Nucleotide salvage, genome instability, and potential therapeutic applications

Oxidized nucleosides

Oxidative stress has long been found to induce DNA oxidation, resulting in a plethora of DNA modifications [57]. Among the oxidatively generated DNA adducts, 8-oxodG has received the most attention due to the lowest oxidation potential of dG among the four canonical 2′-deoxynucleosides [57]. Hence, the occurrence and biological consequences of 8-oxodG salvage have been extensively investigated [6, 54, 58–63]. Research indicates that 8-oxodG can be salvaged into the nucleotide pool, where it is hydrolyzed by purine nucleoside phosphorylase (PNP) to yield 8-oxoguanine instead of being directly phosphorylated. This process then leads to the formation of the corresponding nucleoside triphosphates, 8-oxorGTP and 8-oxodGTP [6]. The 8-oxodGTP can be utilized by DNA polymerases and incorporated into genomic DNA [58, 60–63]. For instance, DNA polymerase β (Pol β) can incorporate 8-oxodGTP with the formation of 8-oxodG:dC and 8-oxodG:dA mispairs within DNA trinucleotide repeat [59]. This also holds true for Pol η and Pol κ, which can efficiently insert 8-oxodGMP opposite a template dA [60, 63, 64]. The reported level of salvaged 8-oxodG incorporation is of the same order of magnitude as basal level of genomic 8-oxodG [6, 65]. The measurement, however, was conducted under experimental conditions involving near-saturating extracellular 8-oxodG, and thus may represent an upper bound for salvage-mediated incorporation [6].

Under normal conditions, this threat is mitigated by NUDT1 (MTH1), which hydrolyzes 8-oxodGTP to 8-oxodGMP [44, 45]. When nucleotide sanitization is inadequate, e.g. during oxidative stress or elevated salvage flux, misincorporation increases substantially, promoting mutagenesis.

Alkylated lesions

Alkylated DNA lesions represent another major type of extensively studied DNA damage [57, 66]. The ubiquitous presence of alkylating agents in the environment and inside cells renders this type of lesions generally unavoidable [67, 68]. Aside from being generated by alkylating agents, DNA alkylation may arise from the salvage of alkylated nucleosides and their subsequent incorporation into genomic DNA. Indeed, Spratt and coworkers [5] demonstrated that N²-substituted dGTP derivatives with methyl (Me), n-butyl (nBu), benzyl or 4-ethynylbenzyl group are robust substrates for human DNA Pol κ, as evidenced by both in vitro and cellular experiments. In addition, work from our laboratory showed that feeding cultured human cells with N²-MedG and N²-nBudG resulted in augmented levels of the two modified nucleosides in genomic DNA [8, 14]. Interestingly, Spratt and we found that incorporation of N²-alkyl-dG occurs preferentially during early S-phase DNA replication [15, 69]. In addition, DNA Pol κ and, to a lesser extent, Pol η facilitate the incorporation of N²-MedG and N²-nBudG into genomic DNA [5, 8, 14]. Furthermore, depletion of ALKBH3 in cultured human cells led to increased accumulation of N²-MedG and N²-nBudG [8].

Food consumption

Cooking food, especially at high temperatures, has long been shown to be associated with adverse health consequences, including higher tendency to develop colorectal cancer and diabetes [70]. Conventionally, DNA adducts induced by chemical species generated during cooking process, e.g. polycyclic aromatic hydrocarbons and heterocyclic aromatic amines, are of primary concern when considering genotoxic agents induced by cooking [71]. The DNA in food itself, however, received little attention. High-temperature treatment can produce oxidized purines, thymine glycol, and hydrolytic deamination products [72, 73]. Consequently, dietary DNA can carry pre-existing lesions formed during cooking and/or food processing/storage. Upon digestion, these pre-existing modified nucleosides can be absorbed, enter salvage pathways, and become available for incorporation into DNA. A recent study by the Kool laboratory showed that HeLa cells exhibit elevated levels of DNA double-strand breaks (DSBs) upon incubation with damaged nucleosides, e.g. dU, 5-hydroxy-2′-deoxyuridine (5-OHdU), 5-hydroxy-2′-deoxycytidine (5-OHdC), 5,6-dihydro-thymidine (DH-dT), and thymidine glycol (dTg) [9]. This finding supports that deaminated and oxidatively damaged nucleosides can be taken up into cells, metabolically activated and incorporated into genomic DNA, ultimately resulting in DSBs [9]. However, it is important to note that dTg nucleotide is more difficult to be utilized by polymerases during DNA synthesis [74]. Furthermore, oral administration of mice with dU led to augmented levels of dU in genomic DNA isolated from small intestine [9]. Therefore, salvage of damaged DNA components from food can be a potential contributor to genome instability [9].

Ribonucleotide oversupply

rNTPs are normally present at levels that are orders of magnitude higher than dNTPs [75]. Although DNA polymerases are capable of excluding rNTPs, such exclusion is incomplete; hence, even high-fidelity replicative DNA polymerases can incorporate rNMPs during DNA synthesis [12, 13]. For instance, human lagging-strand replicative polymerase Pol δ was shown to incorporate approximately one rNTP in every 2 000 dNTPs, although the experiment was performed in vitro using purified protein [12]. Moreover, a substantial amount of inserted rNTPs by leading-strand replicative polymerase Pol ε escape 3′→5′ exonucleolytic proofreading [13]. In addition to replicative polymerases, human Pol η is capable of incorporating rCMP opposite dG and 8-oxodG [76]. The misincorporation of ribonucleotides into genomic DNA has been extensively investigated, and its prevalence and biological implications have been reviewed elsewhere [77, 78].

RNA-derived modifications

Different from its DNA counterpart, RNA carries a diverse array of chemical modifications, with more than 150 distinct modifications identified to date [79]. In addition, some of these modifications occur at high stoichiometries [80, 81]. Degradation of RNA continuously releases these modified nucleosides, which can be recycled through salvage pathways to produce not only rNTPs, but also chemically modified dNTPs. Once in the nucleotide pool, these aberrant substrates may be utilized by DNA polymerases and incorporated into genomic DNA.

This is the case for N⁶-MedA. Several reports in 2015 proposed that N⁶-MedA can act as an epigenetic mark in eukaryotes [82–84], and aberrant deposition of N⁶-MedA was shown to modulate gene expression and contribute to the proliferation of glioblastoma cells [19, 85, 86]. This role of N⁶-MedA, however, remains controversial, where multiple studies revealed extremely low levels of N⁶-MedA in mammalian genomes [19, 87, 88]. Moreover, recent studies indicate that m⁶A from RNA degradation can be metabolically activated to yield N⁶-MedATP, which can be incorporated into genomic DNA by DNA polymerases [10, 11]. Importantly, isotope tracing experiments revealed that a substantial portion of the genomic N⁶-MedA in cultured mammalian cells originates not from DNA methyltransferase-mediated methylation, but rather from m⁶A [10, 11]. These findings challenge the view of N⁶-MedA as an epigenetic mark in mammalian cells, and suggest that N⁶-MedA in DNA may arise from a byproduct of metabolic recycling.

Consistent with stringent cellular control over this process, a recent study by Chen et al. [18] identified an active checkpoint that limits aberrant DNA N⁶-MedA arising from RNA m⁶A turnover. In particular, the adenosine deaminase-like protein ADAL catabolizes N⁶-MedAMP in vivo, thereby preventing its accumulation and subsequent incorporation into DNA. Loss of ADAL results in elevated levels of free m⁶A species and augmented misincorporation of N⁶-MedAMP into DNA, which is suppressed by AK1 depletion [18]. Together, these findings support a model where RNA-derived m⁶A can enter salvage pathways, but is normally intercepted by dedicated sanitization mechanisms, thereby minimizing the incorporation of N⁶-MedA into genomic DNA. In addition, the salvage pathway may constitute an important route for incorporating other RNA-derived modifications into DNA.

Mutagenesis

Once modified nucleotides are incorporated into DNA, the DNA replication machinery often struggles to maintain fidelity. To prevent replication fork collapse, cells rely on TLS DNA polymerases, e.g. Pol η, Pol ι, Pol κ, Pol ζ, and REV1 [56, 89]. These polymerases can accommodate bulky or noncanonical nucleobases into their active sites, facilitating bypass of those lesions that stall replicative polymerases, albeit sometimes at the expense of fidelity. For example, Pol κ often inserts mismatched nucleotides opposite oxidized purines [90], while REV1 specializes in incorporating dCMP regardless of the template base [91]. In addition, modified bases can pair with incorrect nucleotides. A well-studied example is 8-oxodG, which can adopt the syn N-glycosidic bond configuration, facilitating Hoogsteen pairing with dA, though it pairs correctly with dC when in the anti configuration [92]. As such, 8-oxodG can result in G → T transversion [93].

It should be noted that TLS polymerases are required for the error-free bypass of N²-alkyldG lesions [94–97]. In this vein, our laboratory showed that loss of Pol ι, Pol κ, or REV1 diminishes bypass efficiency and results in substantial frequencies of G → A transition and G → T transversion mutations at the site of N²-alkyldG lesions in mammalian cells [95, 97]. Moreover, simultaneous depletion of Pol ζ together with Pol κ or Pol ι further aggravates the frequencies of these mutations, suggesting the cooperation of TLS polymerases in mitigating replication errors [95].

Apart from nucleobase modifications, unrepaired ribonucleotides embedded in DNA can promote genome instability. In mammalian cells, ribonucleotides incorporated into genomic DNA are primarily removed by RNase H2-dependent ribonucleotide excision repair, where impairment in RNase H2 activity leads to ribonucleotide accumulation and increased DNA damage [98, 99]. In biochemical and cellular systems, human topoisomerase I (Top1) exhibits endoribonuclease activity and cleaves ribonucleotide-containing DNA, generating strand breaks and repair intermediates [100]. Moreover, defective ribonucleotide excision repair in mammals is associated with the ID4 signature mutation—an insertion-deletion mutation detected in human cancers characterized by short 2–5 base pair deletion—occurring at a TNT sequence motif, suggesting a mechanism involving TOP1 activity at ribonucleotide sites embedded in the genome [101].

Replication stress and DNA strand breaks

Lesions that escape from repair can stall DNA polymerases and block replication fork progression. Persistent stalling promotes fork reversal, collapse and ultimately DSB induction. Oxidatively generated DNA lesions such as 8-oxodG were shown to elicit structural changes around the replication fork, interfering with DNA replication [102, 103]. N²-alkyldG lesions also impede DNA replication with lower bypass efficiencies [95] and were recently linked to R-loop accumulation [16], leading to genome instability. In addition, salvage of damaged nucleosides, e.g. dU, 8-oxodG, 5-OHdU and 5-OHdC, could give rise to high activities of base excision repair (BER), resulting in DSBs [9]. Feeding cells with dU, 5-OHdU and 5-OHdC also leads to chromosomal aberrations, although this effect was not significant for 8-oxodG [9]. Moreover, mice administered with dU exhibited elevated DNA repair activity and increased level of DSBs [9]. Apart from BER, additional repair pathways are also involved in mitigating salvage-elicited genomic instability.

Mismatch repair (MMR) corrects mispairs like 8-oxodG:dA [102], preventing mutagenic G→ T transversions, where oxidized dNTP pool was shown to result in genomic instability in MMR-deficient Msh2^−/− mouse embryonic fibroblasts [104]. In addition, nucleotide excision repair (NER) excises bulky and minor-groove lesions like N²-alkyl-dG [105], which can otherwise be bypassed accurately by Pol κ [94–97]. Under conditions of high salvage flux, however, the capacity of MMR and NER to process these lesions may be overwhelmed, exacerbating replication stress and genome instability.

Ribonucleotides embedded in DNA confer structural distortions that hinder replication and transcription. Even a single ribonucleotide alters DNA electrostatic potential, groove width, and duplex elasticity, and RNA’s intrinsically high hydrolytic instability accelerates single-strand break formation [98, 106].

Epigenetic regulation

An additional layer of complexity arises when RNA-derived modifications, such as m⁶A, enter the nucleotide pool via salvage. Once converted into N⁶-MedATP and incorporated into DNA [10, 11], this aberrant nucleotide may confer epigenetic-like effects. Indeed, N⁶-MedA has been reported as an emerging DNA modification capable of modulating gene expression, chromatin organization, and processes such as DNA replication and recombination in mammals [19, 87, 107]. Notably, functional roles for DNA N⁶-MedA have also been described in certain pathophysiological settings, including glioblastoma [19, 85]. In this vein, salvage-mediated misincorporation of N⁶-MedA may perturb normal epigenetic regulation by introducing noncanonical modifications into DNA, rather than representing a bona fide epigenetic mark.

Likewise, some oxidatively damaged nucleosides, e.g. 8-oxodG, are also implicated in modulating G-quadruplex folding in gene promoters to regulate gene expression [108–110], and were found to crosstalk with histone epigenetic marks and DNA methylation [111–114]. In this context, 8-oxodG at CpG site was shown to alter DNA methyltransferase binding and modulate DNA cytosine methylation in vitro [112, 114]. In addition, 8-oxodG arising from histone demethylation-induced ROS was shown to recruit 8-oxoguanine-DNA glycosylase 1 and topoisomerase 2β, which elicit chromatin and DNA conformational changes that are essential for estrogen-induced transcription [113]. As such, salvage-driven recycling of 8-oxodG into the nucleotide pool, followed by misincorporation into DNA, may not only compromise genome integrity, but also disrupt epigenetic signaling.

Cancer

Cancer cells frequently upregulate nucleotide salvage to meet elevated dNTP demand associated with rapid proliferation while simultaneously experiencing oxidative stress, hypoxia, inflammation, high nucleic acid turnover, and altered nucleoside transporter activity [115]. Oncogene amplification (e.g. MYC) and activation (e.g. HIF1A) induce expression of salvage enzymes (TK1, dCK, HPRT1, and UCK2) [40, 116, 117], thereby increasing intracellular access to preformed modified nucleosides.

Elevated salvage flux increases the probability that chemically modified nucleosides are phosphorylated to triphosphates and incorporated into DNA. In tumors, this risk is exacerbated by chronic ROS production [118], dysregulated nucleotide sanitization due to altered NUDT1 or SAMHD1 activity [49, 119], and increased reliance on TLS polymerases [120, 121]. Together, these features create an environment that is conducive for salvage-driven mutagenesis and genome instability.

Chronic inflammation

Chronically inflamed tissues generate high levels of reactive oxygen and nitrogen species, as well as hypohalous acids, which damage endogenous DNA, RNA, and dietary or microbial nucleic acids [122–125]. Salvage pathways can convert these modified nucleosides into dNTPs, increasing the likelihood of their incorporation into genomic DNA. Recurrent cycles of damage and repair in conditions such as inflammatory bowel disease, gastritis, and chronic hepatitis may therefore create a cellular environment particularly susceptible to salvage-driven genome instability.

Aging and neurodegeneration

Aging is associated with diminished expression and/or activity of nucleotide sanitization enzymes, e.g. NUDT1 [53], mitochondrial dysfunction [21, 126], increased ROS burden [127], and imbalanced dNTP pool arising from a proof-reading-deficient mutant of replicative DNA polymerase in mitochondria (i.e. Pol γ^D257A) [128]. This mutant Pol γ-elicited dNTP pool imbalance, however, could not reproduced in a subsequent study [129]. These age-related changes collectively confer heightened vulnerability to salvage-derived lesions. Post-mitotic tissues such as neurons, which rely heavily on salvage pathways due to limited de novo nucleotide synthesis [130, 131], may be especially susceptible to incorporation of noncanonical nucleotides [132]. Although neurons do not divide, mitochondrial biogenesis and robust DNA repair required to preserve cell integrity can create opportunities for modified nucleotides to be incorporated into mitochondrial and nuclear DNA [132–134]. Elevated levels of oxidized nucleosides observed in neurodegenerative disorders such as Parkinson’s disease and Alzheimer’s disease raise the possibility that salvage-mediated incorporation of modified nucleotides contributes to mitochondrial genome instability and impaired neuronal survival [135–137].

Microbiome- and diet-derived nucleosides

The gut microbiome represents a continuous source of chemically modified nucleosides, including deaminated purines, alkylated bases, and oxidized products arising from microbial metabolism and host–microbe redox interactions [138, 139]. These nucleosides can traverse the intestinal epithelium and enter systemic circulation [9], rendering them accessible to salvage pathways in mammalian systems.

Dietary intake constitutes an additional exogenous source of damaged nucleosides. Heat-processed foods contain oxidized purines, thymine glycol, and deaminated cytosine or adenine derivatives [9]. Experimental evidence demonstrates that mammalian cells readily uptake such damaged nucleosides, leading to elevated DNA DSBs and chromosomal aberrations in cultured cells and in vivo [9]. Hence, salvage provides a direct metabolic route through which dietary DNA damage may intersect with genome integrity.

Salvage enzyme targets for therapeutic intervention

Several salvage enzymes represent potential therapeutic targets. TK1 is highly upregulated in proliferating cancer cells and is central to the activation of many pyrimidine analog drugs, rendering it a candidate for synthetic lethal strategies in salvage-dependent tumors [140, 141]. dCK activates numerous anticancer nucleoside analogs, including gemcitabine and cytarabine [142]. Inhibition of dCK can sensitize tumors to replication stress [141]. PNP, a key enzyme in purine salvage, has been extensively explored as a therapeutic target in T-cell malignancies [143], and its inhibition may also limit the metabolic activation of damaged purine nucleosides. HPRT1 and APRT, essential for purine base salvage, may be exploited in MYC-driven cancers that exhibit heightened reliance on salvage pathways [144, 145].

Nucleotide sanitization enzymes as therapeutic modulators

Nucleotide sanitization enzymes offer complementary therapeutic leverage. Inhibition of NUDT1 (MTH1) increases incorporation of oxidized purine nucleotides, driving toxic mispairing in ROS-rich tumors [54], and high level of NUDT15 protein expression is positively correlated with tolerance to 6-mercaptopurine [146]. DUT inhibition elevates dUTP levels, promoting lethal cycles of uracil incorporation and BER [147, 148]. ITPA minimizes incorporation of deaminated purines [43] and 6-thioguanine [149], while SAMHD1 constitutes a biomarker for cytarabine response and a therapeutic target in acute myeloid leukemia [50, 51]. Modulating these enzymes can amplify or suppress salvage-associated genotoxic stress.

Nucleoside analog therapies

Current cancer therapies already exploit salvage-dependent activation mechanisms. Examples include 5-fluorouracil, 6-thioguanine, azacytidine/decitabine, and gemcitabine, which rely on salvage enzymes for metabolic activation and incorporation into DNA [150–153]. Understanding salvage-driven mutagenesis provides a framework for interpreting differential sensitivity and resistance to these agents.

Patient-selection biomarkers and precision oncology

Dependence on salvage pathways varies widely across tumors, creating opportunities for biomarker-guided therapy. Candidate biomarkers include expression levels of TK1, dCK, NUDT1, DUT, and SAMHD1 [50, 51, 54, 141, 147], dNTP pool measurements by LC–MS [75], isotope-tracing of salvage flux [141, 154], and mutation spectra indicative of salvage-driven incorporation, such as elevated G→T transversions due to 8-oxodG:dA mispairing [155]. These metrics may guide patient stratification for nucleoside analog therapies or salvage-pathway inhibitors.

Combination strategies

Aside from targeting individual enzymes or steps within salvage pathways, combinatorial strategies may be particularly effective. Coupling enhanced salvage flux with inhibition of nucleotide sanitization is expected to amplify genotoxic stress. For example, driving incorporation of oxidized nucleosides while blocking NUDT1 [54], or coupling nucleoside analog therapy with checkpoint inhibition to exploit replication stress [156, 157] may constitute viable strategies. Tumors lacking de novo synthesis capacity, such as methylthioadenosine phosphorylase (MTAP)-deficient cancers [158], may be particularly vulnerable to such approaches.