Genetic requirements for repair of lesions caused by single genomic ribonucleotides in S phase

Abstract

Single ribonucleoside monophosphates (rNMPs) are transiently present in eukaryotic genomes. The RNase H2-dependent ribonucleotide excision repair (RER) pathway ensures error-free rNMP removal. In some pathological conditions, rNMP removal is impaired. If these rNMPs hydrolyze during, or prior to, S phase, toxic single-ended double-strand breaks (seDSBs) can occur upon an encounter with replication forks. How such rNMP-derived seDSB lesions are repaired is unclear. We expressed a cell cycle phase restricted allele of RNase H2 to nick at rNMPs in S phase and study their repair. Although Top1 is dispensable, the RAD52 epistasis group and Rtt101^Mms1-Mms22 dependent ubiquitylation of histone H3 become essential for rNMP-derived lesion tolerance. Consistently, loss of Rtt101^Mms1-Mms22 combined with RNase H2 dysfunction leads to compromised cellular fitness. We refer to this repair pathway as nick lesion repair (NLR). The NLR genetic network may have important implications in the context of human pathologies.

Fig. 1. SGA screen identifies network of genes required for rNMP-derived lesion tolerance in S phase.

A For the SGA analysis, the illustrated three query strains (G1-RNH202, S-RNH202, and G2-RNH202) were crossed to the non-essential yeast knockout (YKO) collection. The heterozygous diploids were sporulated and fitness of the resulting haploid double mutants was scored based on their colony size. The outcome was compared to the corresponding scores from the wild type (RNH202) cross. For each genotype, four replicates per strain were generated and analyzed. Scheme was created using Adobe Illustrator. B–D Scatter plots of normalized double mutant fitness for the queries compared to wild type (wt). The three queries with cell cycle RNH202 alleles compared to the wild type control (wt). Each data point represents a single mutant in the YKO collection. Significant synthetic sick interactions (fitness query x orfΔ wt x orfΔ <0.8, p < 0.05 in a t-test, corrected for multiple testing using the Benjamini-Hochberg method) are highlighted in magenta. Crosses with mms1Δ, mms22Δ and rtt101Δ mutants are indicated. Top right, total number of significant synthetic sick interactions. E Venn diagram of the number of synthetic sick interactions for the three queries with the cell cycle restricted RNH202 alleles. F GO term enrichment analysis for synthetic sick interactions of the S-RNH202 query. Only Biological Process GO terms are shown, top 10 terms by p-value in a hypergeometric test. G Network summary of the synthetic sick interactions for the three queries with the cell cycle RNH202 alleles. Genes mapped to the GO term „DNA metabolic process” are highlighted in grey. We could exclude genes with linkage to the rnh202Δ locus, and manual tetrad dissection identified false positives (Supplementary Data 1). These false positive hits were excluded from the network (faint appearance in the scheme).

Fig. 2. Rtt101, histone modifiers and HDR factors promote cell viability when genomic rNMPs are hydrolyzed in S phase.

A Tetrads from Rtt101^Mms1-Mms22 complex-deficient diploid strains in combination with S-RNH202-TAP revealed smaller colony sizes (colonies in circles). The genetic suppressor of rtt101Δ, mrc1Δ is sufficient to fully reverse the growth phenotype (colonies in squares). B Tetrads from dissections of Rtt101^Mms1-Mms22 complex-deficient diploid strains in the S-RNH202-TAP pol2-M644G genetic background augmented the sickness. The rtt101Δ S-RNH202-TAP pol2-M644G lethality was less penetrant compared to the mms1Δ and mms22Δ mutants, but after propagating, these small colonies were mostly inviable or acquired suppressor mutations. C Tetrad dissections of rad52Δ with RNH202-TAP cell cycle alleles demonstrates that the S-RNH202-TAP rad52Δ double mutant is inviable. D The phenotype from (C) was exacerbated when increasing the rNMP load (pol2-M644G). E The tetrads for RAD51 deletion with all cell cycle alleles of RNH202 shows that the S-RNH202 rad51Δ double mutant is sick but viable. F Spot assay shows epistasis between S-RNH202-TAP rtt101Δ and S-RNH202-TAP rad51Δ double mutants and the S-RNH202-TAP rtt101Δ rad51Δ triple mutant. G Alkaline gel electrophoresis of the genotypes used in F) showed epistasis in terms of gDNA fragmentation between S-RNH202-TAP rtt101Δ and S-RNH202-TAP rad51Δ double mutant and the S-RNH202-TAP rtt101Δ rad51Δ triple mutant (compare lanes 3, 4, 5). The neutral gel serves as control for gDNA purity and integrity. Two alkaline gels revealed the same result. H Western blot analysis of checkpoint status by phospho-shift analysis of Rad53. The DNA profiles showed a 2n peak accumulation in line with the activated Rad53 checkpoint. I Viability analysis of exponential cultures stained with SYTOX Green shown as Scatter dot blot (line = mean, error bars = SD). “Boil” is the boiled positive control representing 100% dead hence SYTOX green positive cells. The unstained wild type sample serves as background control. Statistical analysis with GraphPadPrism8 (multiple comparison ANOVA test), n = 2 for all samples except rnh201Δ rtt101Δ and S-RNH202-TAP rtt101Δ n = 4 samples. Source data are provided as a Source Data file.

Fig. 3. Rtt101 becomes essential in S phase in a Top1-independent manner to overcome rNMP-derived toxicity.

A The rtt101Δ rnh201Δ double mutant is synthetic lethal in the presence of HU (hydroxyurea). Transformation with RNH1 did not affect the growth of the double mutant on HU-containing agar plates. B Transformation with RNH201 did rescue growth of the rtt101Δ rnh201Δ double mutant on HU plates, while the RER-deficient separation-of-function rnh201-RED mutant had no effect. Two independent transformants of each genotype showed the same result. C Depletion of RNH201-AID* in the presence of IAA (auxin) resulted in a synthetic sick growth phenotype with rtt101Δ, which was amplified into a synthetic lethal phenotype when combined with pol2-M644G. D Complementation of the pol2-M644G rtt101Δ RNH201-AID* triple mutant with wild type RNH201 rescued growth on auxin plates, while the rnh201-RED mutant did have no effect. E The synthetic lethality of the pol2-M644G rtt101Δ RNH201-AID* triple mutant on auxin plates was Top1-independent. Three independent strains from separate tetrads were spotted to confirm Top1-independence of the observed phenotype. F–H Liquid cultures with the indicated genotypes were synchronized with α-factor in the G1 phase in the presence of IAA to deplete RNH201-AID*. The arrested cultures were either directly plated on YPD agar plates (colony count panel G), or released into the S phase in the presence of IAA, followed by plating on YPD agar plates (colony count panel H). Scatter dot blot with bar at mean from n = 7 plate counts per genotype and condition. Statistics were performed with GraphPad Prism8 (multiple comparison ANOVA). I The spot assay with rtt101Δ strains in the RER-deficient rnh201Δ and the rNMP accumulating pol2-M644G background revealed that alkalization of the YPD agar was sufficient to phenocopy the synthetic sick growth defects seen in the presence of HU. Source data are provided as a Source Data file.

Fig. 4. Rtt101 mediates the repair of rNMP-derived DNA damage in S phase through histone H3 ubiquitylation.

A Scheme of the contribution of the histone chaperone Asf1, the histone acetyl transferase Rtt109 and the ubiquitin ligase Rtt101^Mms1 to the replication-coupled nucleosome assembly (RCNA) pathway. Sequence of events (modified from ref. ): Asf1 binds to de novo synthesized H3-H4 dimers and Rtt109 acetylates H3 followed by ubiquitylation of H3 through Rtt101 that lead to the release of Asf1 and facilitates DNA incorporation. Note that an acetylation deficient H3-K56R mutant is reminiscent of RTT109 deletion and an ubiquitylation-deficient H3-3KR mutant is reminiscent of the RTT101 deletion. Created with BioRender.com. B Manual tetrad dissection confirmed the synthetic lethal phenotype between the S-RNH202 allele and the histone remodeler genes RTT109 and ASF1 (double mutant colonies in circles). C Representative tetrads from single CAF-1 complex deletion mutants (cac1Δ, cac2Δ, cac3Δ) in combination with S-RNH202-TAP to check contribution of the RCNA pathway. Note that these genes work redundantly and they are synthetic lethal with each other, hence we cannot exclude their contribution. D Spot assays with RTT109 deletion mutants shows that loss of RTT109 is toxic in RER-deficient strains with high rNMP load (pol2-M644G). E The same is true for loss of the histone chaperone ASF1. F Spot assays with histone H3 mutants deficient for Lysine-56 acetylation (“H3-K56R”) and Rtt101-dependent Lysine-121,122,125 triple ubiquitylation (“H3-3KR”). These plasmid-borne mutant versions of histone H3 replaced the two H3 coding genes (HHT1 and HHT2) that were deleted from the genome (see yeast strain list for the respective complete genotypes). Histone H3 Lysine-56 acetylation became essential in RER-deficient cells (RNH201-AID* on IAA plates). The H3-3KR strain revealed mild sickness in RER-deficient cells but was inviable when rNMP levels increased with the pol2-M644G allele. G Strains from (F) were combined with RTT101 deletion to confirm the epistasis between RTT101-deficiency and H3 ubiquitylation deficiency (compare lanes 4–6).

Fig. 5. Top1-independent NLR pathway is essential when rNMPs cause pre-S phase nicks that result in seDSB.

DNA polymerases transiently incorporate single rNMPs into the genome during replication and repair. The RER pathway removes genomic rNMPs immediately in the subsequent G2 phase. In RER-deficient, or RER-dysfunctional cells the Top1-mediated backup pathway deals with rNMP-removal. However, if high amounts of persistent genomic rNMPs accumulate in RER-deficient, or RER-dysfunctional cells, the likelihood increases that hydrolysis-prone rNMPs form ssDNA nicks. When the replication fork encounters such an rNMP-derived nick, a toxic seDSB is formed. End topology of the seDSB may differ dependent on the location of the nick. To repair the rNMP-derived seDSB lesions, functional RCNA is required. The histone remodelers Asf1 and Rtt109 act upstream of Rtt101^Mms1-Mms22, presumably accompanied by the resection of the seDSB by MRX (Mre11-Rad50-Xrs2), followed by HDR (Rad52, Rad51) and resolution of the HDR intermediates (Mus81-Mms4, Sgs1-Top3-Rmi1) to result in the error-free repair of the seDSB. In RTT101-deficient cells with high rNMP load, histone H3 does not become ubiquitylated and downstream error-free HDR repair of the seDSB is compromised causing genomic instability likely by alternative, error-prone repair attempts. Abbreviations: NLR rNMP-derived nick lesion repair, rNMP single ribonuclesoide monophates, RER ribonucleotide excision repair, seDSB single-ended double strand break, ssDNA single stranded DNA, RCNA replication-coupled nucleosome assembly, HDR homology-directed repair. Created with BioRender.com.