Discovery and properties of a monoclonal antibody targeting 8-oxoA, an oxidized adenine lesion in DNA and RNA

Abstract

8-oxoA, a major oxidation product of adenosine, is a mispairing, mutagenic lesion that arises in DNA and RNA when ^•OH radicals or one-electron oxidants attack the C8 adenine atom or polymerases misincorporate 8-oxo(d)ATP. The danger of 8-oxoA is underscored by the existence of dedicated cellular repair machinery that explicitly excise it from DNA, the attenuation of translation induced by 8-oxoA-mRNA or damaged ribosomes, and its potency as a TLR7 agonist. Here we present the discovery, purification, and biochemical characterization of a new mouse IgGk1 monoclonal antibody (6E4) that specifically targets 8-oxoA. Utilizing an AchE-based competitive ELISA assay, we demonstrate the selectivity of 6E4 for 8-oxoA over a plethora of canonical and chemically modified nucleosides including 8-oxoG, A, m6A, 2-oxoA, and 5-hoU. We further show the ability of 6E4 to exclusively recognize 8-oxoA in nucleoside triphosphates (8-oxoATP) and DNA/RNA oligonucleotides containing a single 8-oxoA. 6E4 also binds 8-oxoA in duplex DNA/RNA antigens where the lesion is either paired correctly or base mismatched. Our findings define the 8-oxoAde nucleobase as the critical epitope and indicate mAb 6E4 is ideally suited for a broad range of immunological applications in nucleic acid detection and quality control.

1. Introduction

Oxidative damage to nucleic acids is a seminal feature of cancer, aging, neurodegenerative diseases, and metabolic syndrome [1]; Beckman and Ames, 1998; [[2], [3], [4]]. The hydroxyl radical (^•OH), formed in cells by ionizing radiation (IR), chemical toxins, and rogue mitochondrial electrons, is the most potent nucleic acid oxidizer generating a dose-dependent assortment of nucleic acid modifications and strand breaks [[5], [6], [7]]. The ^•OH modification of guanine at C8 (8-oxoG) is often considered the most frequent oxidative lesion in DNA due to the low redox potential of the nucleobase and is specifically targeted for removal by DNA repair glycosylases [6,8,9]. Nevertheless, oxidized and potentially mutagenic species of all purine and pyrimidine nucleobases have been observed in both DNA and RNA [[10], [11], [12], [13]]. Their frequency and mispairing potential in comparison to 8-oxoG, may depend on relative oxidation potentials influenced by the specific sequence, base-stacking, and mismatch contexts [[14], [15], [16], [17]].

Like its purine counterpart, adenine bases in nucleic acids readily oxidize at C8 (8-oxoA) though oxidation at C2 (2-oxoA) is also possible [6,8]. 8-oxoA predominantly exists as a keto tautomer distinguished from adenine by a C8 carbonyl and an N7 hydrogen [[18], [19], [20]]. Adenine is chemically considered the 2nd nucleobase most prone to oxidation with levels of 8-oxoA in irradiated cells or purified DNA 3–6 fold less than 8-oxoG [[21], [22], [23], [24], [25]]. However, in some instances 8-oxoA frequency in DNA may be near parity with 8-oxoG as observed in a rat and fish liver cancer and aging model [26]. 8-oxoA, along with the ring cleaved Fapy-A, an 8-oxoA oxidation product, accumulate in a variety of cancers and in the DNA extracted from cells exposed to IR, ischemia-reperfusion, and other sources of ^•OH or one-electron oxidant stress [1,11,20,22,[27], [28], [29], [30]]. This increase arises either by direct oxidation of DNA or plausibly vis-à-vis DNA polymerase misincorporation of 8-oxodATP. To prevent polymerase errors, nucleoside triphosphatases (NTPases) such as eukaryal MTH1 and bacterial MutT are tasked with cleansing cells of 8-oxodATP and other oxidized purine NTPs [[31], [32], [33], [34], [35]]. E. coli DNA polymerase I and Taq polymerase accurately incorporate T across from a template 8-oxoA [19,36,37]. However, mammalian translesion (TLS) polymerases (polη), gap-filling polymerases (polβ), replication initiators (polα), and viral reverse-transcriptases (AMV, MMLV) can misincorporate guanine nucleotides across a template 8-oxoA by geometrically favoring the syn conformation and resultant Hoogsteen base pairing [18,36,[38], [39], [40], [41]]. Sulfolobus DNA polymerase IV (Dpo4) also stabilizes an 8-oxoA:dGTP mispair in Watson-Crick configuration [42]. The mutagenic potential of 8-oxoA is driven by its structural propensity to form mispairs, especially when encountered by an error prone DNA polymerase [36,[41], [42], [43]], its influence on the excision rate of tandem abasic sites [44], and the tendency of C2 to form protein and DNA interstrand cross-links [45,46].

Cells harbor dedicated repair machinery to excise certain 8-oxoA adducts from DNA [6]. Base excision repair (BER) glycosylases, including eukaryotic 8-oxoguanine DNA glycosylase (OGG1) and uracil/thymine DNA glycosylase (TDG), can seek and remove 8-oxoA in both nuclear and mitochondrial genomes [39,[47], [48], [49]]. The 8-oxoA:C mispair is preferentially cleaved by OGG1 and its bacterial counterpart MutM (Fpg) [34,39]. In mammals, a backup 8-oxoA:C repair pathway initiated by NEIL1 endonuclease and polynucleotide kinase phosphatase (PNKP) reinforces the significance of repairing this mismatch [50]. However, 8-oxoA pairs with T or G are not OGG1 or NEIL1 substrates but rather are eliminated by TDG and its E. coli homologue, mismatch uracil glycosylase (MUG) [49]. The chromatin remodeling and strand exchange activity of Cockayne syndrome B protein (CSB) also helps prevent accumulation of 8-oxoA after IR treatment of primary fibroblasts [29,51]. Defective CSB interferes with a variety of nuclear and mitochondrial DNA repair pathways and is a cause of premature aging. What role exonucleases have in removing 8-oxoA is unclear, though the lesion impairs the 3′-5′ exonuclease of Werner Syndrome helicase and the Klenow fragment [52].

The 8-oxoA lesion also surfaces in RNA and its precursors [20]. Xanthine oxidase produces 8-oxoA as an intermediate in purine catabolism and irradiation of polyadenylic acid (polyA) generates 8-oxoA and other adenine adducts [53,54]. 8-oxoA was purified from Torula yeast RNA and may be important in primordial RNA redox catalysis [12,55]. In highly damaged regions of the late-stage Alzheimer's brain, mRNA accumulated 8-oxoA moieties at levels greater than 8-oxoG, hinting at a role for the polyA tail as a redox sink [14]. Unlike 8-oxoG, when 8-oxoA arises in the DNA template, elongation by the RNA polymerase II-transcription factor IIS complex is inhibited [56]. Indiscriminate 8-oxoATP incorporation by RNA or polyA polymerases plausibly drives the accrual of the adduct in cellular RNA, though this is currently unclear. When 8-oxoA is localized to coding regions, translation arrest produces truncated proteins and mRNA endonucleases are recruited to the ribosome to initiate no-go decay (NGD) [[57], [58], [59]]. Similarly, translation is attenuated when a catalytic adenosine (A2451) in 23 S ribosomal RNA is C8 oxidized [60]. Interestingly, 8-oxoA derivatives are potent site 1 agonists of Toll-like receptor 7 (TLR7), a central innate immune sensor [[61], [62], [63], [64], [65], [66], [67], [68]], raising the prospect that lysosomal processing of damaged RNA or its precursors activate the innate immune system.

DNA and RNA specific antibody complexes are diagnostics hallmark of autoimmune diseases such as systemic lupus erythematosus (SLE) and Sjögren syndrome [[69], [70]]. The advent of hybridoma technology in the 1970's spawned the ability to create monoclonal antibodies (mAbs) specific for the detection of DNA or RNA [[71], [72]]. The technology proved particularly useful for recognition of nucleobase modifications such as m6A, 5 mC, Ψ, and m7G [72]. These nucleobase modification mAbs have found diverse scientific applications in immunoaffinity chromatography, ELISA, nucleic acid westerns, and high-resolution immunohistochemistry (IHC). More recently, immunoprecipitation (IP) by nucleobase modification mAbs has been coupled with next generation DNA/RNA sequencing (DIP/RIP-seq) to locate, quantify, and characterize modifications across the genome and transcriptome [72]. While these technologies hold great promise for fundamental discovery and clinical applications, their utility is dependent on rigorous biochemical characterization of the mAb employed to ensure high on-target affinity and low cross reactivity.

Investigations and isolation of oxidatively damaged nucleic acids have similarly been aided by mAbs, such as the well characterized anti-8-oxoG and anti-thymidine glycol antibodies [13,73,74]. In the present study, we sought to expand the available toolkit for investigating nucleobase damage modifications to include the 8-oxoA lesion. We developed a mouse, IgGk1 monoclonal antibody against 8-oxoA via immunization with keyhole limpet hemocyanin (KLH) fused to periodate oxidized 8-oxoA (KLH-8-oxoA). A competitive ELISA assay is used to systematically determine relative affinities of the purified mAb 6E4 for 8-oxoA in comparison to a suite of native and damaged purine and pyrimidine nucleosides, 8-oxo-adenosine triphosphates (8-oxoATP), and DNA/RNA oligonucleotides harboring a single 8-oxoA. We demonstrate mAb 6E4 has exquisite specificity for 8-oxoA with no cross-reactivity to the closely related 8-oxoG or A. The nucleobase of 8-oxoA is the essential epitope and detectable with high affinity as a free nucleobase, ribonucleoside, 2′-deoxyribonucleoside, and nucleoside triphosphate. 6E4 detects 8-oxoA with equal proficiency in single and double strand DNA/RNA, though affinity is influenced by certain 8-oxoA base mispairs. Our results indicate the 6E4 monoclonal antibody may have broad applications in 8-oxoA immunodetection platforms.

2. Results

2.1. Design, construction, and selection of an anti-8-oxoA monoclonal antibody

The 8-oxoA antigen was constructed by fusing periodate oxidized 8-oxoA (8-oxoA-dialdehyde) to keyhole limpet hemocyanin (KLH-8-oxoA) and then immunizing three BALB/c female mice. Serum was screened for 8-oxoA affinity and specificity by a competitive ELISA assay whereby free antigens are used to disrupt antibody binding to an 8-oxoA acetylcholinesterase conjugate (AchE-8-oxoA) (Fig. 1). Binding to 8-oxoA is measured by monitoring hydrolysis of acetylthiocholine by the 8-oxoA fused AchE and monitoring absorbance at 414 nm after addition of Ellman's reagent.

Fig. 1.

Design of the 8-oxoA ELISA. Experimental overview of the competitive ELISA assay used to measure binding to 8-oxoA. Wells are coated with goat anti-mouse IgG Fc domain and BSA as a blocking agent. A mixture of free 8-oxoA, AChE conjugated to 8-oxoA, and the indicated antibody are added to the well, followed by washing with buffer to remove unbound material. Ellman's reagent and acetylthiocholine are added as substrate for bound AchE-8-oxoA. The emission of yellow light is then monitored at 414 nm on a BioTek Epoch microplate spectrophotometer and absorbance plotted as % B/Bo. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

Serum from the mouse exhibiting the strongest 8-oxoA specific titer was sacrificed and prepared for splenocyte fusion with the plasmacytoma line P3X63Ag8.653. Of the ∼90 stable hybridoma clones, we proceeded to characterize clone 6E4 due to its affinity for both 8-oxoA and a 21 mer RNA oligonucleotide harboring a central 8-oxoA (ssRNA-8-oxoA) and its low cross-reactivity with nucleosides A, G, and 8-oxoG (>1000-fold competitor concentration required relative to 8-oxoA). A mouse specific, lateral flow assay was used to confirm the isotope encoded in clone 6E4 is IgGk1. We also confirmed an IgGk1 isotype for a second hybridoma, clone 5G7, with robust specificity for 8-oxoA.

2.2. Purification and 8-oxoadenine specificity

To discern its nucleobase specificity, we purified mAb 6E4 by protein G chromatography from the supernatant of hybridoma 6E4 cultured in ultra-low IgG fetal bovine serum (Fig. 2A). We then titrated increasing mass of purified 6E4 and measured its ability to bind the 8-oxoA-AChE conjugate in a non-competitive ELISA assay (Fig. 2B). 6E4 exhibited a dose-dependent ability to capture 8-oxoA-AChE. All subsequent assays were then carried out using an antibody concentration extrapolated from the linear range of the titration curve.

Fig. 2.

Purification and Antibody Titration.A. SDS-PAGE of 5 μg each of non-reduced (NR) and reduced (R) purified mAb 6E4 is shown after staining by Coomassie Blue. A control, mouse IgGk1 mAb is displayed for comparison. The location and molecular weight (kDa) of protein standards are exhibited on the left. B. Titration of purified mAb 6E4 binding to AChE-8-oxoA in the absence of competitor. The absorbance (A) at 414 nm is plotted as a function of the mass (ng) of purified mAb 6E4. The binding assay was performed as described in Materials and Methods.

Purified 6E4 was then assayed by competitive ELISA on a series of 8-oxoA derivatives to determine the requirement of the 8-oxo moiety for antibody binding and how nucleobase specificity is affected by the presence of a ribose, 2′-deoxyribose, or 5′-triphosphate (Fig. 3). The structure of 8-oxoA is depicted in Fig. 3A, highlighting the C8-oxo, the additional H-atom at N7, and the N-glycosidic bond to ribose or 2′-deoxyribose. A complete list of all competitor structures used in the study is available in Supplementary Table S1. The ability of increasing concentrations of competitive 8-oxoA variants (nM) to dislodge 6E4 from 8-oxoA-AChE is depicted in Fig. 3. The relative affinities, as gauged by IC₅₀ (Table 1), indicate a preference for 8-oxoadenine (8-oxoAde), the free nucleobase (IC₅₀ = 20 nM) (8-oxoAde>8-oxodA>8-oxodATP>8-oxoA>8-oxoATP > A ≈ Fapy-Ade). The presence of a ribose 2′-OH is slightly repulsive (2.4-fold decrease in IC₅₀ of 8-oxodA vs. 8-oxoA) and a 5′-triphosphate is well tolerated when fused to 8-oxoA or 8-oxodA nucleosides (IC₅₀ = 91 nM and 29 nM, respectively). The 8-oxoAde oxidative cleavage product, Fapy-Ade, is an inadequate competitor.

Fig. 3.

Antibody Binding Specificity for 8-oxoA Variants. A. The structure of 8-oxoA. The O8 and H7 atoms that distinguish 8-oxoA from A are shaded pink and grey. The green box labeled R demarcates the attached ribose or 2′-deoxyribose. B. Specificity for 8-oxoA determined by an AChE-8-oxoA competitive ELISA. The binding assay was performed as described in Materials and Methods. The % B/Bo is plotted as a function of the concentration (nM) of the indicated competitor: 8-oxoadenosine (8-oxoA); 8-oxoadenosine 5′-triphosphate (8-oxoATP); 8-oxo-2′-deoxyadenosine (8-oxodA); 8-oxo-2′-deoxyadenosine 5′-triphosphate (8-oxodATP); 8-oxoadenine (8-oxoAde); Fapy-adenine (Fapy-Ade); adenosine (A). Each datum is the average of at least 2 independent experiments±SEM. % B/Bo is the ratio of A₄₁₄ nm for a specific well divided by the maximum absorbance, Bo, of a competitor-free well. 100% B/Bo indicates no inhibition by the indicated competitor. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

Table 1.

Summary of IC₅₀ and Cross Reactivities for mAb 6E4.

Competitor	IC50 (nM)	Fold Changea
8-oxoAde	20	0.3
8-oxodA	27	0.4
8-oxodATP	29	0.5
8-oxoA	65	1.0
8-oxoATP	91	1.4
dsRNA-8-oxoA:C	632	10
dsRNA-8-oxoA:A	671	10
ssRNA-8-oxoA	692	11
ssDNA-8-oxoA	774	12
dsRNA-8-oxoA:U	861	13
dsDNA-8-oxodA:dA	875	13
dsDNA-8-oxodA:dG	1151	18
dsDNA-8-oxodA:dC	1206	18
dsDNA-8-oxodA:dT	1298	20
dsRNA-8-oxoA:G	4597	70
ssDNA	ncb	/
dsDNA	nc	/
8-aminoA	nc	/
2-oxoA	nc	/
ssRNA	nc	/
dsRNA	nc	/
8-oxoG	nc	/
FapyAde	nc	/
C	nc	/
5-hoU	nc	/
G	nc	/
U	nc	/
A	nc	/
I	nc	/
m1A	nc	/
m6A	nc	/
5-hoC	nc	/

^a(IC₅₀ Competitor)/(IC₅₀ 8oxoA).

^bNot calculated.

2.3. Pyrimidines, purines, and methylations

Selectivity of an 8-oxoA mAb must be robust in the presence of other pyrimidine and purine nucleosides to be effective in complex mixtures of nucleic acids that include precursors, degradation products, and polynucleotide chains. Furthermore, oxidized variants of the other nucleosides can also be elevated by oxidative stress and display exposed oxygen atoms that could serve as 6E4 epitopes. We thus expanded our investigation of the binding properties of mAb 6E4 to the native pyrimidine ribonucleosides, uridine (U) and cytidine (C), along with their most common oxidation products, 5-hydroxyuridine (5-hoU) and 5-hydroxycytidine (5-hoC), both of which occur as native modifications in cells and are enriched after oxidative stress [[10], [11], [12], [13]] (Fig. 4A). The native pyrimidine nucleosides U and C were unable to compete off mAb 6E4 bound to AChE-8-oxoA. Similarly, the pyrimidine C5 oxidation products, 5-hoU and 5-hoC, were also poor competitors to the AChE-8-oxoA antigen. These findings indicate neither native nor oxidized pyrimidine nucleosides are targeted by mAb 6E4, and that the antibody specifically detects 8-oxoA in the context of native or commonly oxidized nucleosides.

Fig. 4.

Pyrimidines, Purines, and Methylations. A. The % B/Bo is plotted as a function of the concentration (nM) of free 8-oxoA or the indicated competing pyrimidine nucleoside: Uridine (U); 5-hydroxyuridine (5-hoU); cytidine (C); 5-hydroxycytidine (5-hoC). B. The % B/Bo is plotted as a function of the concentration (nM) of the indicated competing purine nucleoside: Adenosine (A); 8-oxoguanosine (8-oxoG); guanosine (G); 2-oxoadenosine (2-oxoA); inosine (I). 8-oxoA is used as a control. Each datum is the average of at least 2 independent experiments±SEM. C. The % B/Bo is plotted as a function of the concentration (nM) of the indicated competing modified adenosine: N1-methyladenosine (m1A); N6-methyladenosine (m6A); and 8-aminoadenosine (8-aminoA). 8-oxoA is used as a control. Each datum is the average of at least 2 independent experiments±SEM.

We next surveyed native purine nucleosides adenosine (A) and guanosine (G) along with important oxidized, purine variants: 2-oxoadenosine (2-oxoA), a cytotoxic, oxidation product of adenosine [75] and 8-oxoguanosine (8-oxoG), the most ubiquitous oxidized base which like 8-oxoA harbors a protonated N7 and a C8 oxo moiety [9]. Inosine (I), an N6 deaminated adenosine with essential biological roles in both DNA and RNA, was also studied. The nucleosides A, G, and I were not competitors for mAb 6E4, even as concentrations were increased to 100 μM, indicating N6/O6 variants or an exocyclic amine at C2 do not induce cross-reactivity. 8-oxoG, a purine nucleoside harboring the same oxygen atom at C8 as 8-oxoA, was similarly ineffective in displacing mAb 6E4 from AChE-8-oxoA. Finally, 2-oxoA (isoguanosine) was also a poor competitor out to 1 μM (B/Bo = 100%), with modest antibody binding at doses >10 μM (IC₅₀ = 73 μM). These outcomes further substantiate the 8-oxoA specificity of mAb 6E4.

We proceeded to measure cross-reactivity with three distinct, modified adenosine analogues. Binding by 6E4 to either N1-methyladenosine (m1A), a modification originally discovered in non-coding RNA and important for RNA structure and stability [76], or N6-methyladenosine (m6A), a pervasive, dynamic modification in eukaryotic mRNA with roles in splicing, stability, and translation [77,78] was undetectable by competitive ELISA (Fig. 4C). We then examined 8-aminoadenosine (8-aminoA), a chemotherapeutic agent that inhibits multiple steps in RNA synthesis and processing including polyadenylation [79]. Unlike m1A and m6A, 8-aminoA is structurally comparable to 8-oxoA in that it harbors a C8 modification (though one with H-bond donors). Nevertheless, mAb 6E4 can effectively discriminate 8-aminoA from 8-oxoA (>1000-fold increase in competitor required to reach 50% B/Bo), though 8-aminoA is a superior antigen to the described methylated adenosines.

2.4. Recognition of 8-oxodA in single and double-strand DNA

The most prevalent adenine lesion in DNA damaged by ionizing radiation is 8-oxodA [80]. A single 8-oxodA in single-strand DNA alters adjacent base-stacking and phosphodiester conformation [81]. When correctly paired with dT, 8-oxodA does not alter the structural properties of the DNA duplex by adopting the anti-orientation [19]. The structures of an 8-oxodA:G base pair in dsDNA shows a syn:anti Hoogsteen arrangement that involves H-bonds between the 8-oxo and N7–H atoms with the N1 and O6 of guanine, respectively [82]. To determine the ability of mAb 6E4 to identify 8-oxodA embedded in single and correctly paired double-stranded DNA, we synthesized a 21 nt DNA oligonucleotide of random sequence containing a central 8-oxodA residue at position 11 and then annealed the ssDNA-8-oxodA to its complementary anti-sense (AS) strand to create a dsDNA with a central 8-oxodA (Fig. 5). Versions of the ssDNA and dsDNA antigens in which position 11 is an unmodified dA were used as controls. mAb 6E4 effectively binds both ssDNA-8-oxodA and dsDNA-8-oxodA, though the affinity is inferior to the nucleoside 8-oxoA (Fig. 5). Recognition of 8-oxoA is efficient in the single and double-strand DNA (IC₅₀ = 774 nM and 1.3 μM, respectively). The corresponding unmodified ssDNA and dsDNA oligonucleotides are both poor antigens, with a 6.2-fold and 4.3-fold decrement, respectively, at 6.25 μM, though modest competition occurs at higher concentrations. These findings highlight that mAb 6E4 is able to detect a site-specific, individual 8-oxodA in single and double-strand DNA.

Fig. 5.

Detection of 8-oxodA in Single and Double Strand DNA. The structure of the 21 nt 8-oxodA DNA oligonucleotide duplex is shown at the top. The 8-oxodA base (blue) is located at position 11 in the sense strand (ssDNA-8-oxodA) and pairs with T in the antisense strand (dsDNA-8-oxodA). The ssDNA and dsDNA contain an unmodified A at position 21 of the sense strand. The ELISA binding assay was performed as described in Materials and Methods. The % B/Bo is plotted as a function of the concentration (nM) of the indicated competing DNA oligonucleotide. 8-oxoA is used as a control. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

2.5. Effect of 8-oxodA base mispairs embedded in dsDNA

We then created 4 distinct duplex DNA oligonucleotides in which we varied the base (X) across from 8-oxodA to better understand the properties of 6E4 when encountering potential mutagenic 8-oxodA mispairs (Fig. 6). The results of the competitive ELISA shown in Fig. 6 show clear dose-dependent binding to 8-oxodA in all DNA oligonucleotides tested. Varying the base (T, C, A, G) that pairs with 8-oxodA only modestly alters 6E4's ability to detect the lesion in dsDNA, with comparable IC₅₀ magnitudes in all 4 duplex, 8-oxodA-DNA antigens. Additionally, 8-oxodA is distinguished with similar facility in single and double-strand DNA contexts. Amongst the mispairs, 6E4 marginally prefers 8-oxodA pairing in the order A > G > C > T. Interestingly, 6E4's affinity for the 8-oxodA:dA pair (IC₅₀ = 875 nM) is nearly equivalent to that of ssDNA-8-oxodA (IC₅₀ = 775 nM) and moderately superior to the other 3 mispairs. We conclude recognition of 8-oxodA in DNA is largely driven by nucleobase specificity, a property in duplex DNA shaped by the identity, and perhaps architecture, of the 8-oxodA base pair.

Fig. 6.

Detection of 8-oxodA in Mismatched DNA Oligonucleotides. The structure of the 21 nt 8-oxodA DNA oligonucleotide duplex is shown at the top. The 8-oxodA base (blue) at position 11 in the sense strand, is variably paired with X in the antisense strand. The ssDNA-8-oxodA is the unpaired sense strand. The ELISA binding assay was performed as described in Materials and Methods. The % B/Bo is plotted as a function of the concentration (nM) of the indicated competing DNA oligonucleotide. 8-oxoA is used as a control. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

2.6. Detection of 8-oxoA in single and double-strand RNA

We proceeded to interrogate the aptitude of 6E4 for the recognition of a single 8-oxoA inserted in RNA (Fig. 7). A ssRNA version of the ssDNA-8-oxodA oligonucleotide described above was synthesized and used as an antigen (ssRNA-8-oxoA) in the competitive ELISA assay. Similarly, the ssRNA-8-oxoA oligonucleotide was hybridized to its RNA complement to form dsRNA-8-oxoA (Fig. 7). Control ssRNA and dsRNA competitors in which an unmodified A replaces 8-oxoA at position 11 were also examined. The ssRNA-8-oxoA antigen is well-recognized by mAb 6E4 (IC₅₀ = 692 nM), exhibiting nM affinity though reduced affinity compared to the 8-oxoA nucleoside (an 11-fold decrement). The unmodified ssRNA counterpart is not effectively bound by mAb 6E4, with a >300-fold concentration necessary relative to ssRNA-8-oxoA to initiate displacement from AchE-8-oxoA. The binding behavior of 6E4 to dsRNA-8-oxoA antigens was reasonably robust. The dsRNA-8-oxoA:U pair is well detected by 6E4 (IC₅₀ = 861 nM), though not with the same affinity as ssRNA-8-oxoA (1.2-fold decrement). The dsRNA control lacking 8-oxoA is feebly recognized, exhibiting a competitor titration curve resembling the unmodified ssRNA. Consistent with its capacity for 8-oxoA detection in nucleosides and DNA oligonucleotides, mAb 6E4 adeptly senses 8-oxoA damage in RNA.

Fig. 7.

Detection of 8-oxoA in Single and Double Strand RNA. The structure of the 21 nt 8-oxoA (blue) RNA oligonucleotide duplex is shown at the top (dsRNA-8-oxoA). The 8-oxoA base is located at position 11 in the sense strand and pairs with U in the antisense strand. The ssRNA-8-oxoA is the unpaired sense strand. The ssRNA and dsRNA contain an unmodified A at position 21 of the sense strand. The ELISA binding assay was performed as described in Materials and Methods. The % B/Bo is plotted as a function of the concentration (nM) of the indicated competing RNA oligonucleotide. 8-oxoA is used as control. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

2.7. Effect of 8-oxodA base mispairs embedded in dsRNA

Double-strand RNAs were also generated in which 8-oxoA is individually paired with each one of the four standard RNA bases (Fig. 8A). Single-strand RNA and double-strand RNA in which 8-oxoA is paired with U, C, or A are all roughly equivalent antigens for mAb 6E4 with IC₅₀ values of 861, 632, and 671 nM, respectively. However, affinity for dsRNA with a central 8-oxoA:G mispair is poor, with 4.6 μM of the competitor required for 50% inhibition—a 5.3-fold decrement in affinity relative to the dsRNA-8-oxoA:U pair and a 6.6-fold decrement relative to ssRNA-8-oxoA. The structure of a syn 8-oxoA:anti G base pair with 2 H-bonds is shown in Fig. 8B. The 2 atomic additions of 8-oxoA—a hydrogen atom at N7 (shaded grey) and an O atom at C8 (shaded pink)—are fundamental to its pairing scheme with guanine. The protonated N7 of 8-oxoA forms H-bonds with the guanine O6, while the O8 adduct Hoogsteen pairs with the guanine N1 hydrogen. These findings indicate the 8-oxoA:G pairs in the dsRNA tested here, may render key epitopes, especially the 8-oxo and N7–H moieties, inaccessible to mAb 6E4.

Fig. 8.

Detection of 8-oxoA in Mismatched RNA Oligonucleotides. A. The structure of the 21 nt 8-oxoA RNA oligonucleotide duplex is shown at the top. The 8-oxoA base (blue) at position 11 in the sense strand, is variably paired with X in the antisense strand. The ssRNA-8-oxoA is the unpaired sense strand. The ELISA binding assay was performed as described in Materials and Methods. The % B/Bo is plotted as a function of the concentration (nM) of the indicated competing RNA oligonucleotide. 8-oxoA is used as control. B. Image of an 8-oxoA:G mispair. 8-oxoA is shown in the syn conformation forming two H-bonds with G in the anti-conformation. The 2 unique atoms that define 8-oxoA are highlighted pink (8-oxo) and grey (N7 hydrogen). (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

3. Discussion

Defects in macromolecular quality control due to accumulated oxidative damage drive aberrant cell death and the spilling of cell debris, including nucleic acids and their precursors, to the extracellular space [83,84]. Detection of this debris by self-reactive B-cells leads to the production of nucleic acid autoantibodies, diagnostic trademarks of the autoimmune disease lupus. When internalized by Fc receptors, these autoantibodies potently stimulate type I interferon (IFN) synthesis via nucleic acid presentation to endosomal TLRs [[69], [70], [85], [86], [87], [88]]. Elevated levels of 8-oxodG and 8-oxodA are present in circulating anti-DNA immune complexes isolated from SLE patients and lesions to all 4 nucleobases have been observed in provoked SLE neutrophils [27,[89], [90], [91], [92]]. Oxidized mitochondrial DNA, discharged from neutrophils, is a formidable catalyst of the described type I IFN synthesis characteristic of SLE [93,94]. Oxidized mitochondrial RNA has been linked with aging and cell death, though existence of extracellular RNA oxidation is largely virgin territory [[95], [96], [97], [98], [99]]. Intriguingly, 8-oxoG RNA antibodies have been extracted from a lupus (antiphospholipid) mouse model [100]. Whether 8-oxoA accumulates in circulating DNA or RNA and is specifically targeted by antibodies is worthy of investigation given the 8-oxoA nucleobase potency as a TLR7 agonist and a marker of oxidative damage. New analytical tools will be valuable in answering these questions.

Here we reported the design, construction, purification, and characterization of a novel mouse IgGk1 monoclonal antibody targeting the damaged nucleobase, 8-oxoA. We demonstrated that the specificity and nanomolar affinity of mAb 6E4 for 8-oxoA is driven by the 8-oxoAde nucleobase moiety and fundamentally requires C8 oxidation of adenine. The presence of a ribose or 2′-deoxyribose only modestly decreases affinity for 8-oxoA, and 5′-triphosphates are well tolerated. 6E4 discriminates against both native and oxidized variants of a suite of common pyrimidine and purine nucleosides. 8-oxoG, considered the most frequent nucleobase oxidation product and modified, like 8-oxoA, with an N7 hydrogen and C8 oxygen, is poorly recognized by 6E4. Similarly, Fapy-Ade, an 8-oxoAde degradation product is an inadequate competitor.

Given our interest in technologies to detect damaged RNA, we initially selected mAb 6E4 based on specificity for the 8-oxoA nucleoside and a ssRNA harboring a single 8-oxoA. Another 8-oxoA specific clone (3E9) was not biochemically pursued given its poor affinity for ssRNA-8-oxoA. Consistent with our initial hybridoma selection criterion, we demonstrate mAb 6E4 is proficient in binding ssRNA-8-oxoA with high specificity relative to its unmodified ssRNA counterpart and a slightly better affinity than its ssDNA-8-oxodA analogue. Since RNA often occurs as a duplex, we examined in dsRNA the detection of 8-oxoA when faithfully paired and as mutagenic mispairs. 6E4 is adept at binding 8-oxoA in duplex RNA when the lesion is paired with C, A, or U, with only a modest decrement relative to ssRNA-8-oxoA. However, antibody binding to 8-oxoA:G pair in dsRNA is severely impaired. These findings may suggest the 8-oxoA nucleobase stabilizes the anti-conformation when paired with C, A, or U in the dsRNA sequence utilized here. Alternatively, the 6E4 lesion recognition complex more efficiently senses the 8-oxoA embedded in these mispairs. In contrast, the 8-oxoA:G pair may favor syn 8-oxoA and anti G conformations with Hoogsteen pairing in the employed RNA. Distinct from the 8-oxoG:A pair, the C8 oxygen atom of 8-oxoA forms a H-bond with the N1 of guanine in the 8-oxoA:G pair. The relative preference of 6E4 for 8-oxoA mispairs (C > A > U»>G) raise the prospect the C8-oxo epitope is presented on the surface of the recognized dsRNA mispairs or easily dislodged by 6E4, but is buried when Hoogsteen paired with G. These observations are consistent with the proposal that the promiscuity of 8-oxoA pairing in dsRNA influences the degree of A-form helix distortion by a particular mispair [15,20,101]. These findings highlight the utility of 6E4 in detecting 8-oxoA in single and double strand RNA with the exception of 8-oxoA:G pairs. 6E4 may thus be particularly promising for the detection of single-strand RNA regions that may be enriched with 8-oxoA, such as cytoplasmic stress granules and P-bodies sequestering polyadenylated mRNA [102], as long as background 8-oxoA nucleosides are removed.

With respect to single and double-strand DNA, 6E4 is also effective in recognizing the 8-oxodA lesion. However, overall affinity for 8-oxodA-DNA is comparatively less than its RNA analogues with the noted exception of the 8-oxoA:G mispair. This pattern is altered from that observed with the surveyed nucleosides whereby 8-oxoAde>8-oxodA>8-oxoA, suggesting that additional 2′-OH contacts outside of the 8-oxoA binding site may drive specificity for ssRNA-8-oxoA over ssDNA-8-oxodA. In the case of duplex antigens, the dimensions of B-form DNA may obscure some portion of the 8-oxodA nucleobase in comparison to A-form RNA or more broadly influence 6E4 binding. The affinity for 8-oxodA varies with differing degrees amongst specific mispairs, with the 8-oxodA:dA pair being the most preferred DNA antigen tested. Interestingly, the facility with which hTDG excises 8-oxodA paired with dA or dT may be dependent on assay and annealing conditions, despite comparable, robust repair activity when the lesion is paired with dC or dG [49,103]. The 8-oxodA:dG mispair may not exclusively exist as a syn:anti Hoogsteen pair in the dsDNA sequence annealed here as the antigen is productively bound, suggesting the C8-oxo and N7–H are efficiently sensed by 6E4. Structures of the error prone mammalian polymerase polβ illustrate how key protein residues (K280, R283) are required to stabilize a template 8-oxodA syn conformation to pair with the incoming anti dGTP [40]. The Hoogsteen base pair is induced by polβ and utilizes the C8-oxo in one of its 3 H-bonds yielding Watson-Crick geometry comparable to a dC:dGTP pair. This is different from the 8-oxodA:dG Hoogsteen pair previously observed in dsDNA that adopts wobble geometry [82], depicted in Fig. 8A with 2 H-bonds that utilize the 8-oxo and N7–H moieties of 8-oxoA. Our findings are consistent with recent observations that some properties of 8-oxoA pairing in dsRNA are distinct from those in dsDNA [15]. The ability of 6E4 to detect 8-oxodA:dG pairs in DNA will likely be dependent on how the pair was created—e.g., by a TLS polymerase or direct ^•OH radical attack—and its resultant architecture.

The high affinity and selectivity of mAb 6E4 for 8-oxoATP and 8-oxodATP was an unexpected finding of our antigen binding survey. Intracellular ATP, produced by anaerobic glycolysis in the cytoplasm and aerobic respiration in the mitochondria, is present in millimolar concentrations and is the most abundant ribonucleoside triphosphate [[104], [105], [106]]. Given its modus operandi in aerobic ATP production and its propensity to form oxygen radicals [7], the mitochondria may be a particularly important fabricator of 8-oxoATP. The broad specificity of human MTH1 for oxidized purine nucleotides and the existence of a mitochondrial isoform reflect, in part, the importance of sanitizing oxidized ATP/dATP variants [20,[31], [32], [33],107,108]. Nevertheless, oxidative injuries still accumulate in mitochondrial nucleic acids and are well documented contributors to aging, cancer, autoimmunity, and neurodegeneration [94]. The relatively poor efficiency of hMTH1 on 8-oxoATP in comparison to other oxidized, purine NTPs may play an important role [108]. Thus, the selective, regulated destruction of oxidatively damaged mitochondria (mitophagy) is critical to cell survival and disease prevention [83,93].

Concentrations of extracellular ATP (eATP) in healthy tissues are carefully regulated by cell surface nucleotidases [106]. However, in the tumor microenvironment (TME), eATP released by necrosis, apoptosis, exosomes, or stress-activated membrane pores can accumulate in the micromolar range and promote immune-stimulated tumor inhibition [[109], [110], [111], [112], [113]]. Antibodies co-targeting eATP and tumor receptors have recently been proposed as a TME-selective targeting mechanism [[112], [114]]. The fraction of eATP in the TME that exists as 8-oxoATP is unknown, as are the consequences of extracellular 8-oxoATP on immune signaling [115]. Periodate oxidized eATP, where ribose 2′,3′ diols are converted into reactive dialdehydes, is a covalent inhibitor of ATP-gated P2 receptors and impairs ATP ectonucleotidases [116,117]. The net antagonistic effect is a diminished immune response. Given the stress-induced conditions of ATP release, the propensity of the mitochondria to generate oxygen radicals, and that free nucleotides are more prone to oxidation than nucleic acids [118], extracellular 8-oxoATP and its cleavage products may be significant species in the TME. The biochemical specificities of mAb 6E4 make it a promising tool to target and interrogate the major ribo- and 2′-deoxyribonucleoside species displaying the 8-oxoA lesion.

The 8-oxoA modification has previously been identified by a variety of methods. High performance liquid chromatography coupled with electrochemical detection (HPLC-ECD) and gas chromatography-mass spectrometry with single-ion monitoring (GC-MS) [22,91,119,120] have each been employed for oxidative lesion detection including 8-oxoA. HPLC coupled with tandem mass-spectrometry (LC-MS/MS) after nuclease and phosphatase digestion overcame specific limitations of the previous assays by both using soft ionization to minimize base decay and allowing simultaneous monitoring of multiple modifications [24]. While the LC-MS/MS method offers high sensitivity and specificity for the nucleoside modification, it is dependent on nucleic acid degradation. 8-oxoA containing synthetic RNA oligonucleotides have been characterized by MALDI-TOF MS [15,101] and gel electrophoretic methods have been utilized to detect the lesion in longer nucleic acids after Ir (IV) oxidation [121]. Antibody based approaches have also been employed for 8-oxoA detection in nucleic acids. A polyclonal radioimmunoassay for 8-oxoA was developed [122] and later rabbit polyclonal antibodies coupled with HPLC-ECD was used to detect and quantify 8-oxoA (IC₅₀ = 8 μM) in gamma irradiated DNA [21]. Mouse monoclonal antibodies to 8-oxoA were described in the 2005 patent of Holmes and Greene (US6900291B2) [123]. Two of the mAb clones (8A6 and 8A9) discriminate 8-oxoAMP from AMP and Fapy-Ade, though quantitative affinities are unclear given that culture supernatants rather than purified, titrated mAbs were assayed. Both 8A6 and 8A9 supernatants identified 8-oxodA in denatured DNA by Southwestern blot. The distinguishing feature of the 6E4 8-oxoA antibody described here, is its extensive characterization on a suite of competing nucleosides and defined, DNA and RNA oligonucleotides. These properties make 6E4 attractive for use in 8-oxoA immunodetection methods including competitive ELISA, immunoblotting, and DNA/RNA immunoprecipitations (DIP/RIP), with the understanding that its affinity for the 8-oxoAde nucleobase will present a comprehensive 8-oxoA status in a given biological sample. Thus, an 8-oxoA ELISA employing 6E4 would be expected to generate higher 8-oxoA values than LC-MS/MS for a defined 8-oxoA nucleoside. The specificity and nanomolar affinity for ssRNA-8-oxoA render mAb 6E4 particularly interesting for RIP technologies, especially as an enrichment step after RNA purification. 8-oxoG-RIP-seq was recently developed by coupling an 8-oxoG monoclonal antibody with next generation sequencing [124]. Utilizing 6E4 as the IP agent should render the method adaptable to 8-oxoA-RIP-seq, as long as sufficient antibody concentration is used to account for any reduced 8-oxoA affinity on longer RNA. Nanopore direct mRNA sequencing of 6E4 purified 8-oxoA transcripts, coupled with LC-MS/MS controls, may be an elegant approach to 8-oxoA-RIP-seq to reduce the risk of error prone cDNA construction and specifically locate coding region damage [38,125,126].

4. Conclusion

In this study, we described the design, discovery, and biochemical examination of a novel, mouse monoclonal antibody (6E4) targeting 8-oxoA, a C8-oxidized adenine lesion arising in DNA, RNA, and free nucleosides. 6E4 exhibits exquisite selectivity for 8-oxoA over a chemically diverse mixture of native, methylated, and oxidized nucleoside antigens including 8-oxoG, 2-oxoA, m6A, and 5-hoU. We further demonstrate the aptitude of 6E4 to specifically sense 8-oxoA in nucleoside triphosphates (8-oxoATP) and DNA/RNA oligonucleotides containing a single 8-oxoA. Importantly, 6E4 discerns 8-oxoA in double-strand DNA/RNA antigens where the lesion is either paired correctly or base mismatched. Our investigation highlights the 8-oxoAde nucleobase as the essential epitope and signal mAb 6E4 as well-suited for a wide span of immunodetection applications in nucleic acid quality control and sequencing.

Funding

This work was funded by the Cymba X Explorations program.

Methods

Construction of the antigen and tracer

A ∼10 mg/mL solution of 8-oxoadenosine (8-oxoA) (TCI America) in ddH₂0 was treated with 2 equivalents of sodium periodate (NaIO₄) overnight at room temperature (RT) and protected from light to convert the ribose 2′,3′-vicinal diol into a dialdehyde. The reaction was quenched by addition of 50 μl of 100% ethylene glycol.

4 mg of 8-oxoA dialdehyde intermediate was mixed with 4 mg of keyhole limpet hemocyanin (KLH) (Thermo-Fisher) in a total volume of 838 μl followed by the addition of 400 μl of 100 mM sodium borate pH 8.5. The reaction mix was incubated for 2 h at RT and then 4 mg of sodium cyanoborohydride (NaBH₃CN) was added and further incubated overnight at RT and protected from light. The 8-oxoA-KLH conjugation reaction was dialyzed twice into 4 L of phosphate buffered saline (PBS) to remove unreacted 8-oxoA and reaction components. Protein concentration in the dialyzed 8-oxoA-KLH was quantified by the bicinchoninic acid assay (Pierce™ BCA assay; Thermo-Fisher). Presence of 8-oxoA in both antigen and tracer was monitored by a wavelength scan with λ_max of 271 nm. 8-oxoA-KLH was frozen and stored at −20 °C until further use. The activated 8-oxoA dialdehyde (1 mg) was similarly fused to 500 U of E. electricus acetylcholinesterase (AChE) (Cayman Chemical) vis-à-vis incubation in 25 mM borate buffer pH 9.5 for ∼2 h at room temperature followed by treatment with NaBH₃CN as described.

Immunization

3 BALB/c female mice (Charles River) were immunized intraperitoneally (i.p.) when 28 days old with 50 μg of KLH-8-oxoA solubilized in equimolar Freund's complete adjuvant in PBS, followed by i. p. Boosters of 50 μg KLH-8-oxoA resuspended in Freund's incomplete adjuvant 14 and 35 days later. Mice were i. p. Injected ∼21 days later with 100 μl of KLH-8-oxoA without adjuvant. Serum was harvested from the submandibular vein ∼24 and 44 days after initial immunization for titer determination by competitive 8-oxoA ELISA. Serum from the mouse with the highest 8-oxoA specific titer was selected for fusion.

Hybridoma production

Mice were euthanized by CO₂ inhalation followed by cervical dislocation. Spleens were harvested and prepared for fusion. Hybridoma's were generated by electrofusion of DMEM washed splenocytes with the P3X63Ag8.653 (ATCC® CRL-1580™) plasmacytoma line in the presence of polyethylene glycol 1450 at 37 °C and 5% CO₂. Fusions were cultured in hypoxanthine-aminopterin-thymidine (HAT) medium. Culture media from HAT-resistant hybridomas was assayed for 8-oxoA specificity by competitive ELISA. 8-oxoA specific parental hybridomas were then cloned by limiting dilution at 1 cell per well. Clones were cultured in RPMI 1640, 2 mM glutamine, 10% fetal bovine serum (FBS), and prepared by centrifugation at 1000 rpm for 10 min at 4 °C. Cell pellets were resuspended at ∼10⁷ cells/ml in RPMI 1640 with 90% fetal bovine serum (FBS) and 10% DMSO. Aliquots of 1 mL were distributed into cryovials, stored overnight at −80 °C, and transferred into liquid N₂.

Total RNA was then purified from clone 6E4 and 5G7 hybridoma cell pellets (Zymogen Quick-RNA) and reverse transcribed using a modified MMLV-RT (SMARTScribe™) primed by oligo dT and a universal template switching oligonucleotide (TSO) to incorporate 5′-terminal transcript sequences. Rapid amplification of cDNA ends (RACE) was performed next to amplify the V_H and V_L domains using a high-fidelity DNA polymerase, the universal TSO as forward primer and an IgG1k constant domain as reverse primer. Products were inserted into the pCR-Blunt-TOPO and transformed into E. coli TOP10. Colony PCR was performed to accurately size select clones. At least 5 colonies harboring accurate V_H and V_L insert lengths were Sanger sequenced. A consensus sequence for the V_H and V_L chains of each hybridoma 6E4 and 5G7 was next determined from nucleotide multiple sequence alignments.

Antibody purification, characterization, and isotyping

The anti-8-oxoA hybridoma clone 6E4 was sequentially adapted to an ultra-low IgG FBS media (Avantor Seradigm). Hybridoma cell culture supernatant was subsequently applied to a protein G chromatography column in the presence of Gentle Ag/Ab Binding Buffer (pH 8.0) (Thermo Scientific). Bound antibodies were eluted with Thermo Gentle Ag/Ab Elution Buffer (pH 6.6) dialyzed first into Tris-buffered saline (TBS) followed by PBS. Fractions were subsequently either stored at −20 °C or mixed 1:1 with ELISA buffer and stored at 4 °C. The purified antibody isotype was determined with the mouse specific Pierce™ Rapid Antibody Isotyping Kit (Thermo Scientific).

Oligonucleotides and nucleoside competitors

Unmodified RNA oligonucleotides and all DNA oligonucleotides were chemically synthesized at Integrated DNA Technologies (IDT). The 8-oxoA RNA oligonucleotide was custom synthesized by ChemGenes. Nucleosides and nucleobases competitors were purchased from Carbosynth-Biosynth, Cayman Chemical, Medchem, LGC Biosearch, and Sigma. 8-oxoATP and 8-oxodATP were purchased from TriLink. Duplexes were annealed by mixing S and AS strands at a 1:1.2 M ratio in 10 mM Tris–HCl (pH 7.5), 50 mM NaCl, and 1 mM EDTA in sterile, nuclease free ddH₂O and heating to 95 °C. The annealing reactions were then slowly cooled to room temperature (∼22 °C) over 45 min followed by centrifugation at ∼10,000 rpm.

Competitive ELISA

A goat polyclonal antibody targeting the Fc domain of mouse IgG was coated on a 96 well surface, blocked in the presence of bovine serum albumin (BSA). Binding assays (150 μl) were carried out in ELISA buffer (100 mM phosphate, 400 mM NaCl, 1 mM EDTA, 0.1% BSA, 0.01% sodium azide) containing AChE-8-oxoA conjugate (3.75 mU), competitive antigen at the indicated concentration, and either dilutions of immunized serum or protein G purified mAb 6E4 (5 ng). Binding was initiated by addition of the immunized serum or mAb and then incubated for 2 h at RT. Antibody titration experiments against AChE-8-oxoA were performed in the absence of competitor.

To visualize binding, Ellman's reagent (5,5′-dithio-bis-2 nitrobenzoic acid) and the AChE substrate acetylthiocholine were added and then plates were developed in the dark for at least 2 h (Bo > 1.0). The emission of yellow light was monitored at 414 nm on a BioTek Epoch microplate spectrophotometer and absorbance plotted as % bound (B/Bo) versus log competitor concentration using a 4-parameter logistic fit. IC₅₀ values were calculated from the 4-paramater logistic fit. Plots and renderings displayed in figures were created in Prism 9 (GraphPad).

Appendix A. Supplementary data

The following is the Supplementary data to this article.