we identified the Rnaseh2bE202X ES clone that had fortuitously acquired an additional mutation resulting in a premature stop codon: c.604G > T, E202X. Further details of gene targeting are available in the Supplemental Information. After karyotyping and blastocyst injection, resulting male chimeras were crossed to C57BL/6J females, giving rise to heterozygous Rnaseh2b knockin mice carrying both A174T and E202X mutations (Rnaseh2btm1-hgu-A174T,E202X, elsewhere referred to as Rnaseh2bE202X). Crosses with p53+/− mice were used to generate Rnaseh2bE202X/+;p53+/− double mutants. Knockout-first Rnaseh2b mice were generated by blastocyst injection of the Rnaseh2btm1a(EUCOMM)Wtsi ES cell clone EPD0087_4_A02 (EUCOMM ID: 24441; elsewhere referred to as Rnaseh2btm1a). The Rnaseh2btm1a allele is designed to prematurely truncate Rnaseh2b transcripts through targeted insertion into intron 4 of a genetrap cassette that contains a strong splice acceptor and an efficient polyadenylation signal.

31Genotypes of mice and embryos were determined by multiplex PCR. The status of early embryos was determined by α-RNase H2 immunofluorescence. For further details and primer sequences, see Supplemental Information and Table S1.

32Enzyme activity assays were performed in triplicate using a FRET-based fluorescent substrate release assay as previously described using 100 ng/μl of total protein from whole-cell extracts. RNase H2-specific activity was determined by subtracting the cellular activity against a sequence-matched DNA duplex without ribonucleotides.

33Immunoblotting was performed on whole-cell extracts as described previously. For immunohistochemistry, tissues and deciduas were dissected into ice-cold PBS and fixed with 4% paraformaldehyde/PBS for 3–16 hr at 4°C with further processing performed by standard methods. Images were collected on Zeiss Axioplan II fluorescence or Nikon A1R confocal microscopes. For full experimental details, see the Supplemental Information; for antibodies and dilutions, see Table S2.

34Total nucleic acids were isolated by mechanical disruption of MEFs or yolk sacks in ice-cold lysis buffer (20 mM Tris-HCl [pH 7.5], 75 mM NaCl, and 50 mM EDTA) and subsequent incubation with 100 μg/ml proteinase K, with Sarcosine then added to final 1% concentration. Nucleic acids were sequentially extracted with TE-equilibrated phenol, phenol:chloroform:isoamylalcohol (25:24:1), and chloroform; precipitated with isopropanol; washed with 75% ethanol; and dissolved in water. Mitochondrial DNA (mtDNA) was isolated from sucrose-gradient purified mitochondria as previously described. For alkaline gel electrophoresis, total nucleic acids were heated for 2 hr at 55°C with 0.3 M NaOH and separated on agarose gels (50 mM NaOH, 1 mM EDTA) as previously described. Control samples were heated with 0.3 M NaCl and separated on 0.5× TBE agarose gels. Alternatively, nucleic acids were treated with RNase H enzymes and heated for 30–60 min in 90% formamide/20 mM EDTA at 37°C before separation on 0.5× TBE agarose gels. Digestions with RNase HI were carried out in 100 μl of 1× reaction buffer (NEB) with 5 U of enzyme for 1 hr at 37°C. Digestions with RNase H2 were carried out for 1 hr at 37°C in 100 μl reaction buffer (50 mM Tris [pH 8], 60 mM KCl, 10 mM MgCl2, 0.01% BSA, 0.01% Triton) using 10 nM of purified recombinant human RNase H2. Nucleic acids were ethanol precipitated and dissolved in 90% formamide/20 mM EDTA. After electrophoresis, gels were stained with SYBR Gold (Invitrogen) or ethidium bromide.

35A genomic fragment of 4.5 kb, which included exon 6 and 7 of the mouse Rnaseh2b locus (nucleotides 62977197-62981727 of Chr 14, Ensembl release 64), was retrieved by gap repair from BAC bMQ454F14 (Source Bioscience Lifesciences), and a floxed neomycin cassette was inserted between exon 6 and 7 by bacterial recombination as previously described, producing a targeting vector with two external homology arms of 2.8 kb and 1.7 kb. A point mutation c.520G > A (A174T) was introduced into Exon 7 of mouse Rnaseh2b by site-directed-mutagenesis (Quikchange II XL site directed mutagenesis kit, Agilent Technologies). The insert was removed from the vector backbone by digestion with NotI and SalI, purified by electroelution using an Elutrap (Schleicher & Schuell), and electroporated into 129/Ola E14Tg2AIV embryonic stem cells. Correctly targeted clones were selected by Southern blotting and long-range PCR. Exons 6 and 7 were sequenced and in one clone an additional c.613G > T mutation identified, which resulted in a premature stop codon (E202X). This ES cell clone was then karyotyped prior to injection into mouse blastocysts. All mouse studies were conducted under guidance and approval issued by the UK Medical Research Council and the UK Home Office.For mouse genotyping, ear-clips were boiled in 50 μl 25 mM NaOH, 0.2 mM EDTA for 30 min at 95°C, cooled, then neutralized with 50 μl 40 mM Tris base and used directly in PCR reactions. For genotyping of embryos whole embryos, tails or yolk sacs were treated with DirectPCR Lysis Reagent (Viagen) according to the manufacturer's instructions. All genotyping PCRs were performed using a ‘multiplex’ three primer strategy and Taq ReddyMix PCR Master Mix (Thermo Scientific). For primer sequences, PCR programs and expected product sizes see Table S1. To genotype early embryos used for immunofluorescence, cryosections were stained with polyclonal affinity purified rabbit α-RNase H2 antiserum (raised against recombinant mouse RNase H2 complex; see Table S2). Embryos negative for RNase H2 staining were designated RNaseH2null; those positive for staining were used as RNase H2 positive controls.MEFs were isolated from individual E10.5 or E11.5 embryos after removing the head, by mincing in 1 ml of medium (DMEM, 10% FCS, 50 U/ml penicillin and 50 μg/ml streptomycin, 1 mM β-Mercaptoethanol). Resulting suspensions were grown at 37°C, 5% CO2 and 3% O2, and non-adherent cells removed after 24 hr.Enzyme activity assays were performed using a FRET-based fluorescent substrate release assay. 10 μM of 3′ fluorescein-labeled oligonucleotides (GATCTGAGCCTGGGaGCT for RNase H2 specific activity, or gatctgagcctgggagct for total RNase H activity; uppercase DNA, lowercase RNA) were annealed to a complementary 5′ DABCYL-labeled DNA oligonucleotide (Eurogentec) in 60 mM KCl, 50 mM Tris-HCl pH 8, by heating for 5 min at 95°C followed by slow cooling to room temperature. Reactions were performed in 100 μl of buffer (60 mM KCl, 50 mM Tris-HCl pH 8, 10 mM MgCl2, 0.01% BSA, 0.01% Triton X-100) with 250 nM substrate in 96-well flat-bottomed plates at 24 ± 2°C. Fluorescence was read for 100 ms using a VICTOR2 1420 multilabel counter (Perkin Elmer), with a 480 nm excitation filter and a 535 nm emission filter.HeLa cells were synchronized at the G1/S border by double thymidine block: 30%–40% confluent HeLa cells were incubated for 19 hr with 2 mM of thymidine, washed three times with PBS, incubated in fresh media for 9 hr, and incubated a second time with thymidine for 15 hr at 37°C, 5% CO2, 21% O2. Cells were released from G1/S arrest by removal of thymidine, and harvested at 2 hr intervals for immunoblotting analysis. The cell cycle phase at each time point was determined by FACS analysis of propidium iodide stained cells fixed in 70% ethanol.For section immunohistochemistry, tissues and decidua were dissected into ice-cold PBS and fixed with 4% paraformaldehyde (PFA)/PBS for 3-16 hr at 4°C. After three 30 min washes with PBS, the samples were passed through a sucrose series, embedded, then frozen in OCT embedding medium (Sakura) and cryosectioned at 10 μM. Frozen sections were post-fixed in ice cold acetone 10 min at −20°C and air-dried prior to use; then washed three times in TBST (1x TBS, 0.1% Triton X-100). Sections were blocked with 10% donkey serum/TBST/1mM NaN3 for 1 hr at room temperature. Sections were incubated with diluted primary antibody (Table S2) in 1% donkey serum/TBST/1mM NaN3 for 14-16 hr at 4°C. Slides were washed three times 10 min with TBST before incubating with secondary antibody (Table S2) diluted in 1% donkey serum/TBST/1mM NaN3 for 4 hr at room temperature with TOTO-3 (Invitrogen, 1:400) and RNase A (50 μg/ml). Samples were washed three times 10 min in TBST and mounted in Prolong Gold (Molecular Probes).For cell proliferation studies, pregnant dams and postnatal animals were given a single intraperitoneal injection of EdU (Invitrogen) at a dose of 50-100 mg/kg body weight in a solution of 10 mg/ml PBS (pH 7.4). After 1-2 hr, animals were sacrificed and tissues harvested and fixed in 4% PFA/PBS with gentle mixing for 3 hr at 4°C and processed as described above. Click-iT reactions (Invitrogen) were performed on cryosections as per manufacturer's instructions.For whole-mount immunohistochemistry of blastocysts, E2.5 embryos were collected from Rnaseh2b+/− x Rnaseh2b+/− plugs and cultured overnight in M16 media (Sigma) to blastocyst stage. Blastocysts were then fixed in 4% PFA/PBS for 20 min and transferred to depression slides for subsequent steps. Blastocysts were washed five times 5 min in PBST (1x PBS, 0.1% Triton X-100), permeabilized in PBS/0.25% Triton X-100 for 20 min and washed five times 5 min in PBST. Blastocysts were blocked in 10% donkey serum/TBST/1 mM NaN3 for 2 hr at room temperature and incubated 14-16 hr in primary antibody (Table S2) diluted in 1% donkey serum/TBST/1 mM NaN3 in a humidified chamber at 4°C. Samples were washed three times 10 min in TBST, prior to incubating with secondary antibody (Table S2) diluted in 1% donkey serum/TBST/1 mM NaN3 for 4 hr at room temperature with TOTO-3 (Invitrogen, 1:400) and RNase A (50 μg/ml). Samples were washed three times 10 min in TBST and mounted on a Superfrost slide between 9 mm Secure-Seal spacers (Molecular Probes) and no. 0 coverslips in Prolong Gold (Molecular Probes). Details of antibodies and dilutions used in this study are given in Table S2.Images were collected on a Zeiss Axioplan II fluorescence microscope or on a Nikon A1R confocal microscope, comprised of a Nikon Eclipse TiE motorized inverted microscope with Perfect Focus System. Image capture and analysis were performed using IPLab Spectrum (Scanalytics Corp) driving a Coolsnap HQ CCD camera (Photometrics Ltd) for the Zeiss system. For the confocal, image capture and analysis was performed using Nikon NIS Elements AR.RNA was extracted from individual embryos using TRIzol reagent (Invitrogen), with further purification using RNeasy Mini Kit columns (QIAGEN) and RNase-free DNase I (QIAGEN). cRNA was generated using an Illumina TotalPrep RNA Amplification Kit (Ambion) and whole-genome gene expression analysis of biological triplicates performed using MouseWG-6 v2.0 Expression BeadChips (Illumina). MIAME compliant microarray data was deposited at the Gene Expression Omnibus (GEO, http://www.ncbi.nlm.nih.gov/geo/) under the accession number GSE37419.Microarray data were analyzed with R 2.14.0, using its beadarray and Limma 3.10.2 packages. Raw, non-normalized bead-summary values were imported from the Illumina BeadStudio software into R using the beadarray package. Quantile normalization was applied to the data to enable comparison between arrays. A linear model was applied to the expression data for each gene. To determine statistically differentially expressed genes the results of the linear model were summarized and a Bayes moderated t test applied. To control for multiple testing, a Benjamini and Hochberg false discovery rate p-value of < 0.05 was used.10 μl reactions consisting of 1 μl template cDNA, 1 X Brilliant II Sybr Green qPCR Master Mix (Stratagene), 0.3 μM passive reference dye (ROX, Stratagene) and either 0.1 μM of each oligonucleotide primer (Ccng1, Cdkn1a) or 0.06 μM of each primer (Actb) in nuclease-free water. Each sample and -RT were analyzed in triplicate and a no template control was include for each master mix. Quantitative real-time PCR was performed using an ABI Prism HT7900 Sequence Detection System (Applied Biosciences) with the following protocol: 50°C (2 min), 95°C (10 min) followed by 40 cycles of 95°C (15 s), 60°C (1 min). The cycle threshold (CT) values for each gene were used to calculate the expression of each target gene normalized to beta-Actin using the comparative CT method. Specificity of the primers was confirmed by agarose gel electrophoresis of the PCR products.Numeric and simulation analysis was performed in R (version 2.10.1). Peaks in the size reference lane were identified under supervision and the linear model lm(y∼log(x)) fitted to produce an electrophoretic distance (y) to nucleotide size (x) calibration curve. Densitometric histograms of electrophoresis lanes (e.g., Figure 5D) corresponding to alkali-treated genomic DNA were smoothed by fitting the smooth.spline function with 40 degrees of freedom. These smoothed distributions were transformed into nucleotide coordinate based histograms using a calibration curve from the same gel derived from 1 kb size standards (Invitrogen), extrapolated where necessary. Histograms were scaled such that they represented genomic fragment size distributions summing to 109 nucleotides. Direct subtraction of the number of genomic fragments in each histogram provides an approximate analytic estimate of the difference in fragment number between lanes. Simulation studies were based on the fragmentation of 109 nucleotide genomes. Random cuts were added to a starting genome, either cut at random or with a starting distribution of cuts estimated from densitometric data as above. A hill climbing strategy with 50 iterations was employed to find the optimal number of random cuts added, to achieve the best fit with a target histogram (e.g., turning the Rnaseh2b+/+ densitometric distribution into the Rnaseh2b−/− distribution). The objective function for optimization was to minimize the Manhattan distance between the target histogram and the histogram of the simulated data after smoothing as above.Mouse embryonic fibroblasts (106) were seeded into T175 flasks and incubated for 14-16 hr at 37°C, 5% CO2 and 3% O2. Cells were cultured for another 48 hr in the absence or presence of hydroxyurea (300 μM). Cells were subsequently trypsinized, washed in PBS and fixed in 70% ice cold ethanol. Fixed cells were pelleted and resuspended in 2 ml pre-warmed pepsin (Sigma) dissolved at 1 mg/ml in 30 mM HCl and incubated at 37°C for 30 min with frequent mixing. Cell nuclei were pelleted at 3350 g for 4 min and resuspended in 100 μl 0.1 mg/ml RNase A diluted in PBS-EDTA before the addition of 100 μl propidium iodide (100 μg/ml) diluted in PBS-EDTA. Each sample was incubated at 4°C for a minimum of 1 hr before acquisition on a BD Biosciences FACSAriaII. Data was analyzed using FlowJo software (v7.6.1, Tree Star).For 2D FISH, metaphases were isolated in hypotonic buffer, fixed with methanol:acetic acid (3:1) and dropped onto slides. Slides were incubated with 100 μg/ml RNase A in 2x SSC for 1 hr at 37°C, washed in 2x SSC and dehydrated through an alcohol series, followed by denaturation in 70% formamide/2x SSC at 70°C. Major (pSAT) and minor (R198) satellite probes were prepared and labeled by nick translation with digoxigenin-11-dUTP or biotin-16 dUTP. Probes (100 ng per slide) were mixed with 5 μg of sonicated salmon sperm DNA, denatured and hybridized to slides for 14-16 hr at 37°C. For chromosome 4 painting, 15 μl per slide of Ready-to-use use paint (Cambio) was denatured at 70°C for 5 min, re-annealed at 37°C for 15 min and hybridized to denatured slides for 14-16 hr at 37°C. Washes and detection were performed as previously described. Two-dimensional slides were examined using a Zeiss Axioplan II fluorescence microscope with Plan-neofluar objectives, a 100 W Hg source (Carl Zeiss) and Chroma #8300 triple band pass filter set (Chroma Technology Corp.) with the excitation filters installed in a motorized filter wheel (Ludl Electronic Products). Grayscale images were captured with a Hamamatsu Orca AG CCD camera (Hamamatsu Photonics Ltd.). Image capture and analysis were performed using in-house scripts written for IPLab Spectrum (Scanalytics Corp.).For embryonic and live-born viability, categorical data was analyzed using Chi-square tests, under the null hypothesis that mutant mice would be present at Mendelian ratios. All quantitative data was analyzed using Student's t tests, under the assumption that data followed a Normal distribution. For cell counting experiments (e.g., Figure 4C, 5B and 7A), the percentage of positive cells was calculated for each experiment. This continuous measure was then analyzed by unpaired t tests, to establish if differences between cell lines were statistically significant, allowing for inter-experimental variation.

36A genomic fragment of 4.5 kb, which included exon 6 and 7 of the mouse Rnaseh2b locus (nucleotides 62977197-62981727 of Chr 14, Ensembl release 64), was retrieved by gap repair from BAC bMQ454F14 (Source Bioscience Lifesciences), and a floxed neomycin cassette was inserted between exon 6 and 7 by bacterial recombination as previously described, producing a targeting vector with two external homology arms of 2.8 kb and 1.7 kb. A point mutation c.520G > A (A174T) was introduced into Exon 7 of mouse Rnaseh2b by site-directed-mutagenesis (Quikchange II XL site directed mutagenesis kit, Agilent Technologies). The insert was removed from the vector backbone by digestion with NotI and SalI, purified by electroelution using an Elutrap (Schleicher & Schuell), and electroporated into 129/Ola E14Tg2AIV embryonic stem cells. Correctly targeted clones were selected by Southern blotting and long-range PCR. Exons 6 and 7 were sequenced and in one clone an additional c.613G > T mutation identified, which resulted in a premature stop codon (E202X). This ES cell clone was then karyotyped prior to injection into mouse blastocysts. All mouse studies were conducted under guidance and approval issued by the UK Medical Research Council and the UK Home Office.

37For mouse genotyping, ear-clips were boiled in 50 μl 25 mM NaOH, 0.2 mM EDTA for 30 min at 95°C, cooled, then neutralized with 50 μl 40 mM Tris base and used directly in PCR reactions. For genotyping of embryos whole embryos, tails or yolk sacs were treated with DirectPCR Lysis Reagent (Viagen) according to the manufacturer's instructions. All genotyping PCRs were performed using a ‘multiplex’ three primer strategy and Taq ReddyMix PCR Master Mix (Thermo Scientific). For primer sequences, PCR programs and expected product sizes see Table S1. To genotype early embryos used for immunofluorescence, cryosections were stained with polyclonal affinity purified rabbit α-RNase H2 antiserum (raised against recombinant mouse RNase H2 complex; see Table S2). Embryos negative for RNase H2 staining were designated RNaseH2null; those positive for staining were used as RNase H2 positive controls.

38MEFs were isolated from individual E10.5 or E11.5 embryos after removing the head, by mincing in 1 ml of medium (DMEM, 10% FCS, 50 U/ml penicillin and 50 μg/ml streptomycin, 1 mM β-Mercaptoethanol). Resulting suspensions were grown at 37°C, 5% CO2 and 3% O2, and non-adherent cells removed after 24 hr.

39Enzyme activity assays were performed using a FRET-based fluorescent substrate release assay. 10 μM of 3′ fluorescein-labeled oligonucleotides (GATCTGAGCCTGGGaGCT for RNase H2 specific activity, or gatctgagcctgggagct for total RNase H activity; uppercase DNA, lowercase RNA) were annealed to a complementary 5′ DABCYL-labeled DNA oligonucleotide (Eurogentec) in 60 mM KCl, 50 mM Tris-HCl pH 8, by heating for 5 min at 95°C followed by slow cooling to room temperature. Reactions were performed in 100 μl of buffer (60 mM KCl, 50 mM Tris-HCl pH 8, 10 mM MgCl2, 0.01% BSA, 0.01% Triton X-100) with 250 nM substrate in 96-well flat-bottomed plates at 24 ± 2°C. Fluorescence was read for 100 ms using a VICTOR2 1420 multilabel counter (Perkin Elmer), with a 480 nm excitation filter and a 535 nm emission filter.

40HeLa cells were synchronized at the G1/S border by double thymidine block: 30%–40% confluent HeLa cells were incubated for 19 hr with 2 mM of thymidine, washed three times with PBS, incubated in fresh media for 9 hr, and incubated a second time with thymidine for 15 hr at 37°C, 5% CO2, 21% O2. Cells were released from G1/S arrest by removal of thymidine, and harvested at 2 hr intervals for immunoblotting analysis. The cell cycle phase at each time point was determined by FACS analysis of propidium iodide stained cells fixed in 70% ethanol.

41For section immunohistochemistry, tissues and decidua were dissected into ice-cold PBS and fixed with 4% paraformaldehyde (PFA)/PBS for 3-16 hr at 4°C. After three 30 min washes with PBS, the samples were passed through a sucrose series, embedded, then frozen in OCT embedding medium (Sakura) and cryosectioned at 10 μM. Frozen sections were post-fixed in ice cold acetone 10 min at −20°C and air-dried prior to use; then washed three times in TBST (1x TBS, 0.1% Triton X-100). Sections were blocked with 10% donkey serum/TBST/1mM NaN3 for 1 hr at room temperature. Sections were incubated with diluted primary antibody (Table S2) in 1% donkey serum/TBST/1mM NaN3 for 14-16 hr at 4°C. Slides were washed three times 10 min with TBST before incubating with secondary antibody (Table S2) diluted in 1% donkey serum/TBST/1mM NaN3 for 4 hr at room temperature with TOTO-3 (Invitrogen, 1:400) and RNase A (50 μg/ml). Samples were washed three times 10 min in TBST and mounted in Prolong Gold (Molecular Probes).

42For cell proliferation studies, pregnant dams and postnatal animals were given a single intraperitoneal injection of EdU (Invitrogen) at a dose of 50-100 mg/kg body weight in a solution of 10 mg/ml PBS (pH 7.4). After 1-2 hr, animals were sacrificed and tissues harvested and fixed in 4% PFA/PBS with gentle mixing for 3 hr at 4°C and processed as described above. Click-iT reactions (Invitrogen) were performed on cryosections as per manufacturer's instructions.

43For whole-mount immunohistochemistry of blastocysts, E2.5 embryos were collected from Rnaseh2b+/− x Rnaseh2b+/− plugs and cultured overnight in M16 media (Sigma) to blastocyst stage. Blastocysts were then fixed in 4% PFA/PBS for 20 min and transferred to depression slides for subsequent steps. Blastocysts were washed five times 5 min in PBST (1x PBS, 0.1% Triton X-100), permeabilized in PBS/0.25% Triton X-100 for 20 min and washed five times 5 min in PBST. Blastocysts were blocked in 10% donkey serum/TBST/1 mM NaN3 for 2 hr at room temperature and incubated 14-16 hr in primary antibody (Table S2) diluted in 1% donkey serum/TBST/1 mM NaN3 in a humidified chamber at 4°C. Samples were washed three times 10 min in TBST, prior to incubating with secondary antibody (Table S2) diluted in 1% donkey serum/TBST/1 mM NaN3 for 4 hr at room temperature with TOTO-3 (Invitrogen, 1:400) and RNase A (50 μg/ml). Samples were washed three times 10 min in TBST and mounted on a Superfrost slide between 9 mm Secure-Seal spacers (Molecular Probes) and no. 0 coverslips in Prolong Gold (Molecular Probes). Details of antibodies and dilutions used in this study are given in Table S2.

44Images were collected on a Zeiss Axioplan II fluorescence microscope or on a Nikon A1R confocal microscope, comprised of a Nikon Eclipse TiE motorized inverted microscope with Perfect Focus System. Image capture and analysis were performed using IPLab Spectrum (Scanalytics Corp) driving a Coolsnap HQ CCD camera (Photometrics Ltd) for the Zeiss system. For the confocal, image capture and analysis was performed using Nikon NIS Elements AR.

45RNA was extracted from individual embryos using TRIzol reagent (Invitrogen), with further purification using RNeasy Mini Kit columns (QIAGEN) and RNase-free DNase I (QIAGEN). cRNA was generated using an Illumina TotalPrep RNA Amplification Kit (Ambion) and whole-genome gene expression analysis of biological triplicates performed using MouseWG-6 v2.0 Expression BeadChips (Illumina). MIAME compliant microarray data was deposited at the Gene Expression Omnibus (GEO, http://www.ncbi.nlm.nih.gov/geo/) under the accession number GSE37419.

46Microarray data were analyzed with R 2.14.0, using its beadarray and Limma 3.10.2 packages. Raw, non-normalized bead-summary values were imported from the Illumina BeadStudio software into R using the beadarray package. Quantile normalization was applied to the data to enable comparison between arrays. A linear model was applied to the expression data for each gene. To determine statistically differentially expressed genes the results of the linear model were summarized and a Bayes moderated t test applied. To control for multiple testing, a Benjamini and Hochberg false discovery rate p-value of < 0.05 was used.

4710 μl reactions consisting of 1 μl template cDNA, 1 X Brilliant II Sybr Green qPCR Master Mix (Stratagene), 0.3 μM passive reference dye (ROX, Stratagene) and either 0.1 μM of each oligonucleotide primer (Ccng1, Cdkn1a) or 0.06 μM of each primer (Actb) in nuclease-free water. Each sample and -RT were analyzed in triplicate and a no template control was include for each master mix. Quantitative real-time PCR was performed using an ABI Prism HT7900 Sequence Detection System (Applied Biosciences) with the following protocol: 50°C (2 min), 95°C (10 min) followed by 40 cycles of 95°C (15 s), 60°C (1 min). The cycle threshold (CT) values for each gene were used to calculate the expression of each target gene normalized to beta-Actin using the comparative CT method. Specificity of the primers was confirmed by agarose gel electrophoresis of the PCR products.

48Numeric and simulation analysis was performed in R (version 2.10.1). Peaks in the size reference lane were identified under supervision and the linear model lm(y∼log(x)) fitted to produce an electrophoretic distance (y) to nucleotide size (x) calibration curve. Densitometric histograms of electrophoresis lanes (e.g., Figure 5D) corresponding to alkali-treated genomic DNA were smoothed by fitting the smooth.spline function with 40 degrees of freedom. These smoothed distributions were transformed into nucleotide coordinate based histograms using a calibration curve from the same gel derived from 1 kb size standards (Invitrogen), extrapolated where necessary. Histograms were scaled such that they represented genomic fragment size distributions summing to 109 nucleotides. Direct subtraction of the number of genomic fragments in each histogram provides an approximate analytic estimate of the difference in fragment number between lanes. Simulation studies were based on the fragmentation of 109 nucleotide genomes. Random cuts were added to a starting genome, either cut at random or with a starting distribution of cuts estimated from densitometric data as above. A hill climbing strategy with 50 iterations was employed to find the optimal number of random cuts added, to achieve the best fit with a target histogram (e.g., turning the Rnaseh2b+/+ densitometric distribution into the Rnaseh2b−/− distribution). The objective function for optimization was to minimize the Manhattan distance between the target histogram and the histogram of the simulated data after smoothing as above.

49Mouse embryonic fibroblasts (106) were seeded into T175 flasks and incubated for 14-16 hr at 37°C, 5% CO2 and 3% O2. Cells were cultured for another 48 hr in the absence or presence of hydroxyurea (300 μM). Cells were subsequently trypsinized, washed in PBS and fixed in 70% ice cold ethanol. Fixed cells were pelleted and resuspended in 2 ml pre-warmed pepsin (Sigma) dissolved at 1 mg/ml in 30 mM HCl and incubated at 37°C for 30 min with frequent mixing. Cell nuclei were pelleted at 3350 g for 4 min and resuspended in 100 μl 0.1 mg/ml RNase A diluted in PBS-EDTA before the addition of 100 μl propidium iodide (100 μg/ml) diluted in PBS-EDTA. Each sample was incubated at 4°C for a minimum of 1 hr before acquisition on a BD Biosciences FACSAriaII. Data was analyzed using FlowJo software (v7.6.1, Tree Star).

50For 2D FISH, metaphases were isolated in hypotonic buffer, fixed with methanol:acetic acid (3:1) and dropped onto slides. Slides were incubated with 100 μg/ml RNase A in 2x SSC for 1 hr at 37°C, washed in 2x SSC and dehydrated through an alcohol series, followed by denaturation in 70% formamide/2x SSC at 70°C. Major (pSAT) and minor (R198) satellite probes were prepared and labeled by nick translation with digoxigenin-11-dUTP or biotin-16 dUTP. Probes (100 ng per slide) were mixed with 5 μg of sonicated salmon sperm DNA, denatured and hybridized to slides for 14-16 hr at 37°C. For chromosome 4 painting, 15 μl per slide of Ready-to-use use paint (Cambio) was denatured at 70°C for 5 min, re-annealed at 37°C for 15 min and hybridized to denatured slides for 14-16 hr at 37°C. Washes and detection were performed as previously described. Two-dimensional slides were examined using a Zeiss Axioplan II fluorescence microscope with Plan-neofluar objectives, a 100 W Hg source (Carl Zeiss) and Chroma #8300 triple band pass filter set (Chroma Technology Corp.) with the excitation filters installed in a motorized filter wheel (Ludl Electronic Products). Grayscale images were captured with a Hamamatsu Orca AG CCD camera (Hamamatsu Photonics Ltd.). Image capture and analysis were performed using in-house scripts written for IPLab Spectrum (Scanalytics Corp.).

51For embryonic and live-born viability, categorical data was analyzed using Chi-square tests, under the null hypothesis that mutant mice would be present at Mendelian ratios. All quantitative data was analyzed using Student's t tests, under the assumption that data followed a Normal distribution. For cell counting experiments (e.g., Figure 4C, 5B and 7A), the percentage of positive cells was calculated for each experiment. This continuous measure was then analyzed by unpaired t tests, to establish if differences between cell lines were statistically significant, allowing for inter-experimental variation.