Ribonucleotide triggered DNA damage and RNA-DNA damage responses

Abstract

Research indicates that the transient contamination of DNA with ribonucleotides exceeds all other known types of DNA damage combined. The consequences of ribose incorporation into DNA, and the identity of protein factors operating in this RNA-DNA realm to protect genomic integrity from RNA-triggered events are emerging. Left unrepaired, the presence of ribonucleotides in genomic DNA impacts cellular proliferation and is associated with chromosome instability, gross chromosomal rearrangements, mutagenesis, and production of previously unrecognized forms of ribonucleotide-triggered DNA damage. Here, we highlight recent findings on the nature and structure of DNA damage arising from ribonucleotides in DNA, and the identification of cellular factors acting in an RNA-DNA damage response (RDDR) to counter RNA-triggered DNA damage.

Abundant Incorporation of ribonucleotides into DNA

In articulating the general transfers of molecular biology: DNA → DNA, DNA → RNA, and RNA→ Protein, Francis Crick presciently noted that the proposed central dogma:

“ says nothing about what the machinery of transfer is made of, and in particular nothing about errors. (It was assumed that, in general, the accuracy of transfer was high.)."¹

Nearly a half-century later, emerging data indicates that for DNA polymerases, the fidelity of the polymerization reaction (i. e. DNA → DNA) with respect to the nucleic acid sugar identity varies widely.² Ribonucleotide incorporation into DNA is abundant and pervasive across phyla. It has been demonstrated to occur in prokaryotes (E. coli)^3,4 yeasts (both Schizosaccharomyces pombe⁵ and Saccharomyces cerevisiae ^2,6,7 and importantly, in vertebrate cells.^8,9 Estimates for the budding yeast replicative polymerases suggest that on average ∼1/6500 polymerization events (∼13,000 rNMPs/genome) are completed using ribonucleotides, rather than deoxyribonucleotides.^2,6,7 In mouse embryonic fibroblasts, it is estimated that greater than ∼1,300,000 ribonucleotides (approximately 1/7600 polymerization reactions) are introduced into the nuclear genome each cell division.^8,9 Ribonucleotides are also utilized by the DNA repair polymerases μ and λ,¹⁰ as well as the mitochondrial DNA polymerase γ.¹¹ Broadly speaking, ribonucleotides outnumber all other forms of abundant DNA damage including apurinic sites, oxidative DNA base damage (e.g. Eight-oxo guanine), and UV induced photoproducts (e.g., T-T dimers) (Fig. 1). In fact, when summed, the number of ribonucleotide incorporation events in cycling cells may exceed, by an order of magnitude, the total of all other known forms of DNA damage combined. How then, are ribonucleotides removed from the genome, and what are the consequences of harboring ribonucleotides in DNA? How is the potential for ribonucleotide triggered DNA damage circumvented, and what factors operate in this RNA-DNA realm to preserve genome integrity?

Figure 1.

Ribonucleotides are an abundant form of DNA damage. Ribonucleotide incorporation in cycling cells is the most commonly observed insult to genomic DNA, with ∼1,000,000 ribonucleotide insertions observed in mammalian genomes⁸ and ∼13,000 budding yeast genomes.⁷ UV-induced damage including 6,4 photoproducts and thymine-thymine dimers are estimated to occur at high frequency, ∼10⁵ per cell/day in keratinocytes.^53,54 Apurinic (AP) sites from spontaneous depurination and base excision repair reactions occur at ∼30,000 APsites/cell/day.⁵⁵ DNA base damage from reactive oxygen species (ROS) generates 8-oxoguanine, a common lesion with an abundance of ∼2,400 lesions/cell/day.⁵⁵

Ribonucleotide excision repair (RER)

Like all abundant DNA damage, ribonucleotides in DNA are efficiently removed (Fig. 2A). A robust ribonucleotide removal pathway, now termed ribonucleotide excision repair or RER (see below), is initiated by the structure specific RNase H2 endoribonuclease. ^7,8,9,12,13 Early work demonstrated that mammalian HeLa or Saccharomyces cerevisiae cell extracts can catalyze the in vitro endonucleolytic cleavage and excision of a single ribonucleotide embedded in an otherwise doubled-stranded DNA (dsDNA) substrate.¹³ Yeast lysates deficient in the catalytic subunit of RNase H2, RNH201, cannot catalyze incision 5′ of the ribonucleotide base, whereas rad27Δ strains deficient for the budding yeast flap endonuclease (FEN-1) homolog display impaired nucleolytic removal of the incised ribonucleotide. Consistent with a role in DNA repair, a budding yeast RNase H2 deficient strain has an increased spontaneous mutation rate, that is dominated by 2–5 bp deletions in repeat sequences.^6,7,14,15 To further test the functional relevance of RNase H2 to ribonucleotide excision in vivo, Kunkel and coworkers⁷ engineered powerful genetic tools to facilitate manipulation of ribonucleotide levels in the budding yeast genome. In this approach, mutants of the leading strand replicative DNA polymerase ε (S. cerevisiae POL2) active site targeting a conserved hydrophobic residue adjacent to the polymerase “steric gate” were created that confer either increased or decreased propensity of the polymerase to utilize rNTPs (rather than dNTPs) during polymerization.⁷ Yeast strains with mutant POL2 have either elevated (pol2-M644G) or suppressed (pol2-M644L) levels of genomic ribonucleotides.^6,7 Significantly, the rate of 2–5 bp deletions are highly elevated in a strain that harbors both an elevated ribonucleotide load, and a deficiency in RNase H2-directed RER (pol2-M644G rnh201Δ).^6,7 Similar genetic relationships for polymerase active site mutants and RNase H2 have been observed in S. pombe.⁵ Genome maintenance roles for RNase H2 have also been established in mice. Mutational inactivation of mouse RNase H2 results in abnormal embryonic development, accumulation of single and double ribonucleotides in nuclear DNA of mouse embryonic fibroblasts, confers severe genome instability phenotypes including gross chromosomal rearrangements, and causes activation of a p53-dependant DNA damage response.^8,9 Altogether, studies in yeasts and mice indicate that RNase H2 directed RER is critical for removal of DNA embedded ribonucleotides to prevent genomic instability.

Figure 2.

Ribonucleotide incorporation by DNA polymerases into the genome creates insults that necessitate the need for multiple methods and repair enzymes for efficient removal. (A) RNase H2 cleavage initiates RER and generates 3′OH and 5′PO₄-Ribo-DNA substrates. Abortive DNA ligation adenylates RER intermediates to generate adenylated-RNA-DNA lesions. Aprataxin catalyzes deadenylation of the 5′-AMP. When RNase H2 is deficient, topoisomerase I incision at a ribonucleotide creates unnatural ends consisting of a 5′OH and 3′Ribo-Top1 cleavage complex. Upon cyclization utilizing the available 2′OH of the ribonucleotide, a 2′-3′-cyclic-PO₄ adduct is generated, similar to hydrolysis. Ribonucleotides stimulate Top2cc formation. Tdp2 repairs Top2-RNA/DNA complexes. (B) The crystal structure of a transition state mimic complex of Aprataxin bound to adenylated RNA-DNA (RCSB ID: 4NDG).²⁵ The adenylated-RNA lesion is bound in A-form conformation, and completely encircled and aligned for the Aptx direct damage reversal reaction. (C) X-ray structure of T. maritima RNase H2 bound to an 5′-RNA-DNA-3′ junction (PDB ID: 4HHT).¹⁷ (D) The crystal structure of a Top1-DNA covalent complex (PDB ID: 1a31).³² A modeled 2′-OH is displayed, indicating active site rearrangement might be required to facilitate the 2′-3′-cycliclization reaction. (E) The structure of an RNA-DNA junction hybrid bound by mouse Tdp2 (PDB ID: 4PUQ).²⁸ The 1.6 Å X-ray structure indicates the Tdp2 binding pocket accommodates RNA substrate, in a strained formation, for a single metal ion mediated phosphotyrosyl cleavage reaction. Glycerol occupies the approximate position of Top2 active site tyrosine residue.

Structure-function studies of RNase H2 homologs have also laid a detailed foundation for understanding the RNase H2 incision reaction.^16,17 Early structures of Archaeoglobus fulgidus RNase H2 predicted that together a canonical RNase H fold, conserved tyrosine-finger (Y164) and helix-loop-helix domain would mediate interactions with the RNA-DNA.¹⁷ Subsequent X-ray structures of T. maritima RNase H2 bound to an RNA-DNA junction¹⁶ confirmed and extended these predictions, unveiling critical properties of RNA-DNA structure-specific substrate recognition (Fig. 2C). Notable interactions to the ribonucleotide 2′-OH are stabilized by key interactions with the tyrosine finger (Y163 of T. maritima RNase H2) of the C-terminal domain that deforms the RNA/DNA backbone, facilitating a proposed substrate assisted catalytic mechanism.¹⁶ The tyrosine finger also participates in ring stacking interactions with the ribose sugar of the +2 position in the RNA-DNA junction, providing selectivity for cleavage at the RNA-DNA junction (Fig. 2C). By manipulating conserved elements of the RNA-DNA interaction observed in the X-ray structures, a mutant of budding yeast RNase H2 has been engineered that is unable to cleave single ribonucleotides in DNA but retains the ability to perform other RNase H2 dependent functions (e.g. degradation of RNA-DNA hybrids).¹⁸ This separation of function mutant allows for the correlation of 2–5 bp deletion mutants accumulated in rnh201Δ strains with single ribonucleotides embedded in DNA, and is thus supportive of RNase H2 RNA-DNA processing roles in the maintenance of genomic integrity in vivo.¹⁸

Ribonucleotide-triggered RNA-DNA Damage

Given the high abundance of ribonucleotide incorporation into DNA, it is also probable that some ribonucleotides escape RER. While differing by a only a single atom at the 2′-position, reactivity of the ribose 2′-hydroxyl renders the sugar-phosphate backbone prone to hydrolysis reactions generating aberrant 5′-hydroxyl and 2′-3′-cyclic phosphate strand break termini (Fig. 2A).¹⁹ Ribonucleotides also induce distortions to surrounding local DNA helical structure,^20,21 and alter elastic properties of nucleic acid.²² The differences in the structure and backbone reactivity of ribonucleotide-containing DNA can further induce differential processing of an RNA-DNA substrate by key nucleic acid metabolizing enzymes including DNA ligases,^23-25 DNA topoisomerase 1 (Top1),^15,26 and DNA topoisomerase 2 (Top2).^27,28 We discuss these outcomes below.

Mutagenic Ribonucleotide Removal by Top1

Topoisomerase 1 (Top1) resolves DNA supercoils to facilitate replication and transcription.^29,30 The Top1 incision-religation reaction cycle initiates with nucleophilic attack by an active site tyrosine on the DNA phosphodiester backbone to generate a transient and reversible Top1 cleavage complex (Top1cc) where Top1 is covalently linked to the 3′-terminal phosphate (Fig. 2A). On ribonucleotide containing substrates, however, Top1 can display strong endoribonuclease activity.^15,26 Ribonucleotides create hot spots for the Top1 incision,^15,26 and within the Top1 active site, the Top1cc is further prone to nucleophilic attack by the vicinal 2′-OH of the ribonucleotide. This cyclization reaction generates a 5′-OH and 2′-3′-cyclic-PO₄ products, rather than resealing of the phosphodiester backbone as in the canonical Top1 reaction.^6,15,26,31 Inspection of the X-ray structure of human Top1cc,³² suggests that a significant rearrangement of Top1cc with a covalently linked 3′-ribonucleotide (modeled, Fig. 2D) would be required to facilitate the activation and alignment of the 2′-OH group for cyclization.

Intriguingly, in the absence of RNase H2, Top1 processing of ribonucleotides drives an alternative ribonucleotide excision repair pathway for removal of ribonucleotides from genomic DNA, albeit at a cost.⁶ In RNase H2-deficient yeast containing a high level of unrepaired ribonucleotides, Top1 processing of ribonucleotides causes activation of the intra S-phase checkpoint, replication stress, and the hallmark generation of 2–5 bp deletions in repetitive sequences.^6,15,31 Following Top1 incision, the 5′-side of the break can be processed by an Srs2-ExoI helicase-nuclease pathway.³³ Yet, the precise downstream processing events that generate deletions remain to be determined in detail (see refs 6,33 for discussion). Also, the candidate processing enzyme(s) for resolving the 2′-3′-cyclic-PO₄ ends has not been reported.

Repair of Ribonucleotide Linked Top2 Cleavage Complexes by Tdp2

The eukaryotic type II topoisomerases (Top2α and Top2β) regulate DNA topology by employing a dsDNA cleavage and religation cycle involving transient formation of Top2-DNA cleavage-complexes (Top2cc).^34,35 Top2 catalytic intermediates are characterized by covalent linkage of the topoisomerase to the DNA 5′-terminus by an active site tyrosine residue (Fig. 2A). However, aberrant DNA structure or targeted chemotherapeutic disruption of the Top2 reaction can generate Top2cc, protein-DNA crosslinks that block transcription and/or collapse DNA replication forks.^29,30 Interestingly, ribonucleotides stimulate the Top2α and Top2β DNA cleavage reactions²⁷ to produce Top2–RNA-DNA cleavage complexes.^27,28 Ribonucleotides are therefore potentially toxic to the normal Top2 reaction cycle.

A major pathway for repair of Top2cc that is present in vertebrates, but absent in lower eukaryotes including yeasts, involves direct reversal of Top2-DNA phosphortyrosyl linkages by tyrosyl DNA phosphodiesterase 2 (Tdp2).^36-38 Tdp2 (aka Ttrap/Eap2/VPg unlinkase) also catalyzes reversal of protein-RNA covalent linkages during RNA replication of picornaviruses (e.g., poliovirus and rhinovirus), suggestive of broader RNA repair functions for Tdp2.³⁹ Recently, we demonstrated that Tdp2 also efficiently processes protease digestion fragments of Top2cc covalently adducted to 5′ ribonucleotides.²⁸ The high resolution structure of Tdp2 bound to a 5′-ribonucleotide containing substrate further defines a mechanism through which RNA-containing substrates are engaged by Tdp2 in a manner similar to that of DNA-only substrates (Fig. 2E).^40,41 Overall, in vitro results support the possibility that genomic instability triggered by ribonucleotides in DNA may also be mediated by formation of Top2–RNA-DNA cleavage complexes and production of DNA single strand and double strand breaks. Correspondingly Tdp2 may also protect from RNA-triggered Top2cc formation in vivo by acting as an RNA-DNA repair factor.

Adenylated RNA-DNA Repair by Aptx

Embedded ribonucleotides are not the only threats to genomic integrity. We hypothesized that abundant incised RER intermediates from RNase H2 cleavage might also impact frequently occurring DNA transactions.²⁵ One example of this is DNA ligation. Eukaryotic ATP-dependent DNA ligases catalyze DNA nick sealing during DNA replication and repair with a 3 step mechanism involving active site adenylation of the ligase, adenylate transfer to the DNA 5′-phosphate, and DNA nick sealing with release of AMP. However, when DNA ligases engage nicked DNA substrates with preexisting DNA damage, for instance an RNA-DNA junction from RER, DNA ligase can undergo “abortive ligation” where the enzyme dissociates prematurely from its substrate following DNA adenylation.^42,43,44 In this context, rather than sealing a DNA nick to finalize DNA replication or repair, DNA ligase may exacerbate preexisting DNA damage by catalyzing further addition of bulky AMP adducts (Fig. 2A). RNA-DNA junctions arising from RER are indeed subject to abortive ligation by human DNA ligases 1 and 3.²⁵ Thus, RER intermediates trigger ligation failure, and production of compounded DNA damage in the form of adenylated RNA-DNA lesions. The mechanism through which DNA ligase is impacted by ribonucleotides merits further investigation. This process might involve localized distortion of the DNA 5′-terminus that impacts the ligase nick-sealing reaction and/or modulation of DNA ligase's ability to stably encircle and align the nick junction.

Aprataxin (Aptx) is a member of the histidine triad (HIT) family of nucleoside hydrolases, and catalyzes direct reversal of DNA 5′-adenylation resulting from abortive ligation.^25,42 Aptx therefore may be critical for genome stability in cells undergoing abortive ligation during RER. Consistent with a role for Aprataxin in metabolizing RNA derived damage, budding yeast cells with elevated genomic ribonucleotides that lack the Aptx homolog, Hnt3, display marked defects in cellular proliferation, activation of the S-phase DNA damage checkpoint, and are sensitive to the replication inhibitor hydroxyl urea (HU).²⁵ These phenotypes are attenuated in an rnh201Δ background, supporting a model where Aptx is required for efficient repair of adenylated RNA-DNA junctions that arise from RNase H2 incision followed by abortive ligase metabolism. Human and yeast aprataxins all efficiently catalyze reversal of adenylated RNA-DNA in vitro.²⁵ Further, a series of X-ray crystal structures of hAptx bound to adenylated RNA-DNA structures defines a molecular basis for adenylated RNA-DNA damage processing involving structure-specific A-form RNA-DNA recognition of the adenylated RNA 5′ terminus, and encirclement of the lesion pyrophosphate linkage to facilitate deadenylation (Fig. 2B).

Conclusions and Outlook

More research is needed to define the impacts of ribonucleotides on human health. For example, to what extent does RNA-triggered DNA damage contribute to carcinogenesis? Given that Aptx⁴⁵, RNase H2^46,47 and Tdp2⁴⁸ mutations are linked to neurological disease, how do ribonucleotides in DNA modulate these disease outcomes? Mutations of the genes encoding human RNase H2 subunits are all associated with the autosomal-recessive neuroinflammatory Aicardi-Goutières Syndrome (AGS).⁴⁶ AGS phenotypes may be in part due to defective RER and the accumulation of endogenous ribonucleotides and added forms of RNA-triggered DNA damage (Fig. 2A). Human APTX mutations are linked to heritable progressive cerebellar degenerative disease Ataxia with Oculomotor Apraxia 1 (AOA1).⁴⁵ The facts that disease-causing AOA1 mutants (e.g. the missense mutation K197Q) significantly impair RNA-DNA deadenylation and directly distort the Aptx substrate RNA-DNA binding pocket makes it tempting to speculate that RNA-triggered abortive ligation contributes to the neurodegenerative pathology in AOA1.²⁵

We note that in addition to the variety of potentially deleterious impacts on genomic stability discussed herein, intriguing examples for functions of embedded nuclear ribonucleotides are coming to light. These include the stable incorporation of ribonucleotides into the nuclear genome that regulate the mating type switch in S. pombe^49,50 and a role for ribonucleotides as a DNA strand discrimination signal in DNA mismatch repair.^51,52 Undoubtedly, genome-wide mapping of distribution, composition, and stability of genomic ribonucleotides promises to illuminate new functions and vulnerabilities imparted by introduction of ribonucleotides into DNA.

In searching for unifying principles that define the impacts of ribonucleotides on DNA transactions, we find few. Rather, ribonucleotides unpredictably alter enzymatic transactions, with important implications for genome stability, function, and disease. Ribonucleotides can vary reaction rates and equilibrium (e.g., Top2cc formation), alter the reaction course of DNA metabolizing enzymes (e.g. conversion of Top1 to an RNA-DNA endonuclease), and turn key genome guardians into genome threats (e.g., RNAse H2-dependant triggering of abortive DNA ligation). The emerging structure/activity studies are revealing that genome maintenance direct reversal enzymes and nucleases can indeed cope with varying chemistries triggered by ribose incorporation. The active sites of enzymes such as Tdp2 and Aptx reveal these enzymes do not deselect ribose, but rather can effectively accommodate lesions arising in both DNA and RNA-DNA contexts. Similarly, broad substrate specificity of the RER nuclease FEN-1 dually accommodates both RNA- and DNA in its incised strand.^56,57 As the lines between RNA and DNA worlds become blurred, we anticipate new exciting twists, and added RNA-DNA metabolizing enzymes to emerge as our understanding of the impacts of ribonucleotides on genome stability and function deepens.

Disclosure of Potential Conflicts of Interest

No potential conflicts of interest were disclosed.

Acknowledgments

We thank J Williams, S Andres and D Appel for comments.

Funding

Our studies are supported by the NIH Intramural Program: National Institute of Environmental Health Sciences, 1Z01ES102765 to R.S.W.