Nucleic Acid Adductomics – the Next Generation of Adductomics towards Assessing Environmental Health Risks

Abstract

This Discussion article aims to explore the potential for a new generation of assay to emerge from cellular and urinary DNA adductomics which brings together DNA- RNA- and, to some extent, protein adductomics, to better understand the role of the exposome in environmental health. Components of the exposome have been linked to an increased risk of various, major diseases, and to identify the precise nature, and size, of risk, in this complex mixture of exposures, powerful tools are needed. Modification of nucleic acids (NA) is a key consequence of environmental exposures, and a goal of cellular DNA adductomics is to evaluate the totality of DNA modifications in the genome, on the basis that this will be most informative. Consequently, an approach which encompasses modifications of all nucleic acids (NA) would be potentially yet more informative. This article focuses on NA adductomics, which brings together the assessment of both DNA and RNA modifications, including modified (2’-deoxy)ribonucleosides (2’-dN/rN), modified nucleobases (nB), plus: DNA-DNA, RNA-RNA, DNA-RNA, DNA-protein, and RNA-protein crosslinks (DDCL, RRCL, DRCL, DPCL, and RPCL, respectively). We discuss the need for NA adductomics, plus the pros and cons of cellular vs. urinary NA adductomics, and present some evidence for the feasibility of this approach. We propose that NA adductomics provides a more comprehensive approach to the study of nucleic acid modifications, which will facilitate a range of advances, including the identification of novel, unexpected modifications e.g., RNA-RNA, and DNA-RNA crosslinks; key modifications associated with mutagenesis; agent-specific mechanisms; and adductome signatures of key environmental agents, leading to the dissection of the exposome, and its role in human health/disease, across the life course.

Graphical Abstract

1. Introduction to the Exposome and Biomarkers

Components of the Environment contribute 80-90% of the risk of developing cancer, and degenerative diseases ^{1, 2}. Specifically, this risk of disease is driven by components of the exposome i.e., all the internal and external, environmental factors to which humans are exposed, across the life course, including diet, air pollution, social interactions, living environment, and lifestyle choices ³ (summarized in Figure 1). Exposome-related health issues therefore represent a significant challenge to public health ⁴. Interactions between the exposome and the genome impact individual health, and contribute to the risk of disease ⁵. However, the exposome represents a complex mix of exposures, processes, and factors, and therefore to identify the precise nature, and size, of risk powerful tools are needed to link exposure, cellular consequences, and health/disease ⁶. Considerable effort is focused currently on methodology for estimating the specific and general external exposomes (the concepts of which are illustrated Figure 1) i.e., the origins of exposure lie exclusively outside the body. From a health perspective, the focus is on the agents which reach the body’s cells, and may then lead to disease ⁷. In this regard, it is the internal exposome which is of primary interest as it represents (i) the total target (which could be a particular cell type, organ, or whole-body burden), and (ii) dose of physico-chemical agents in the body (irrespective of their source i.e., being external or internal to the body, or how they are derived e.g., processes such as inflammation, metabolism, and microbiome⁸ (Figure 1) ^{6, 7}. (We have added ‘physico-’ to the definition to include agents such as radiation.) A biomarker can be described as “a measurable substance in an organism whose presence is indicative of some phenomenon such as disease, infection, or environmental exposure” ⁹. Therefore, biomarkers of the internal exposome e.g., modifications of cellular biomolecules may be particularly biologically informative, as they may represent the internal dose, effect(s) on the biological system, and possess a role in disease ¹⁰ (see Figure 1). On this basis, a major goal of toxicology is to use biomarkers of the internal exposome to (i) identify and (ii) act as proxies for key health-modifying elements of the exposome without the need for detailed measurements of the external environment ⁶.

Figure 1. Role of the exposome in the generation of cellular molecular modifications.

The exposome (encompassing both external and internal exposomes) represents a wide, and diverse range of exposures, that may occur across the life course. Species arising from the exposome have the potential to damage, or otherwise modify, DNA, RNA, proteins. These modified molecules may result in adverse outcomes for the cell, and lead to disease in the organism. Example exposures from the exposome are illustrated.

Like other ‘omics techniques (e.g., genomics, metabolomics, proteomics), a major strength of DNA adductomics is the untargeted assessment of the totality of a target, in this case the DNA adductome i.e., all of the covalently modifications of DNA nucleobases (DNA adducts) derived from environmental and dietary genotoxicants and endogenously produced electrophiles ¹¹. In this Discussion, we have summarized some of the recent advances in DNA adductomics, and then indicated how DNA adductomics may advance to become yet more comprehensive through the detection of the totality of nucleic acid (NA)-derived adducts, and how NA adductomics may thereby better inform on both exposures and their biological consequences.

2. Consequences and causes of nucleic acid modifications

There is continual exposure to agents, and processes which damage ¹², or otherwise modify NA [which includes DNA, RNA, and the 2’-deoxyribonucleotide and ribonucleotide i.e., (d)NTP pools] ¹³, and their associated proteins (represented by “molecular modifications” in Figure 1). The definition of a genotoxin is a chemical or agent that can cause DNA or chromosomal damage ¹⁴. Currently there is not an equivalent term for RNA damage, although analogy clearly exists ¹⁵. We propose that DNA and RNA can, and should, be considered together when adopting a holistic view of the effect of genotoxin exposure, particularly as damage to RNA, like DNA, is increasingly being linked to disease (reviewed in ^16–19). As discussed elsewhere, whether a particular alteration of NA is referred to as damage or a modification depends upon the modification, and context ²⁰. Modifications induced intentionally by normal cellular process, such as epigenetic programming (e.g., DNA methylation), may be viewed as modifications, whereas those formed unintentionally are damage ²⁰, although a grey area may exist in which context is key to interpretation ²¹.

To date most focus has been upon damage to DNA (and 2’-dNTP pool) as this impacts cell function ²², and plays a role in the pathogenesis of, arguably, all major human diseases, in particular cancer, but also neurodegeneration, and cardiovascular disease, plus aging ²³. However, there is increasing evidence that damage to, or modification of, RNA (and the ribonucleotide pool) also affects cell function, and plays a role in pathogenesis ^{17, 18, 24}. Although DNA and RNA have distinctly different functions within a cell, and hence damage to them has different implications, they both represent targets for agents which can damage nucleic acids ^{25, 26}. Damage to DNA can interfere with fundamental cellular processes, such as DNA replication, and transcription, and can thereby lead to a variety of outcomes, for example mutation, microsatellite instability, alterations in transcription factor binding, loss of heterozygosity ²², and cytotoxicity ²⁷. Damage to RNA (both coding and non-coding) accumulates at much higher levels than DNA ^{28, 29}, alters the chemical properties and hence the function of RNA leading to downstream consequences, such as defects in protein synthesis ^30–32. The consequences of damage to non-coding RNAs (which represent ~95% of the cellular RNA) is largely unknown, but the presence of mechanisms to detect modified rRNA and tRNA suggest their presence could have detrimental outcomes for the cell (reviewed in ³³). As they are the origin of the nucleobases in DNA and RNA, damage to the 2’-dNTP and NTP pools is a potentially significant source of damaged nucleobases in these biomolecules ^34–37, resulting in similar cellular consequences as listed above. The presence of mechanisms to sanitize damage from the 2’-dNTP and NTP pools (reviewed in ³⁸) supports the significance of these as sources of damage to NA.

Processes which alter NA can have exogenous and/or endogenous origins. An example of an endogenous origin is the “leakage” of electrons from normal cellular metabolism, which may lead to the generation of reactive oxygen species, and can damage NA resulting in background levels of oxidatively modified NA ³⁹. Another example is DNA methylation, where specialised enzymes add or remove methyl groups from DNA nucleobases to control gene expression, chromosomal structure, and stability ⁴⁰. In contrast, environmental stressors have exogenous origins, and while the damage that they cause to NA can be specific, and hence characteristic, to a particular stressor e.g., cyclobutane pyrimidine dimers (from UV radiation ⁴¹), 8,9-dihydro-8-(N7-guanyl)-9-hydroxyaflatoxin B1 (from dietary aflatoxin B1 ⁴²), 7-(2’-deoxyadenosin-N⁶-yl)aristolactam I (from Aristolochic acid ⁴³), they may also impact endogenous processes, and influence levels of endogenously-derived NA modifications. For example, exposure to UVR or ionising radiation, and metabolism of certain xenobiotics may all lead to the generation of free electrons, impact redox homeodynamics, and increase levels of oxidatively modified NA above baseline levels e.g., 8-oxo-7,8-dihydro-2’-deoxyguanosine and 8-oxo-7,8-dihydroguanosine, from DNA and RNA, respectively ⁴⁴. Environmental factors such as UVR and ozone can transform chemicals, which will also lead to the production of reactive species with the potential to modify DNA ⁴⁵. Similarly, some exogenous stressors e.g., benzo(a)pyrene ⁴⁶, polychlorinated biphenyls, methylmercury, and organochlorine pesticides ⁴⁷, can influence methylation, and other epigenetic processes. Consequently, there is some overlap in the types of damage arising from both endogenous and exogenous sources. While not every stressor may cause a characteristic single form of damage, like those noted above, we propose that it might be possible to determine unique ‘signatures’ of a particular stressor if multiple forms of damage are considered, or this might identify the precise mechanisms by which a stressor may act ⁴⁶. In both cases, the greater the types of damage considered, the more precise the signature or mechanism.

Of the processes which induce DNA damage, oxidation is a prime example of the potential to induce a multiplicity of different forms of damage. Recently defined as a hallmark of environmental insult ⁴⁸, oxidative eustress/distress ⁴⁹ can lead to the generation over 24 DNA nucleobase (nB) products, and the total number of adducts exceeds 100, when 2-deoxyribose (2-dR), and phosphate backbone modifications are considered ²³, and this does not include DNA-DNA and DNA-protein crosslinks, or the adducts derived from secondary processes, such as lipid peroxidation ⁵⁰. These numbers are increased further, when RNA-derived modifications, and epigenetic modifications of DNA and RNA are also included, bringing the total number of potential oxidatively-generated modifications/adducts into the high hundreds, if not greater.

Most of the literature describes the targeted analysis of DNA damage, rather than RNA, and the measurement of a single, or a few adducts ⁵¹. While not precluding the use of targeted analysis, this approach cannot assess the spectrum of NA adducts present in the cell, which gives rise to several weaknesses: e.g., potential biomarkers of exposure may be excluded, it is impossible to identify modifications which act in concert to increase disease risk, and it precludes the identification of yet-to-be-discovered modifications with a key role in disease. Indeed, there is growing evidence that RNA modifications are involved in a growing number of human diseases ^{24, 38, 52} (see also the MODOMICS website for a comprehensive review, https://iimcb.genesilico.pl/modomics/), and RNA modifications have a broad range of functions, including RNA processing, splicing, polyadenylation, editing, structure, stability, localization, translation initiation, and gene expression regulation ⁵³. To date, approximately 151 modified RNA nucleobase residues have been identified, increasing to ~334 as nucleoside and nucleotide forms are included ^54–59 however their precise effects upon cellular function remain to be discovered ¹³. Other types of NA-associated adducts also have significant biological consequences e.g., DNA-DNA, and DNA-protein crosslinks (DDCL, DPCL) interfere with DNA replication and transcription ^{60, 61}, RNA-protein crosslinks (RPCL) can modulate, or stabilize, RNA, interfering with function ⁶² [DNA-RNA crosslinks (DRCL), and RNA-RNA crosslinks (RRCL) may also occur, but given that their existence has not yet been reported, their function is unknown]. The breadth of NA modifications is therefore significant and, just as DNA adductomics aims to be most informative by reporting on the totality of DNA adducts ⁶³, we propose that a non-targeted NA adductomics will be yet more informative by aiming to comprehensively inform on the totality of NA modifications, simultaneously.

3. Cellular DNA adductomics - towards cellular NA adductomics.

Kanaly et al. ⁶⁴ first proposed a cellular DNA adductomic strategy to simultaneously measure multiple DNA adducts using liquid chromatography-triple quadrupole tandem mass spectrometry (LC-QqQ-MS/MS) with electrospray ionization (ESI), operating in the selected reaction monitoring (SRM) mode ⁶⁵. Due to the labile nature of the 2-dN glycosidic bond, this method detects a number of contiguous SRM of [M+H]⁺ to [M+H-116]⁺ transitions, in which 116 Da corresponds to the neutral loss of the 2-dR moiety. Lately, high resolution MS (HRMS)^{66, 67} has also been adopted for DNA adductomics analysis, for example, the hybrid quadrupole-linear ion trap Orbitrap MS (Q-LIT-OT-MS) ⁶⁸. The mass of the ions detected by OT can be measured with ppm accuracy (≤ 1-5 ppm mass error). This high mass accuracy significantly improves/enhances confidence in DNA adduct identification, compared to the nominal mass accuracy obtained by QqQ-MS/MS. The advent of DNA adductomics by HRMS has led to the proposal of a “top-down” approach, by which patterns of adducts may be used to trace, and identify the originating exposure source ^{69, 70} (summarised in Figure 2).

Figure 2. The concept of a top-down approach for HRMS-based DNA adductomics for linking human, environmental carcinogen exposure, and cancer risk.

Cellular DNA adductome maps are developed, using HRMS, from exposed and unexposed individuals. DNA adductome maps are compared, and potential candidate adducts, reflective of exposure, are short-listed. Expected (known) adducts are identified from characteristics held within an DNA adductome library ¹⁴⁰, and unexpected (unknown) adducts are identified by in silico prediction. Precursor reactive species/metabolites are identified from the profile of adducts formed, and the responsible environmental agents, to which the individuals were exposure, are then proposed.

Previously, we reported a cellular DNA adductomics approach ⁷¹, then a HRMS approach ⁷², and broadened the range of adducts to include DDCL (i.e., crosslinkomics) ⁷³. Recently, RNA adductomics was reported, which uses neutral loss targeting of the [M+H]⁺ > [M+H-132]⁺ transitions, in which 132 Da corresponds to the neutral loss of the 2-ribose (R) ²⁵. This reveals the potential for genotoxins to cause analogous forms of damage to both DNA, and RNA ²⁵, further emphasising the importance of studying NA, more broadly, as a target for genotoxins. However, to date there are no adductomics methods for the combined detection of DNA, and RNA modifications, let alone in conjunction with DDCL, RRCL, DRCL, DPCL, and RPCL.

4. Cellular nucleic acid adductomics

We propose that NA adductomics will offer the most comprehensive approach for the study of nucleic acid modifications, to date, by combining the assessment of DNA, and RNA modifications, plus DDCL, RRCL, DRCL, DPCL, and RPCL. To illustrate the feasibility of NA adductomics we present supportive data. For example, we illustrated that different NA modifications produce their own specific fragmentation patterns in collision-induced dissociation (CID; Figure 3) by HRMS (i.e., Q-LIT-OT-MS). The experimental details and method applied are given in the Supplementary information. For example: modified 2’-dN tend to lose dR, resulting in the product ion of [M+H-dR]⁺; modified ribonucleosides (rN) tend to lose ribose (R), resulting in the product ion of [M+H-R]⁺ or [M+H-MeR]⁺, respectively. The origin of [M+H-MeR]⁺ can be ascribed to ribose 2’-O-methylation, which is one of the most common and widespread types of RNA modification, commonly found in rRNA, tRNA, small nucleolar RNA (snRNA) and, as reported recently, in mRNA ⁷⁴. Modified nB produce specific product ions including [B+H]⁺ and [B+H-NH₃]⁺; DDCL generate product ions of [M+H-2dR]⁺, and [M+H-dR]⁺; RRCL have product ions of [M+H-2R]⁺, and [M+H-R]⁺; DRCL have product ions of [M+H-dR]⁺, [M+H-R]⁺, and [M+H-dR-R]⁺; DPCL give [M+H-dR]⁺, [M+H-Cys ]⁺ and [M+H-dR-Cys]⁺; RPCL generate [M+H-R]⁺, [M+H-Cys]⁺, and [M+H-R-Cys]⁺. The cysteine moieties arise from the proteins with which the nucleic acids were crosslinked ⁷³. The above exact mass of neutral loss (Da) or product ions (m/z) monitored by HRMS are provided in the Supplementary Table S1 together with the supporting studies reported in the literature. In general, for modified 2’-dN, rN, nB, DDCL and DPCL, the fragmentation patterns observed are consistent with the previous studies (see Table S1). However, to the best of our knowledge, no study has been reported the others (RRCL, DRCL and RPCL) in the literature. It is worth noting that adductomics is not limited solely to adducts formed via damaging reactions i.e., the products of which are unintentional. For example, intentional products including those arising from endogenous processes, such as epigenetic modifications ²¹, are also detected by NA adductomics. The goal is to evaluate the maximal number of modifications, in a single analytical run, to provide the most comprehensive characterization of the role of NA adducts in environmental exposures, and disease risk. Additionally, this could lead to the generation of stressor-specific, signature patterns of adducts (characteristic NA adductome maps), and help inform on the mode of action of stressors.

Figure. 3. Fragmentation patterns of (A) modified 2’-dN, (B) modified rN, (C) modified nB nB , (D) DDCL, (E) RRCL, (F) DRCL, (G) DPCL, and (H) RPCL, by LC-Q-LIT-OT-MS with CID fragmentation mode.

Test solutions were prepared by dissolving various commercially available standards in deionized water, including modified 2’-dN, modified rN, modified nB. For the DDCL, RRCL, DPCL and RPCL, these were synthesized by incubating native 2’-deoxyribonucleosides (i.e., dGuo, dAdo, dCyd and dThd), ribonucleosides (i.e., Guo, Ado, Cyd and Urd), and/or cysteine, with formaldehyde ⁷³. The fragmentation patterns were obtained using a LC-Q-LIT-OT-MS equipped with a heated electrospray ionization source, operated in positive ESI-CID mode. All the informative, and specific product ions for each type of NA modification were generated in MS². The experimental details are provided in the Supplementary information.

While cellular DNA, and hence NA, adductomics have multiple, potential applications ⁶⁹, these are likely to be particularly valuable in identifying environmental exposures associated with increased cancer risk in humans. However, to date, there have been few reports of cellular DNA adductomics analysis of human tissue ^{64, 75}, although more are beginning to appear ⁶⁵, but this has not yet extended to human populations per se. Human tissue can represent a severe analysis challenge due to the need for high sensitivity to detect multiple adducts ⁷⁶. Recent advances in linear ion trap, and Orbitrap instruments offer improved sensitivity ⁵¹, but the availability of such instrumentation is limited, not least due to cost. Furthermore, cellular DNA adductomics requires a significant amount of DNA, typically 50-500 μg, depending upon adduct frequency ^{67, 77}, which is more than that needed for targeted analyses ⁵¹. Obtaining tissue is necessarily invasive, can be challenging logistically, adds additional ethics review board scrutiny, may discourage volunteer participation ^{78, 79}, and limits access to vulnerable individuals (e.g., the young, or elderly). Isolating DNA from exfoliated epithelial cells (e.g., from the urogenital tract or buccal cavity) represents a potential way forward, and a means to circumvent some of these issues ⁸⁰, albeit while restricting the investigation to a particular organ. In contrast, biomonitoring NA damage products in urine may be a viable alternative, with a number of advantages over cellular NA ⁵⁰ (see below).

5. Biomonitoring NA adducts in urine.

The presence of DNA adducts in urine is generally considered to be the consequence of DNA repair ^{81, 82} i.e., base excision repair (BER) is the source of nB adducts ⁸³, and global genome-, and transcriptional coupled-nucleotide excision repair (GG-NER, and TCR-NER, respectively), results in 2’-dN adducts ^84–87 (Figure 4). More specific evidence for this comes from the reported presence of certain 2’-dN adducts in urine, such as (5’R)-, (5’S)-8,5-cyclo-2’deoxyadenosines ^88–90, and 3-(2-deoxy-β-d-erythro-pentofuranosyl)pyrimido[1,2-α]purin-10(3H)-one ^{91, 92}, all of which appear to be repaired by NER ^92–94. Additionally, sanitization of the (2’-deoxy)ribonucleotide pools represents another logical source of urinary 2’-dN (and rN) adducts (Figure 4) ⁹⁵. Spontaneous depurination may also be a contributor to urinary adduct levels ⁹⁶ e.g., unstable N7-guanine adducts which have been reported to occur dose-dependently in the urine of rodents, and which appear proportionally to levels in the liver (reviewed in ⁹⁷). Given on-going exposure to environmental agents which may induce DNA damage, and the usual steady-state repair, ⁹⁸ the urinary NA adductome will be reflective of the body burden of adducts. While it may be argued that these repaired adducts are unlikely to be the ones specifically responsible for inducing mutations, or other adverse cellular events, they can reflect the risk of mutations occurring i.e., the larger the whole organism adductome burden, the greater the risk of mutation occurring within that organism ⁹⁹. Furthermore, defects in the post- adduct excision steps of DNA repair may also have detrimental consequences for the cell ¹⁰⁰.

Figure 4. Representative sources of NA modifications in urine.

Broadly, base excision repair (BER) of DNA leads to the presence of modified nucleobases in urine, and there is some evidence that urinary, modified 2’-deoxyribonucleosides are derived from the activity of global genome, and/or transcription-coupled, nucleotide excision repair (NER). The activity of enzymes, such as the Nudix hydrolases (e.g., NUDT1), which hydrolyse modified (2’-d)NTPs to NMPs, may also contribute to urinary levels of modified (2’-deoxy)ribonucleosides. The figure uses oxidatively modified DNA and RNA to illustrate examples of the processes. The repair products of DNA-DNA crosslinks (DDCL), DNA-protein crosslinks (DPCL), RNA-RNA crosslinks (RRCL), and RNA-protein crosslinks (RPCL) are less well defined.

There is a growing number of reports where targeted approaches have been used to successfully detect specific, environmentally induced DNA adducts in urine. Examples of urinary adducts include: 3-(4-dihydroxyphenyl)adenine, following benzene exposure of mice ¹⁰¹; aflatoxin N7-guanine, and 8,9-dihydro-8-(2,6-diamino-4-oxo-3,4-dihydropyrimid-5-yl-formamido)-9-hydroxy-aflatoxin B1, following exposure of rats to aflatoxin ¹⁰²; N7-(1-hydroxy-3-buten-2-yl)guanine following exposure of mice to 1,3-butadiene (BD) ¹⁰³, and occupationally- or recreationally- (tobacco smoke) exposed humans ¹⁰⁴; N7-(1-hydroxy-3-buten-2-yl) guanine adducts were also detected in the same human urine samples from the previous citation ¹⁰³, and bis-N7-guanine DNA-DNA crosslinks in mice exposed to BD ¹⁰⁵. The above reports, using targeted approaches, are beginning to successfully link exposure to environmental agents to specific urinary DNA adducts.

The literature describes a growing number of reports describing processes for the removal of RNA-derived adducts, the activity of which is supported by the presence of RNA products in urine (reviewed in ³⁸; Figure 4). The processes for the repair of crosslinks (NA-NA, and NA-protein) are more complex, and less well understood than for mono-adducts. DDCL are repaired by processes that involves NER ¹⁰⁶, with the potential involvement of BER ^{107, 108}. It has been suggested that the crosslink is removed in the context of an oligonucleotide, which is subsequently digested down to a smaller, lesion-containing product, prior to excretion into the urine ⁸⁷. There are reports of multiple mechanisms for cells to cope with DNA-protein crosslinks ¹⁰⁹, which may be broadly defined as: direct hydrolysis of the covalent linkage between protein and DNA; excision of the lesion-containing DNA fragment; and proteolysis of the protein component ¹¹⁰. Given these possibilities, the post-repair processing of the DNA-protein adduct itself is less clear than for DDCL (and for RRCL, and RPCL, even less so) ¹¹¹. Nevertheless, unless the DNA adduct is reversed ¹¹², some form of adduct-derived product will remain, which needs to be removed ⁶¹, and excreted (as evidenced from our data, presented here). Taken together, the above findings are highly supportive of the potential of urinary NA adducts to inform on whole body environmental exposures.

Measuring adducts in urine is not without its limitations. For example, while urine may inform on the body burden of adducts, to date it is not always possible to relate adducts in urine to a particular organ (skin and UVR-induced cyclobutane pyrimidine dimers may be a notable exception^{113, 114}, although contributions from peripheral blood mononuclear cells are possible). In contrast, cellular DNA adductomics can report on the adducts that are not repaired, and enables a direct link to be formed with the target tissue. Early reports examining urinary 8-oxoGua, 8-oxodGuo, and thymine glycol suggested possible contributions from the diet to urinary modified nucleobase concentrations, whereas the modified 2’-deoxynucleoside was unaffected ^{115, 116}. However, careful review of these studies revealed limitations in their study design ¹¹⁷, and more robust studies, in both mice and humans, subsequently revealed no contribution from diet to either 8-oxoGua or 8-oxodGuo (or indeed uric acid, a major purine breakdown product) in urine ^{81, 118, 119}, suggesting that this might be true for other adducts. Repair of adducts from the gut microbiome might also be considered a potential contributor to urinary adduct levels, although given that there appears to be no uptake of adducts from the diet, uptake of adducts from the microbiome is also unlikely to be a source; whether or not this is the case has yet to be reported in the literature. Normalization, or correcting for urine concentration can be a challenge, and may prevent comparisons between study groups. However, this has been well-studied for the targeted analysis of adducts such 8-oxodGuo, with 24 h collections being described as the gold standard, and spot urines corrected for creatinine a suitable alternative ¹²⁰. Additionally, we have proposed the use of a cocktail of stable isotopically labelled standards in DNA adductomics which will aid quantification, and correction for matrix effects ⁷¹.

Urine has numerous benefits as a matrix in which to study biomarkers. It is obtained non-invasively, easily collected, transported, and stored, with low biological hazard. Furthermore, no pre-processing is needed prior to storage, small volumes are required for analysis (typically less than 500 μL), and adduct stability is maintained for >15 y at −80 °C (in the case of the representative lesion, 8-oxodGuo, at least) ¹²¹, allowing the use of previously banked samples ^{117, 122}. Urine sample workup is generally simpler than for DNA. It is a well-established matrix in which to study biomarkers of exposure ¹²³, including a wide variety of DNA adducts e.g., oxidized nucleic acid products (including RNA) ^{124, 125}, N-nitroso-derived DNA adducts ¹²⁶, alkylated purines ¹²⁷, and aristolochic acid-derived DNA adducts ¹²⁸, illustrating the breadth of adducts in urine, which parallels that in NA. However, as noted above, such targeted analyses severely restrict the amount of information that can be gained. Applying adductomics to the study of urine therefore represents a convenient means to non-invasively assess the totality of NA modifications, and therefore exposures. Like most assays, cellular and urinary DNA adductomics both possess strengths and weaknesses but, as the literature for cellular and urinary targeted assays show, their appropriate application can be highly informative, and there is growing evidence that this is particularly true for adductomics, where a strength is being as comprehensive as possible, and applicable to a wide range of in vivo and in vitro systems ^129–132.

6. Why aim for the most comprehensive approach to studying nucleic acid damage?

While cellular and urinary DNA adductomics aim to assess the totality of DNA adducts in the genome, these approaches are not entirely comprehensive for genomic damage, as they have excluded RNA modifications (i.e., damage, and intentional modifications, such as epigenetic changes), and other classes of adducts (e.g., crosslinks), all of which have downstream consequences for the cell, and are consequently implicated in pathogenesis ^{38, 60, 62, 133–135}. We propose that the NA adductome represents an improved, functional, multi-faceted biomarker of the interaction between exposome, and genome. Furthermore, comprehensively assessing the totality of damage to all NA moieties, is critical to quantifying, and characterizing exposures, along with defining the role of nucleic acid damage in disease. The strengths of this approach include consideration of: (i) the factors which influence adduct formation, arising from the physicochemical nature of the NA moiety, e.g., due to abundance, conformation (double-, single-stranded, or free nucleoside/nucleobase), physical associations (e.g., with proteins, or metal ions), location (nucleus, cytoplasm, proximity to mitochondria), and propensity to modification (e.g., RNA is more easily oxidised than DNA, ^{28, 33} and probably more easily damaged generally²⁶); (ii) the variety of downstream effects that modified nucleic acids can cause, and their relevance to pathogenesis, e.g., mutations (DNA), epigenetic changes (DNA/RNA), error-containing proteins, and altered protein synthesis (RNA) ³⁸; and (iii) the ability to perform non-invasive assessments via urine.

7. Is NA adductomics feasible?

Cellular DNA adductomics of biological samples is becoming increasingly well-established ^{11, 25, 130, 136, 137}. However, we propose that, in line with the goal of ‘omics techniques to “determine the totality of a target”, the next advance for DNA adductomics is to evaluate the totality of modifications to nucleic acids, in cells, tissues, and urine. We therefore propose the development of cellular and urinary NA adductomics, extending our cellular ⁷¹, and urinary ¹³⁸ DNA adductomics assays to include modified 2’-dN, nB, rN, DDCL, RRCL, DRCL, DPCL, and RPCL. Urine is a convenient matrix in which to assess the NA adductome as it contains the totality of adducts following the repair of cells in the entire body, producing levels significantly higher than in individual tissues, and reflecting the totality of exposures. We therefore propose an NA adductomics approach, using a LC-Q-LIT-OT-MS with CID fragmentation, to achieve the detection of the widest, and most comprehensive, range of nucleic acid modifications. To support the feasibility of our proposed NA adductomics approach, our preliminary, proof-of-principle results revealed that at least six types of NA modifications were present in the urine of healthy subjects (Figure 5), including modified 2’-dN, modified rN, modified nB, DRCL, RRCL and RPCL. DDCL and DPCL were not detected in the urine, from healthy human subjects. Of the NA modifications detected, three out of 61 modified 2’-dN, 16 out of 265 modified rN, six out of 325 modified nB were fully identified, and confirmed using commercially available standards (summarized in Table S2). The detailed LC-Q-LIT-OT-MS analysis is provided in the Supplementary information. Among these fully identified NA modifications, it is worth noting that several are novel NA modifications not previously identified in the human urine, such as N⁴-methyl-2’-deoxycytidine (N⁴-mdCyd). At present, we cannot identify whether they derive from exogenous, or endogenous sources (or both). As expected, several well-known NA modifications were also successfully detected such as 8-oxodGuo and 5-methyl-2’-deoxycytidine (5-mdCyd; see Figure 5A), N⁶-methyladenine (m⁶A) and N⁶,2′-O-dimethyladenosine (m⁶Am; see Figure 5B), together with N7-methylguanine (N7-mGua), N3-methyladenine (N3-mAde) and 8-oxoGua (Figure 5C). In addition to the modified purine nucleobases (Figure 5C), a large number of modified pyrimidine nucleobases were also observed (> 400 adducts; data not shown). Regarding NA-associated crosslinks, for the first time we identified one DRCL (Figure 5D) and six RRCL (Figure 5E), which we propose to be AP apurinic/apyrimidinic (AP) site-derived crosslinks ¹³⁹ (DRCL: Ion 1 = 2’-deoxyadenosine-AP (dAdo-AP); RRCL: Ion 2 = 2’-O-methylguanosine-AP (Gm-AP); Ion 3 = 2’-O-methyluridine-AP (Um-AP); Ion 4 = deazaguanosine-AP (dzGuo-AP); Ion 5 = adenosine-AP (Ado-AP); Ion 6 = 2’-O-methyladenosine-AP (Am-AP); Ion 7 = adenosine-di-AP (Ado-Di-AP). Moreover, a novel RPCL (Ion 8) was newly identified, which we propose is adenosine-CH₂-cysteine (Ado-CH₂-Cys). These assumptions were supported by the presence of their protonated precursor ions [M + H]⁺, as well as their specific fragmentation features. The proposed chemical structures, and product ion spectra of NA-associated crosslinks are provided in Supplementary Figure S1–S3, and Table S3. In addition to the above NA modifications, we detected numerous other NA modifications in human urine, but these are yet to be fully identified (e.g., Figures 5A–5C). The structure elucidation of these NA modifications requires a searchable mass spectral database, akin to those currently under development for DNA adductomics ^{140, 141}. In total, we detected over 1,000 modified nucleic acid moieties in urine, giving the first indication of the significant breadth of modifications present in the cellular, and urinary, adductomes.

Figure 5. Proof-of-principle: successful generation of NA adductome maps for (A) 2’-dN adducts, (B) rN adducts, (C) nB nB adducts (modified purines), (D) DRCL, (E) RRCL and (F) RPCL, from a pooled urine derived from six healthy subjects, by LC-Q-LIT-OT-MS with CID fragmentation.

Data are reported as accurate mass-to-charge ratios (m/z; Y axis, left) with retention times (RT; X axis) and associated peak intensities (spot color; Y axis, right), using OriginPro. Eight NA-associated crosslinks (Ions 1-8) were newly identified, and the corresponding isomers of RRCL were further labelled with letter “a” to “d”. Additionally, over 300 ions were noted corresponding to pyrimidine nB (data not shown). The identification of each type of NA modification was achieved by the observation of their protonated precursor ions [M + H]⁺, and their specific fragmentation features in MS². The detailed HRMS monitoring parameters are given in the Supplementary information.

HRMS-based ‘omics analysis always generates a massive dataset of product ion spectra. In the present study, the above NA modifications were characterized, and identified one by one using an ion-map function of Thermo Scientific Xcalibur software (ver. 4.1, Thermo Fisher Scientific Inc., USA) combined with extensive manual inspection (as described in Supplementary Experimental Information), which is a complicated and extremely time-consuming process. This laborious process highlights the need of dedicated, post-data analysis software which can characterize, and identify all the specific fragmentation features derived from all types of NA adducts. To date, post-data analysis software is only available for the DNA adductomics ^{76, 142, 143}. Although several data processing methods have been developed for proteomics or metabolomics (e.g., the software XCMS ¹⁴⁴), there is still a lack of open-access/freely available automated data analysis methods that can detect compounds through tandem reactions¹⁴³, such as by tracking specific product ions or a neutral loss MS² fragmentation, in untargeted NA adductomic approaches.

Despite of the lack of post-data processing software for NA adductomics, we demonstrated the ability to detect novel, and/or unexpected NA modifications such as the modified 2’-dN, rN, DRCL, RRCL and RPCL noted here, and several of the crosslinks we reported previously ⁷³. Combined, this represents a major strength of NA adductomics, and offers a means to fully characterise exposures based upon the spectra of adducts that they induce, together with providing a better understanding of the modifications which lead to mutagenicity.

8. Challenges

There are several challenges to successfully performing NA adductomics, using HRMS, these include: (1) Analytical software. There is a need for high-throughput post-data processing software, as the data sets generated are massive and complex, consisting of various features derived from different types of NA modifications. (2) A comprehensive NA adductome database. Such a database would assist in identification of expected and unexpected adducts. Unlike proteomics or metabolomics, a NA adductome database is not available yet, although DNA adductome databases are in preparation ^{140, 141}. This would allow for searches based upon exact mass, fragmentation pattern, and chemical structures, as well as offering visual representations of the 3D structure, all of which will aid in the identification of NA modifications. (3) Harmonization of LC–HRMS-based DNA, RNA, and NA adductomics approaches. NA adductomics can be performed using a range of HRMS instruments. Many pre-analytical, analytical, and post-analytical factors can impact NA adduct detection, identification, and quantification. Strategies are needed to account for these differences including optimal sample work-up and extraction, removing matrix effects, data normalization, and ion suppression, or using pooled quality control samples and internal standards, for harmonizing cross-platform and cross-laboratory data collected from untargeted HRMS-based NA adductome studies. For example, a reference standardization protocol has been recently proposed for harmonizing large-scale metabolomics data collected across different studies or different HRMS methods ¹⁴⁵. Analogous issues have been addressed for the measurement of cellular and urinary nucleic biomarkers of oxidative stress via the formation of international consortia, such as the European Standards Committee on Oxidative DNA Damage, the European Standards Committee on Urinary Lesion Analysis, and the European Comet Assay Validation Group ¹⁴⁶. We propose that an equivalent consortium, encompassing all that NA adductomics represents (perhaps in conjunction with protein adductomics), would be the beginning of a strategy towards addressing the above challenges. (4) Biological meaning. Adductomics offers the potential for detecting unexpected adducts whose identity and role is unknown, but their presence potentially significant. As has been seen from other ‘omics approaches, this kind of unbiased, untargeted approach has the potential to reveal new features for mechanistic investigation, but the feature needs to be detected before a biological meaning can be sought. Achieving this requires effort across multiple fields of study. However, a biological meaning is not a prerequisite for use. For example, a particular pattern of gene expression can be ascribed to be the consequence of exposure to a particular agent, without knowing the function of all the genes. The equivalent is likely to be true for NA modifications.

9. Future perspectives, and conclusions

We suggest that the NA adductomic approach (cellular, urinary, and other biomatrices) to biomonitoring environmental exposure will offer a valuable means to contribute to evaluating the role of damage to all nucleic acids in health and disease. For the first time, through the detection of over 1,000 modified NA moieties, we illustrated the scale of the NA adductome further emphasising the need, and potential benefits of the NA adductomic approach, which is analogous to the other ‘omics approaches. The impact of this methodology extends to the study of any disease for which there is an environmental component, and for which our approach will facilitate Adductome-Wide Association Studies (AWAS; an untargeted, agnostic, hypothesis-generating approach for exploring nucleic acid/protein modifications associated with health outcomes, complementary to GWAS, Epigenome-WAS, and Exposome-WAS). As show in Figure 6, we propose applying our approach to the creation of a library of NA adductome maps, consisting of the integrative and informative NA adductome profiles for environmental agents. More broadly, this can be combined with the study of associations between DNA, and RNA modifications, and the relationships between environment-induced chemical modifications, and epigenomic modification (which we term the “NA adductome network”). Our approach also has utility for mechanistic studies e.g., for more comprehensively identifying the processes responsible for genotoxicity; elucidating the spectrum of physiologically relevant NA modifications that are targeted by discrete repair pathways; improving our understanding of how environmental agents impact epigenetics; and performing more detailed exploration of the relationship between NA adductome signatures, mutational signatures ^{77, 147, 148}, epigenome/epitranscriptome signatures ^{53, 149}, and metabolome signatures ¹⁵⁰. Combined, this will provide the most comprehensive, multi-adductomic characterizations of environmental agents to date, vital to identifying the components of the exposome which are associated with disease, their mechanisms, and a potential utility in disease risk assessment.

Figure 6. Illustration of how Adductome-Wide Association Studies (AWAS) may be used to explore the environmental causes of cancer.

The proposed approach may be used to investigate the link between exposure to environmental agents, and health risk. Top panel. The premise is that the exposome results in modification of all types of nucleic acids [DNA, RNA, (d)NTPs], which is associated with an increased risk of developing major diseases, such as cancer. Main figure. In vitro and in vivo models are exposed to a panel of environmental toxicants, individually, and cellular and urinary NA adductome maps are created. These are used to develop a library of NA adductome maps and a network of inter-related NA modifications. Associations are then investigated between NA adductomic “signatures” of specific exposures and mutational, epigenomic, metabolomic, and epitranscriptomic which, together, may elucidate the underlying etiology and/or mechanism(s) of disease.