DNA-protein crosslink formation by endogenous aldehydes and AP sites

Abstract

Covalent binding between proteins and a DNA strand produces DNA-protein crosslinks (DPC). DPC are one of the most deleterious types of DNA damage, leading to the blockage of DNA replication and transcription. Both DNA lesions and endogenous products with carbonyl functional groups can produce DPC in genomic DNA under normal physiological conditions. For example, formaldehyde, the most abundant endogenous human carcinogen, and apurinic/apyrimidinic (AP) sites, the most common type of endogenous DNA lesions, has been shown to crosslink proteins and/or DNA through their carbonyl functional groups. Unfortunately, compared to other types of DNA damage, DPC have been less studied and understood. However, a recent advancement has allowed researchers to determine accurate yields of various DNA lesions including formaldehyde-derived DPC with high sensitivity and specificity, paving the way for new developments in this field of research. Here, we review the current literature and remaining unanswered questions on DPC formation by endogenous formaldehyde and various aldehydic 2-deoxyribose lesions.

1. Introduction

The cells in our body are constantly exposed to a mixture of various endogenous and exogenous electrophiles. These reactive compounds can damage lipids, proteins, RNA and DNA, leading to age-associated human diseases including cancers. In fact, even under normal physiological conditions, there is a wide variety of endogenous electrophilic molecules in our cells that are continuously damaging our DNA. Steady-state levels of endogenous DNA damage in mammalian cells have been measured using different methods, and the total number of endogenous DNA lesions reportedly exceeds 37,000 lesions per genome, as quantitated by the Swenberg and Nakamura groups (Table 1, steady-state levels are expressed as lesions/genome or lesions/dG in this article) [1,2,11,3–10]. The electrophilic molecules producing endogenous DNA damage include reactive oxygen species (ROS), which are often derived from the mitochondria [12] and play important roles in inflammation [13]; reactive nitrogen species generated during inflammation [14]; S-adenosylmethionine, which is a major biological methyl donor [15]; N-nitroso-dimethylamine produced by the intestinal flora [16]; and formaldehyde (FA) which is generated as a natural by-product during cellular metabolism [17] and is an essential compound that feeds into the one-carbon cycle important for the generation of DNA and proteins [18].

Table 1.

Steady-State Levels of Endogenous DNA Damage1)

Endogenous DNA lesions	Number per cell	Potential crosslinking mechanism	Disadvantage	Method [Ref]
AP sites (including oxidative deoxyriboses)	30,000	Native AP sites/oxidative 2-deoxyribose	Limitted specificity	ASB [25]
8-OxodG	3,500	Native AP sites (BER)	More prone to artifact	LC-MS/MS [5]
N7-MethylG	2,300	Direct DPC at C8 position/Native AP sites (depurination)		LC-MS/MS [6]
N7-OEG	550	Aldehydic lesions/Native AP sites (depurination)		LC-MS/MS [3]
N6-HOCH2-dA adducts (Formaldehyde)	400	Bi-functional-like chemical	Reversible adduct	LC-MS/MS [70]
N7-HEG	300	Native AP sites (depurination)		GC-MS [2]
N2-HOCH2-dG adducts (Formaldehyde)	300	Bi-functional-like chemical	Reversible adduct	LC-MS/MS [68]
N2-Ethylidene-dG adducts (Acetaldehyde)	230	Bi-functional-like chemical	Reversible adduct	LC-MS/MS [8]
Acrolein-dG (Acrolein)	100	Bi-functional chemical (ring-opening)		LC-MS/MS [3]
M1dG Malondialdehyde)	50	Bi-functional chemical (ring-opening)		LC-MS/MS [3]
N2,3-EthenodG (4-hydroxy-2-nonenal)	30	Bi-functional chemical (ring-opening)		LC-MS/MS [3]
dG-Me-Cys crosslink (DPC, Formaldehyde)	30	Bi-functional-like chemical	Reversible adduct	LC-MS/MS [68]
1,N2-EthenodG (4-hydroxy-2-nonenal)	25	Bi-functional chemical (ring-opening)		LC-MS/MS [3]
dG-Me-dG crosslink (Formaldehyde)	20	Bi-functional-like chemical	Reversible adduct	LC-MS/MS [7]
1,N6-EthenodA (4-hydroxy-2-nonenal)	20	Bi-functional chemical (ring-opening)		LC-MS/MS [3]
4-HNEdG (4-hydroxy-2-nonenal)	11	Bi-functional chemical (ring-opening)		LC-MS/MS [3]
CrdG (Crotonaldehyde)	3	Bi-functional chemical (ring-opening)		LC-MS/MS [3]
O6-MethyldG	3			LC-MS/MS [6]
1) Adapted from [1] with permission

DNA damaged by endogenous electrophilic molecules can result in gene mutations, DNA replication stress, and transcriptional errors. Among the various DNA lesions, DNA-protein crosslinks (DPC) cause the largest DNA damage and are thought to result in toxicity. Compared to other types of DNA lesions, there is less understood about the mechanisms of formation, pathophysiological effects, and repair pathways of DPC. Particularly lacking in the research on DPC-mediated biological effects has been a sensitive and specific quantitative analysis for DPC, particularly DPC produced by endogenous electrophilic molecules.

Currently, there are three known mechanisms underlying endogenous DPC formation: (1) endogenous aldehydic DNA lesions crosslinking with proteins (e.g., native apurinic/apyrimidinic (AP) sites and oxidized 2-deoxyriboses); (2) reactive aldehydes crosslinking DNA and proteins; and (3) proximal enzymes transiently crosslinking to DNA, forming covalent DNA-enzyme intermediates (e.g., topoisomerase I and II).

In this review, we first discuss the nature, yields, and potential for DPC formation of aldehydic DNA lesions, with a highlight on AP sites which are the most abundant type of endogenous DNA lesions that exist under normal physiological conditions. We then focus on recent advances in understanding the yield and distribution of DPC caused by endogenous FA, which is the most abundant reactive aldehyde present in our cells under normal conditions. DPC caused by topoisomerase and PARP-1 through AP sites will not be discussed in this review (see [19,20] and other articles in this special issue).

2. AP sites

In the last 25 years, a significant advancement in analytical instruments, such as liquid chromatography-tandem mass spectrometry (LC-MS/MS), has allowed researchers to determine accurate yields of various DNA base adducts with high sensitivity and specificity. Consequently, steady-state levels of endogenous DNA base lesions have been able to be reproducibly quantitated in many studies (Table 1) [2,4–9,21]. While some research groups have demonstrated significant progress in the quantitative analysis of endogenous 2-deoxyribose lesions, such as native AP sites and oxidized 2-deoxyribose lesions using LC-MS/MS [22,23], the technology available for accurate and reproducible quantification of 2-deoxyribose lesions is still lacking, especially compared to DNA base lesion analysis [24]. In part, this may be due to the chemically labile characteristic of 2-deoxyribose lesions. For example, 2-deoxyribonolactone, which is a type of oxidized 2-deoxyribose lesion, is an alkali-labile lesion. DNA isolation and preparation under basic and/or heated conditions can cause considerable β-/δ-elimination, leading to the underestimation of such sugar lesions. Quantifying, identifying the structure, and understanding the characteristics of endogenous 2-deoxyribose lesions is very important, as different 2-deoxyribose lesions can cause genotoxicity at different efficiencies (e.g., DPC and inter-strand crosslinking lesion (ICL) formation, gene mutations, and DNA replication errors).

In 1998, the Nakamura group first established an Aldehyde Reactive Probe (ARP, N-(aminooxyacetyl)-N’-biotinylhydrazine)-Slot Blot (ASB) assay for quantitating AP sites in genomic DNA [25]. ARP is a biotinylated reagent with a hydroxylamine group that forms a very stable Schiff base with aldehydic AP sites (Fig. 1). ARP, as indicated by the name itself, has the ability to make stable bonds with many aldehydic DNA lesions including AP sites [26]. In this assay, AP sites were excised from the genomic DNA backbone by treatment with AP endonuclease and DNA polymerase β or chemicals inducing β-elimination. As the initial AP site signals were mostly eliminated after this reaction, it was concluded that most of the AP sites detected by the ASB assay are aldehydic 2-deoxyribose lesions [25,27,28]). As discussed later, all 2-deoxyribose lesions recognized by ARP contain aldehydic moieties; therefore, such electrophilic DNA lesions have potential for DPC formation (Table 1) (Fig. 2) [29]. Through this assay, steady-state levels of AP sites in genomic DNA extracted from animal and human tissues were determined, and it was discovered that AP sites are one of the most abundant endogenous DNA lesions (~30,000 lesions/genome) (Table 1) [10,27].

Fig. 1. Derivatization Method for Native AP Sites and C1-Oxidized Deoxyriboses.

ARP [25], O-(tetrahydro-2H-pyran-2-yl)hydroxylamine [11], and O-(pyridin-3-yl-methyl)hydroxylamine [23] covalently react with native aldehydic AP sites through the O-hydroxylamine moiety. Pentafluorophenyl hydrazine covalently binds to 5-methylene-2(5H)-furanone derived from 2-deoxyribonolactone [22].

Fig. 2. Native AP site chemical reactions forming N-CH₂-N and N-CH₂-S linkages from FA, proteins, and nucleobases.

Native AP sites are in equilibrium between the ring-closed form and the ring-opened aldehydic form. The latter can reversibly bind to the amino group of proteins (green) via a Schiff base formation, leading to the formation of an unstable DPC. Subsequently, there is a cleavage reaction of the DNA strand through β-elimination, after which the protein can be dissociated from the 3’-AP sites. Alternatively, DPC can be formed between native aldehydic AP sites and the N-terminal Cys of proteins (blue). The AP sites can reversibly bind to the α-amino group of proteins via a Schiff base formation, leading to DPC formation. The DPC then rearranges to produce a reversible thiazolidine linkage between the AP site, the α-amino group, and the sulfhydryl moiety of its amino-terminal Cys.

There are 2 types of AP sites that exist in genomic DNA. The first category includes native AP sites (Fig. 2), which are derived from spontaneous or enzymatic cleavage of the N-glycosidic bond between 2-deoxyribose and a base. The second category includes 2-deoxyriboses oxidized by ROS (hereafter referred as oxidized 2-deoxyriboses) (Fig. 3). In this section, we mainly focus on the nature of AP sites in cells and tissues under normal conditions and their potential to crosslink with proteins. The detailed mechanisms underlying AP site-derived DPC formation and their repair pathways will be discussed by other authors in this special issue.

Fig. 3. Structures of a nucleoside and the major aldehydic/ketonic oxidized 2’-deoxyriboses attached to a phosphodiester backbone.

(modified from [24] with permission)

2.1. Native AP sites

Native AP sites exist in an equilibrium between the ring-closed form and the ring-opened aldehydic form (Fig. 2). Native AP sites are generated by (1) the spontaneous depurination and depyrimidination of normal bases as well as unstable modified bases (e.g., N⁷-methylG, Table 1), or (2) the elimination of modified bases (e.g., 8-oxoG and N3-methylA, Table 1) by DNA glycosylase in the first step of the base excision repair (BER) pathway.

There have been several studies that have attempted to quantify the spontaneous depurination rate of bases. In 1972, Lindahl and Nyberg [30] measured the spontaneous depurination rate at 70°C by quantitating depurinated purines using bacterial DNA with radioisotope-labeled bases. The results were extrapolated to depurination rates at 37°C. They estimated the depurination frequency at rates ranging from 2,400 to 12,000 events/cell/day (Note: the rate of formation is described as events/day in this article). In 1998, the Nakamura group directly measured the depurination rate by quantitating an increase in AP sites during a 10-day incubation of DNA in phosphate-buffered saline (PBS) at 37°C. The rate of depurination as determined by the ASB assay was approximately 9,000 events/cell/day [25]. The depurination rate of packed genomic DNA in cells has yet to be determined; however, if we estimate that the depurination rate in cells is similar to the rate of AP site formation in tubes, the rate of spontaneous depurination in our bodies can be calculated as follows. Our body consists of approximately 3×10¹² nucleated cells, as was determined by a recent re-evaluation [31]. The total number of nucleated cells in our body is multiplied by the rate of AP site formation under physiological conditions, leading to the total depurination rate in our body, which is approximately 2.7×10¹⁶ events/human/day.

Researchers have also tried to quantify the number of native AP sites. In 2013, Chan and coworkers measured the number of native AP sites in human cultured cells using LC-MS/MS combined with a derivatization of pentafluorophenyl hydrazine (containing an aldehyde-reactive - NH-NH₂ moiety) [32]. They detected about ~7,500,000 native AP sites/genome in intact TK6 cells. Furthermore, their group determined an increase in the number of native AP sites in the cells exposed to methyl methanesulfonate (MMS). However, the number of native AP sites induced by MMS was at odds with published results showing a much fewer number of N⁷-methylG, which is the dominant base lesion caused by MMS [33]. These high levels of AP sites in the intact and MMS-treated cells are likely due to artifactual AP site formation during DNA preparation. Data recently published by Turesky’s group quantitated the number of native AP sites in nuclei isolated from rat liver using LC-MS/MS combined with a derivatization of O-(pyridin-3-ylmethyl)hydroxylamine (Fig. 1, containing the aldehyde-reactive -O-NH₂ moiety) [23]. This method detected approximately 550 native AP sites/genome; however, based on the published depurination rate, steady-state levels of AP sites can reach at least 1100 AP sites/genome or higher due to artifactual depurination during DNA preparation [25]. This raises the question about the accessibility and reactivity of O-(pyridin-3-ylmethyl)hydroxylamine to the nuclear native AP sites in the lysis buffer that was used, potentially leading to the underestimation of the number of native AP sites.

Once formed, native AP sites are predominantly repaired by BER enzymes by a hydrolytic reaction without covalent DPC formation (Fig. 2). Furthermore, our cells are normally equipped to prevent genomic instability caused by native AP sites through DPC formation. A complex of native AP sites and DPC formed by the protein HMCES (5-hydroxymethylcytosine binding, the embryonic stem cell-specific protein) has recently been reported to protect native AP sites on single stranded DNA during DNA replication [34,35]. HMCES recognizes native AP sites on the DNA and makes a reversible thiazolidine linkage between a ring-opened native AP site, an α-amino group, and the sulfhydryl moiety of its amino-terminal cysteine (Cys) (Fig. 2). This DPC with a thiazolidine linkage appears to protect AP sites from mutations caused by error-prone polymerases as well as strand breaks through β-elimination. Most recently, Chan et al. published data on endogenous DPC levels in intact HeLa cells by measuring native AP site-Cys (S-glycosidic) linkages using LC-MS/MS [36]. Again, they found a large number of endogenous DPC at around 250,000 DPC/genome.

Importantly, further progress on LC-MS/MS-based assays will be critical for improved sensitivity and accuracy in the detection of native AP sites and AP site-mediated DPC complexes, for example in the assessment of artifactual DPC formation, the removal of existing native AP sites through β-elimination and the dissociation of proteins from DPC during DNA preparation, and the accessibility of derivatizing agents to target DNA lesions.

2.2. Oxidative 2-deoxyribose

Aldehydic oxidative 2-deoxyriboses generated from exogenous factors are often produced by ionizing radiation or iron-mediated Fenton and copper-mediated Fenton-like reactions [28,37–43]. However, oxidative 2-deoxyriboses can also be generated endogenously. In fact, endogenous AP sites in intact cells and tissues appear to possess similar characteristics to aldehydic oxidative 2-deoxyriboses exogenously caused by Fenton/Fenton-like reactions [27,28]. Endogenous oxidative 2-deoxyriboses are generated under ROS conditions, whereby hydrogen atoms are abstracted at each of the five carbon positions in 2-deoxyribose, resulting in a wide assortment of oxidative 2-deoxyriboses (Fig. 3). The resulting oxidative 2-deoxyriboses frequently have aldehyde and ketone moieties in their molecules.

Efforts have been made for the quantification of oxidative 2-deoxyriboses. Dedon’s group has established several new assays for measuring oxidative 2-deoxyriboses (C1- and C5-oxidation [22]; C3-oxidation [44]; C4-oxidation [45]). They have quantitated oxidation products at the C1 and C5 positions of 2-deoxyriboses using LC-MS/MS combined with pentafluorophenyl hydrazine (Fig. 1) [22]. Steady-state/background levels of 2-deoxyribonolactone (products of C1-oxidation) and nucleoside 5’-aldehyde products of 2-deoxyribose oxidation (products of C5-oxidation) were measured at ~56,000 and ~423,000 lesions/genome, respectively. Compared to native AP sites in the liver of intact rats, 100 to 800 times higher amounts of 2-deoxyribose lesions exist as a result of ROS reactions under normal physiological conditions [22,23]. Further investigation utilizing these assays after minimizing the contribution of oxidative stress during DNA preparation would be informative. Although accurate quantification of true steady-state levels of endogenous native AP sites and various oxidative 2-deoxyriboses has not yet been achieved, we predict that this will be accomplished in the near future with the establishment of sensitive and specific LC-MS/MS-based analyses and further improvements to instruments.

As previously mentioned, both native AP sites and oxidative 2-deoxyriboses contain electrophilic aldehydes and ketones in their molecules; therefore, 2-deoxyribose lesions have the potential for generating DPC via reactions with the amino moieties in proteins [29,46,47]. While aldehydic native AP sites are predominantly repaired by BER enzymes without forming a covalent DPC (Fig. 2), 2-deoxyribonolactone and its β-elimination product bind to the lysines of BER proteins, leading to a stable, irreversible DPC [48–50]. The biological importance of 2-deoxyribonolactone-derived DPC will be discussed by Quiñones et al. [51] in this special issue.

As endogenous oxidative 2-deoxyriboses are generated in ROS conditions, it is important to study these genetic instabilities caused by normal and pathological levels of oxidative stress. Importantly, intermittent H₂O₂ exposure (every 10 min for 1 h) to cultured cells at low concentrations exponentially increases AP sites, likely oxidative 2-deoxyriboses [39]. Interestingly, intermittent H₂O₂ exposure at a lower concentration causes a markedly larger number of AP sites in the cells compared to bolus H₂O₂ treatment with the same total amount of H₂O₂. Furthermore, H₂O₂ at low concentrations induces G1-dependent double strand breaks (DSBs) and oxidative clustered DNA lesions (OCDLs) in cultured cells [5]. Clustered AP sites appear to efficiently cause DSBs through histone-derived DPC formation [52]. With increasing age, we become more susceptible to chronic non-infectious inflammatory diseases including nonalcoholic steatohepatitis wherein there is continuous elevation of oxidative stress. Under such high oxidative stress conditions, OCDLs consisting of native AP sites, oxidative 2-deoxyriboses, and oxidized bases may result in even more DPC, leading to the formation of DSBs and complexed clustered DNA lesions with DPC.

ROS generates a wide variety of oxidative 2-deoxyribose lesions, and 2-deoxyribonolactone is just one of the many oxidative 2-deoxyriboses. A better understanding of the structure of major oxidative 2-deoxyriboses that accumulate in cells under normal physiological conditions as well as under oxidative stress environments will help elucidate the biological importance of DPC caused by such DNA lesions.

3. Other endogenous base lesions

Following AP sites the second, fourth and sixth most abundant types of DNA lesions are 8-oxodG, N⁷-(2-oxoethyl)G (N⁷-OEG) and N⁷-(2-hydroxyethyl)G (N⁷-HEG) (Table 1), all of which are generated by ROS in cells under normal conditions. N⁷-HEG and N⁷-OEG are also produced by exogenous exposure to ethylene oxide [2] and vinyl chloride [21], respectively. N⁷-OEG itself has the potential for making DPC through its aldehydic moiety. In addition, 8-oxodG, N⁷-OEG, and N⁷-HEG all produce native AP sites through spontaneous depurination or by DNA glycosylase; therefore, all of these base lesions could potentially produce DPC.

Lipid peroxidation can also produce very reactive α,β-unsaturated aldehydes, such as acrolein, 4-hydroxy-2-nonenal, and malondialdehyde (MDA, β-hydoxyacrolein). These electrophiles produce exocyclic base damage, which can rearrange during ring-opening, resulting in aldehydic base lesions (Fig. 4) [53–57]. The ring-opened base lesions caused by α,β-unsaturated aldehydes also have the potential to make DPC. In fact, DPC formation in vitro and in cells has been reported after exposure to α,β-unsaturated aldehydes [58].

Fig. 4. α,-β Unsaturated aldehyde leading to aldehydic base lesions.

Lipid peroxidation-derived α,-β unsaturated aldehydes produce dG adducts, which are in equilibrium between the ring-opened aldehydic form and the ring-closed exocyclic form (modified from [53] with permission)

Interestingly, a recent study demonstrated that N⁷-methylG, the third most abundant DNA lesion (Table 1) which is likely derived from an endogenous methyl donor, reversibly binds to proteins at the C8 position of guanine [59,60]. However, the abundance and biological significance of N⁷-methylG need more investigation.

Overall, most of the endogenous DNA lesions are either aldehydic DNA lesions or the source of native aldehydic AP sites; therefore, these lesions could directly or indirectly make DPC under physiological conditions. If these initial endogenous DNA lesions are not repaired in a timely manner, the further increase in DPC formation can cause serious biological consequences.

4. Formaldehyde (FA)

4.1. Exogenous FA

Before we discuss the effects of endogenous FA, the biological consequences after exposure to exogenous FA needs to be mentioned for a better understanding of this toxic electrophilic molecule. Exogenous FA has been classified as a known human and animal carcinogen by the International Agency for Research on Cancer (IARC) [61–63]. FA is continuously released into the environment through combustion processes, such as automobile exhaust and tobacco smoke [62]. Even without combustion, heating tobacco products generates a significant amount of FA through the degradation of propylene glycol and glycerol, both of which are utilized for vapor generation [64]. In addition, FA has been widely used for industrial manufacturing processes, such as the production of paper, textiles, carpets, chipboard and plywood, thermal insulation foams, adhesives, resin, and others; therefore, workers at plants manufacturing such products can be exposed to high levels of FA [62]. Inhaled exogenous FA is first deposited on the nasal mucous membrane and absorbed into the nasopharyngeal epithelium as the first site of contact. Extended exposure to high levels of exogenous FA can eventually lead to the formation of nasopharyngeal squamous cell carcinoma (SCC) [61,62]. Studies have demonstrated that exogenous FA inhalation exposure causes an increase in the incidence of nasopharyngeal SCC in rats at 6 ppm or higher (Fig. 5A, data in Fig. 5 are adapted from [61,65,66]).

Fig. 5. Tumor and DPC formation in the nasal tissue of animals exposed to formaldehyde.

A. Nasal tumor (SCC) incidence in rats exposed to formaldehyde by inhalation for 18 and 24 months (6 hour/day, 5 days/week) [61,65]. B. Endogenous and exogenous FA-induced DPC (dG-Me-Cys) in the nasal tissue of rats exposed to air control vs. 15 ppm of [13CD2]-formaldehyde (6 hour/day) [9]. C. Endogenous and exogenous FA-induced DPC (dG-Me-Cys) in the nasal tissue of rats exposed to 2 ppm of [13CD2]-formaldehyde (6 hour/day, 5 days/week) [9]. D. Effects of exogenous FA exposure on endogenous FA-derived DPC (dG-Me-Cys) in rats exposed to air control vs. 15 ppm of [13CD2]-formaldehyde for 4 consecutive days (6 hour/day) [9] E. Endogenous and exogenous FA-induced DPC (dG-Me-Cys) in the nose and bone marrow of non-human primates (NHP, Cynomolgus Macaque) exposed to air control vs. 6 ppm of [13CD2]-formaldehyde for 2 days (6 hour/day) [9].

FA has been most extensively studied in the mechanism of DPC formation in the context of FA-mediated carcinogenesis. Numerous studies on FA carcinogenesis have proposed that the etiology of nasopharyngeal SCC caused by FA exposure is due to a combination of DPC formation, cytotoxicity, inflammation, and regenerative cell proliferation [67]. Recently, a very sensitive LC-MS/MS assay for quantitation of endogenous and exogenous FA-induced DPC has been developed and applied to tissues of animals exposed to exogenous FA [9,68,69]. While the details of FA-derived DPC will be discussed later in this review, it is important to mention here that exogenous FA causes an increase in DPC that can persist in animal tissues for an extended period of time. Specifically, after inhalation exposure to stable isotope-labeled exogenous FA ([¹³CD₂]-FA) in rats and monkeys, endogenous (unlabeled) and exogenous ([¹³CD₂-labeled) DPC were measured in the nasal tissues as well as other organs/tissues by LC-MS/MS [9,68,69]. The nasal epithelium of rats exposed to [¹³CD₂]-FA at 15 ppm for 4 consecutive days (6 hours/day) reportedly showed an accumulation of exogenous DPC that reached 1.8 crosslinks/10⁷dG on Day 4, which is much higher than the level of endogenous DPC formation (0.37 crosslinks/10⁷dG) (Fig. 5B). Likewise, exogenous FA exposure at 2 ppm for 4 weeks (6 hours/day) also induced an accumulation of exogenous DPC; however, the DPC yield did not exceed endogenous levels (Fig. 5C). Furthermore, exogenous DPC levels persisted for up to 1 week during the recovery period after a 4-week exposure (Fig. 5C). Likewise, monkeys exposed to exogenous FA at 6 ppm for 2 consecutive days also showed very similar results to data from rats (Fig. 5E). Both the accumulation of FA-derived DPC during the exposure period and its persistence during the recovery period are strongly suggestive of a poor repair efficiency of FA-induced DPC. As mentioned above, the etiology of inhaled FA-mediated nasopharyngeal SCC has been proposed to be due to a combination of FA-derived cytotoxicity, inflammation, regenerative cell proliferation, and DPC formation. However, FA induces many different types of DNA lesions in addition to DPC. Therefore, the importance of DPC formation in FA carcinogenesis needs to be further elucidated. In addition to nasopharyngeal SCC, previous epidemiological studies have shown that FA exposure may result in the development of leukemia in humans [62,63]. In contrast, although rats and monkeys exposed to exogenous inhaled FA did induce exogenous DPC in nasal tissue, there were no exogenous DPC in the bone marrow or other organs/tissues (Fig. 5E) [9,68]. Based on these results, Swenberg’s group has proposed that it is unlikely that inhaled FA increases the risk for leukemia [9,68,69]. With these conflicting results, it is important to continue to investigate whether and how inhaled exogenous FA may cause leukemia.

4.2. Endogenous FA

4.2.1. Sources of endogenous FA

The serendipitous discovery of endogenous base damage was made while researchers were investigating and identifying exogenous DNA base damage [2,21]. Similarly, FA-derived DNA lesions were identified in cultured cells and animal tissues in conditions without any exposure to exogenous FA [7,9,70]. Our body can in fact naturally produce FA under both normal physiological conditions as well as in pathophysiological conditions such as at sites of inflammation. Thus, a significant amount of endogenous FA can always be detected in our body. Endogenous FA can be generated through many different mechanisms. For example, oxidative demethylation of methylated histones produces FA as a by-product during epigenetic regulation of gene expression [17]. Likewise, enzymatic oxidative demethylation of methylated RNA and DNA produces FA by a similar reaction [71]. In some cases, in these oxidative demethylation reactions, hydrogen peroxide (H₂O₂) can also be generated in addition to FA, including during the demethylation of histones. Both FA and H₂O₂ are genotoxic, particularly when they are produced in close proximity to DNA. FA can also be produced as a by-product in demethylation and deamination reactions outside of the nucleus. For example, FA is generated as a by-product by myeloperoxidase [72] released from neutrophils. In addition, FA is produced by vascular adhesion protein-1 [73], which is expressed in vascular endothelial cells and regulates leukocyte adhesion and extravasation at various sites of inflammation. Additionally, FA can be generated during the one-carbon cycle metabolism, likely derived from the spontaneous degradation of folate derivatives [18].

4.2.2. Toxicity of Endogenous FA

The average endogenous FA concentration has been reported at around 100 μM in human blood [62,74–77]. In cultured cells, this level of FA is high enough to cause toxicity. In 2007, the Nakamura group discovered that levels of FA comparable to those in plasma caused severe toxicity in chicken-derived B-lymphoblastoid cells deficient in various genes involved in homologous recombination, nucleotide excision repair, translesion DNA synthesis, and ICL repair (i.e. Fanconia anomia complementation group FANC genes), as determined by differential cell toxicity assays [78,79]. Acetaldehyde at much higher concentrations compared to FA also caused toxicity in cells deficient in FANC genes [78]. These results suggest that endogenous FA and alcohol consumption-derived acetaldehyde may cause genotoxicity through crosslinking products in humans. In contrast, bi-functional aldehydes such as methylglyoxal, acrolein, and crotonaldehyde did not induce differential toxicity in FANCD2-deficient chicken cells [78]. Ide’s group also demonstrated the differential toxicity of FA and acetaldehyde in a series of human cells deficient in FANC genes and CHO cells deficient in homologous recombination, nucleotide excision repair, or Fanc genes, and less or no toxicity with acrolein, crotonaldehyde, an malondialdehyde using LD10 doses [58]. They concluded that FA and acetaldehyde may cause differential toxicity in the cells deficient in FANC and homologous recombination due to DNA lesions other than DPC (e.g., ICL). This potential toxicity of endogenous FA is corroborated by evidence that chicken-derived B-lymphoblastoid cells deficient in both FANCD2 and the formaldehyde removal enzyme ADH5 are not viable [80]. Cells deficient in ADH5 are believed to have abnormally high endogenous FA concentrations, and mice deficient in ADH5 and FANCD2 show accelerated bone marrow failure [81]. These results indicate that endogenous FA can cause bone marrow failure when the ICL repair pathway is compromised.

4.2.3. FA and DPC formation (in vitro)

FA is a very powerful electrophile that can cause DPC formation through N-CH₂-N and N-CH₂-S linkages (Fig. 6) [82]. Due to a higher nucleophilic potential with proteins compared nucleobases, the carbon of FA is likely attacked first by the primary amine of lysine, leading to methylol adducts of proteins [83,84]. Dehydration leads to the formation of a Schiff base, which subsequently reacts with the amino moieties of nucleobases, resulting in DPC formation with methylene linkages. To try to better understand the different players involved in DPC linkage, Lu et al. incubated FA with different Nα-Boc-protected amino acids (Lys, Cys, histidine, tryptophan, arginine, asparagine, glutamine, and tyrosine) and each nucleoside (dA, dT, dC and dG) in potassium phosphate buffer (pH 7.2) for 48 hours at ambient temperature [85]. They discovered that FA made linkages between nucleosides and four amino acids including Lys, Cys, histidine, and tryptophan. The most abundant crosslink was between dG and the side chain amino group of Lys (dG-me-Lys) (Fig. 6A). Unfortunately, this crosslink product was not stable enough for the characterization and quantification of DPC due to the reversible crosslink reaction. The second most abundant crosslink product was a linkage between dG and the sulfhydryl group of Cys (dG-me-Cys) (Fig. 6B). In contrast to the dG-me-Lys crosslink, the dG-me-Cys crosslink was stable enough to purify and quantitate the target molecules. Based on these results, the researchers decided to utilize dG-me-Cys crosslink products as a biomarker of DPC caused by FA in cells and animals [85].

Fig. 6. Chemical reactions forming N-CH₂-N and N-CH₂-S linkages from FA, protein, and nucleobase.

A. N-CH₂-N linkage: FA reacts with the amino group of proteins, resulting in the formation of methylol adducts on proteins. After dehydration, the Schiff base further binds to the nucleophilic amino moiety of the nucleobase through a methylene linkage (N-CH₂-N). B. N-CH₂-S linkage: FA first binds to the thiol group of Cys in proteins, leading to methylol adducts on the protein. The methylol adducts react with nucleobases after dehydration. All of the reactions (A and B) are reversible (modified from [83,103] with permission).

4.2.4. Endogenous FA-induced DPC in animals

Before the development of an LC-MS/MS analysis for measuring dG-Me-Cys, DPC induced by FA were quantitated by different nonspecific DPC assays, such as the sodium dodecyl sulfate/potassium (SDS/K⁺) precipitation method [86,87]. The SDS/K⁺ precipitation method requires the incubation of cell extracts in the presence of SDS/K⁺ at 65°C for around 10 min in order to dissociate proteins from non-covalently bound DNA. Most of the nuclear matrix proteins [88] not covalently bound to DNA can be dissociated during this process. The amount of DNA covalently bound to protein is then quantitated after protein precipitation to get rid of non-covalently bound DNA. Unfortunately, these non-specific methods cannot distinguish between DPC directly produced by FA, DPC indirectly generated by FA (e.g., DPC caused by oxidative stress through FA-mediated inflammation), and DPC induced by other crosslinking agents coexisting with FA. Furthermore, the dissociation of proteins and non-covalently bound DNA at high temperatures in the presence of SDS/K⁺ may cause the contamination of unwanted proteins non-covalently bound DNA, resulting in an overestimation of DPC. Furthermore, such high temperatures can also accelerate the dissociation of reversible DPC, leading to the underestimation of true DPC.

Recently, a very sensitive LC-MS/MS method was established for quantitating FA-derived DPC (dG-Me-Cys) with high sensitivity and specificity [9,68,69]. This method allows for researchers to specifically quantitate FA-derived DPC but also distinguish between endogenous and exogenous FA-derived DPC using stable isotope-labeled FA ([¹³CD₂]-FA). In this review, we will not cover the detailed methods of the DNA preparation and LC-MS/MS analysis (see [9,68,69]). But briefly, DPC were first isolated from animal tissues and cultured cells using DNAzol in the presence of endopeptidase (e.g., proteinase K). After ethanol precipitation, DNA and proteins were further enzymatically digested to nucleoside and amino acid levels. Lastly, LC-MS/MS analysis was then performed after HPLC purification to eliminate enzymes and unwanted digested products. As mentioned earlier, endogenous DNA damage can look identical to exogenous DNA lesions. In such cases, the exposure of stable isotope-labeled compounds to animals and cultured cells allows for the measurement of endogenous (unlabeled) and exogenous (labeled) DNA damage simultaneously using LC-MS/MS. This methodology can also determine the effects of exogenous agents on the repair or accumulation of DNA damage using identical endogenous reactive agents. For example, when animals are exposed to exogenous FA ([¹³CD₂]-FA, 3Da larger than native FA), tissues are also exposed to endogenous (non-labeled) FA at the same time. Consequently, DNA isolated from tissue samples of the exposed animals contain DPC (dG-[¹³CD₂]-Me-Cys) derived from exogenous FA and DPC (dG-Me-Cys) produced by endogenous FA. LC-MS/MS distinguishes between the two products by the 3Da mass difference. In this particular study, F344 rats were exposed to exogenous [¹³CD₂]-FA by inhalation for the quantitation of endogenous and exogenous FA-derived DPC as well as mono dG adducts (N²-HOMe-dG) [9,68].

As mentioned earlier, exogenous DPC (dG-[¹³CD₂]-Me-Cys) were markedly increased in the nasal mucosa in rats exposed to [¹³CD₂]-FA at 15 ppm for 4 consecutive days [9] (Fig. 4B). In contrast, endogenous DPC (dG-Me-Cys) were significantly decreased on Day 4 compared those in the air control group (Figs. 4B and D). The decrease in endogenous DPC was not observed in the peripheral blood cells or bone marrow of the same FA-treated rats. While the lower FA concentration (2 ppm) group showed similar trends to a lesser extent, at very low concentrations (0.3 ppm or lower) no such effects were detected [68]. Therefore, it can be postulated that exogenous FA-derived DPC could stimulate repair of endogenous FA-derived DPC at certain threshold levels. If rats were exposed to exogenous FA at 15 ppm for longer than 4 days, the repair of endogenous DPC could be saturated, perhaps leading to an increase in endogenous DPC as well as exogenous DPC.

Endogenous DPC levels in rats without any exogenous FA exposure have also been measured. In one study, the levels of endogenous FA-derived DPC were compared between different tissues and organs (nasal epithelium, bone marrow, peripheral blood mononuclear cell (PBMC), trachea, liver, brain and lung) [68]. Interestingly, there were large amounts of endogenous DPC (dG-Me-Cys) in the liver of the rats (Fig. 7B). Endogenous DPC (0.727 ± 0.166 crosslinks/10⁷ dG) in the liver were 2.7 times higher than the amount of endogenous DPC (0.266 ± 0.054 crosslinks/10⁷ dG) in the nasal tissues [68]. In the same study, they measured endogenous dG mono (N²-HOMe-dG) adducts as a marker for FA exposure (Fig. 7A, data in Fig. 7 are adapted from [68]). N²-HOMe-dG adducts have been proposed to be generated mainly as a dissociation product of DPC caused by FA [84]. The number of endogenous N²-HOMe-dG was measured at approximately one order magnitude higher than the amount of endogenous DPC in rats, and the mono dG adducts presented at similar levels in all tissues and organs (nasal epithelium, bone marrow, PMBS, trachea, liver, brain and lung) measured in this study (Fig. 7A). To better understand the DPC accumulation rate, the number of endogenous DPC was divided by the number of mono dG adducts. Similar to the number of DPC (Fig. 7B), DPC/mono adduct ratios demonstrated markedly high levels of endogenous DPC over mono adducts in the rat liver (Fig. 7C). Although the number of tissues and organs was limited, endogenous DPC levels in monkeys (1.546 ± 0.198 crosslinks/10⁷ dG) were also 4.3 times higher in the liver compared to the nasal tissue (0.359 ± 0.101 crosslinks/10⁷ dG) [9]. Moreover, the levels of endogenous DPC in monkey livers were slightly below those of exogenous DPC in the rat nasal tissues (1.818 ± 0.723 crosslinks/10⁷ dG) exposed to 15 ppm FA (Fig. 5B) [9], which is a carcinogenic concentration (Fig. 5A) [61,65]. While Adh3/Adh5 expression in the liver was highest among several organs analyzed in mice [81], endogenous FA concentration in the liver appears to be lower than that in the nasal mucosa in rats [89]. At this time, endogenous FA concentrations in the liver cannot explain such high levels of endogenous DPC (dG-Me-Cys) in the liver.

Fig. 7. Endogenous FA-derived dG mono adducts vs DPC in tissues/organs of rats and monkeys.

A. Endogenous FA-derived dG mono adducts (N2-HOMe-dG adducts/10⁷ dG) in rat tissues [68]. B. Endogenous FA-derived DPC (dG-Me-Cys/10⁷ dG) in rat tissues [68]. C. Endogenous FA-derived DPC/dG mono adduct ratio in rat tissues [68]. D. Endogenous FA-derived DPC (dG-Me-Cys/10⁷ dG) in non-human primate (NHP) tissues [9].

Recently several research groups have demonstrated that SPRTN is the DNA-dependent metalloprotease involved in DPC repair [90–92]. C. elegans and cultured mouse and human cells deficient in SPRTN were all hypersensitive to FA [90,91,93]. The persistence of FA-induced DPC measured by a non-specific DPC assay was also detected in cells deficient in SPRTN [90,91,93]; in addition, an increase in the amount of endogenous DPC was also observed in cultured human cells deficient in SPRTN [91]. Interestingly, both Ruijs–Aalfs syndrome (RJALS) patients, who are deficient in SPRTN, and Sprtn hypomorphic mice developed spontaneous formation of hepatocellular adenoma/carcinomas [94,95]. The mechanism by which SPRTN deficiency leads to liver tumors remains unknown. However, observations of a large number of FA-specific endogenous DPC in rat and monkey livers [9,68] point to the possibility of the etiology of liver tumors in the absence of functional SPRTN.

In summary, endogenous FA is abundantly produced by a wide range of biological processes, and its endogenous concentration in our body appears to be at or just slightly below toxic levels. Normally, there is adequate expression of the enzymes and proteins important for counteracting this toxic endogenous aldehyde. However, patients deficient in these protective pathways, such as those with Fanconi anemia or Ruijs-Aalfs syndrome have compromised capacity to repair ICLs or DPC compared to healthy individuals. Perhaps due to the accumulation of FA-induced ICLs or DPC, these patients may experience early onset of bone marrow failure/leukemia or hepatocellular adenoma/carcinoma.

4.3. Super-stable FA-derived hybrid DNA/protein adducts

The toxicology field has particularly been focused on the risk assessment of chemical mixtures for the past two decades [96]. Similarly to environmental contaminants, endogenous and exogenous electrophilic molecules exist as a mixture in our cells. Several studies have investigated whether these electrophiles may additively and/or synergistically produce more toxic DNA and protein damage than those induced by individual agents [72,97,98]. As discussed earlier, MDA derived from lipid peroxidation can generate DPC as a bi-functional aldehyde [58]. However, it has not been well recognized that MDA has the ability to covalently bind to other aldehydic molecules using its center carbon [72]. For example, FA can react with 2 molecules of MDA in the presence of Lys, leading to a very stable hybrid adduct, 1,4-dihydropyridine-type M2FA-Lys (Fig. 8A) [72]. In fact, it makes sense that in physiological conditions, the diffusible, reactive MDA would come in contact and react with FA, which is the most abundant endogenous aldehyde, to make M2FA-Lys. As seen in Figure 8A, this molecule has two functional aldehydic moieties attached to a dihydropyridine ring; therefore, M2FA-Lys can induce crosslinking between proteins. It is possible that FA first produces M2FA-Lys molecules on the target protein, which may further react with amine moieties of nucleobases using the aldehydic moiety on the dihydropyridine ring. In 2000, Kronberg’s group reported that FA reacts with deoxyadenosine in the presence of MDA, resulting in the formation of 9-(2’-deoxy-β-D-ribofuranosyl)-6-(3,5-diformyl-1,4-dihydro-1-pyridyl)purine (M2FA-dA) adducts [98] (Fig. 8B). Unlike FA-induced nucleoside mono adducts (N²-HOMe-dG and N⁶-HOMe-dA), these M2FA-dA adducts appear to be very stable even at high temperatures; thus, it is possible that M2FA-dA could have the potential for inducing DPC through a reaction with the amino moiety of Lys side chain. M2FA-Lys and/or 1,4-dihydropyridine-type Lys adducts has been reported to induce inflammation, antibody production, toxicity, and fibrosis [72,99–102]; therefore, it could be that M2FA-Lys is a pathogenic protein adduct that contributes to the inflammation and toxicity caused by FA inhalation. In fact, it has recently been demonstrated that M2FA-Lys-adducted molecules are present in the plaque of atherosclerosis lesions in animals [72].

Fig. 8. Structures of M2FA-Lys and M2FA-dA.

A. 1,4-dihydropyridine-type malondialdehyde (MDA)-FA-Lys (M2FA-Lys) adducts are formed by a reaction between FA (red) and two equivalents of MDA (blue) with a Lys (black) [72]. B. Likewise, incubation of FA, MDA, and dA produces M2FA-dA [98].

5. Conclusions and Perspectives

DPC derived from endogenous FA and aldehydic DNA lesions – such as native AP sites and oxidized 2’-deoxyribose – are believed to be highly toxic DNA lesions that may cause gene mutations, chromosomal damage, and replication and transcriptional stress under normal physiological conditions. If left unrepaired, the genomic instability caused by DPC can lead to age-associated human diseases including cancers. Although DNA base adductions and their repair pathways have been extensively investigated by many researchers, research on DPC formation and its repair has unfortunately received much less attention. Furthermore, research on endogenous DPC have been limited by technical difficulties, particularly in the availability of reproducible, sensitive, and selective quantitative analyses. Recently, a very sensitive LC-MS/MS assay was developed for quantitating endogenous FA-derived DPC by measuring dG-Me-Cys linkages. With this new technology, we can now ask new questions that previously could not be answered. For example, did dG-Me-Cys that is detected in tissues exist originally as a bulky DPC, a GSH-DPC or a small dG-Me-Cys adduct and is there any association between FA-derived DPC and human diseases? Importantly, questions remain on the reason for the high steady-state levels of endogenous FA-derived DPC in the liver. Is there any association between the etiology of spontaneous liver cancer, such as nonalcoholic steatohepatitis-derived hepatocellular carcinomas, and endogenous FA-derived DPC? Also, it has been reported that dG-me-Lys adducts may not be ideal biomarkers for DPC due to their unstable linkages; however, as dG-meLys is the most abundant FA-induced DPC, it would be informative to understand the yields of this DPC after stabilization of its linkage. Lastly, a better characterization of the structure of the abundant oxidized 2’-deoxyriboses would be helpful in understanding the mechanisms of endogenous deoxyribose-mediated DPC formation. Elucidating these details on endogenous DPC formation and its repair could greatly accelerate our understanding of age-related human diseases.

Abbreviations

AP site

apurinic/apyrimidinic site

ARP

Aldehyde Reactive Probe

ASB assay

ARP-Slot-Blot assay

BER

base excision repair

Cys

cysteine

dA

2’-deoxyadenosine

dC

2’-deoxycytidine

dG

2’-deoxyguanosine

DPC

DNA-protein crosslinks

DSBs

double strand breaks

dT

thymidine

FA

formaldehyde

HMCES

5-hydroxymethylcytosine binding, embryonic stem cell-specific protein

H₂O₂

hydrogen peroxide

IARC

International Agency for Research on Cancer

ICL

inter-strand crosslinking lesion

LC-MS/MS

liquid chromatography-tandem mass spectrometry

Lys

lysine

M2FA-dA

9-(2’-deoxy-β-D-ribofuranosyl)-6-(3,5-diformyl-1,4-dihydro-1-pyridyl)purine

M2FA-lysine

S-2-amino-6-(3,5-diformyl-pyridin-1(4H)-yl)hexanoic acid

MDA

malondialdehyde

MMS

methyl methanesulfonate

N⁷-HEG

N⁷-(2-hydroxyethyl)G

N⁷-OEG

N⁷-(2-oxoethyl)G

OCDLs

oxidative clustered DNA lesions

PBMC

peripheral blood mononuclear cell

PBS

phosphate-buffered saline

ROS

reactive oxygen species

RJALS

Ruijs–Aalfs syndrome

SDS/K⁺

sodium dodecyl sulfate/potassium

SSC

squamous cell carcinoma