A Mass Spectrometry-Based Method to Measure Aflatoxin B₁ DNA Adducts in Formalin-Fixed-Paraffin-Embedded Tissues

Abstract

Aflatoxin B₁ (AFB₁) is a potent human liver carcinogen produced by certain molds, particularly Aspergillus flavus and Aspergillus parasiticus, which contaminate peanuts, corn, rice, cottonseed, and ground and tree nuts, principally in warm and humid climates. AFB₁ undergoes bioactivation in the liver to produce AFB₁-exo-8,9-epoxide, which forms the covalently bound cationic AFB₁-N7-guanine (AFB₁-N7-Gua) DNA adduct. This adduct is unstable and undergoes base-catalyzed opening of the guanine imidazolium ring to form two ring-opened diastereomeric 8,9-dihydro-8-(2,6-diamino-4-oxo-3,4-dihydropyrimid-5-yl-formamido)-9-hydroxy-aflatoxin B₁ (AFB₁-FapyGua) adducts. The AFB₁ formamidopyrimidine (Fapy) adducts induce G → T transversion mutations and are likely responsible for the carcinogenic effects of AFB₁. Quantitative liquid chromatography-mass spectrometry (LC-MS) methods have shown that AFB₁-N7-Gua is eliminated in rodent and human urine, whereas ring-opened AFB₁-FapyGua adducts persist in rodent liver. However, fresh frozen biopsy tissues are seldom available for biomonitoring AFB₁-DNA adducts in humans, impeding research advances on this potent liver carcinogen. In contrast, formalin-fixed-paraffin-embedded (FFPE) specimens used for histopathological analysis are often accessible for molecular studies. However, ensuring nucleic acid quality presents a challenge due to incomplete reversal of formalin-mediated DNA crosslinks, which can preclude accurate quantitative measurements of DNA adducts. In this study, employing ion trap or high-resolution accurate Orbitrap mass spectrometry, we demonstrate that ring-opened AFB₁-FapyGua adducts formed in AFB₁-exposed newborn mice are stable to the formalin fixation and DNA de-crosslinking retrieval processes that we have established. The AFB₁-FapyGua adducts can be detected at levels comparable to those levels in matching fresh frozen liver. Orbitrap MS² measurements can detect AFB₁-FapyGua at a quantification limit of 4.0 adducts per 10⁸ bases when assaying only 0.8 μg DNA on the column. Thus, our breakthrough DNA retrieval technology can be adapted to screen for AFB₁-DNA adducts in FFPE human liver specimens from cohorts at risk of this potent liver carcinogen.

Graphical Abstract

Introduction

AFB₁ is produced by several strains of Aspergillus, mostly A. flavus and A. parasiticus, which can grow on foods and contaminate peanuts, corn, rice, cottonseed, ground and tree nuts, especially in the warm and humid climates of Eastern and Southeastern Asia, Central America, and sub-Saharan Africa.^1,2 Consumption of foods contaminated with AFB₁ is an established potent risk factor associated with elevated frequencies of hepatocellular carcinomas (HCC). The International Agency for Research on Cancer (IARC) has classified AFB₁ as a Group 1 carcinogen.² Epidemiological data demonstrate that ingestion of foods contaminated with aflatoxin-producing molds, along with concomitant acute and chronic hepatitis infections and drinking water contaminated with microcystins, are well-correlated with early onset HCC, with males being preferentially susceptible to tumor formation.² Even with large-scale disease prevention strategies such as mass vaccination against hepatitis B and increased quality control measures to minimize aflatoxin exposures, AFB₁ remains a contributor to hepatocellular carcinogenesis worldwide, affecting upward of 800,000 deaths each year.³ Regulatory agencies in many countries have established limits on the acceptable levels of aflatoxin in food and feed products to protect public health; however, exposure to this deadly carcinogen still occurs.

The bioactivation of AFB₁ occurs predominantly by CYP3A4 and CYP1A2 oxidation in the liver,⁴ generating the 8,9-exo- and endo-epoxides (Scheme 1). The exo-epoxide reacts at the N7 atom of Gua to form the trans-8,9-dihydro-8-(N7-guanyl)-9-hydroxy-aflatoxin B₁ adduct (AFB₁–N7-Gua) in DNA.^5–7 The cationic AFB₁-N7-Gua adduct is labile and undergoes spontaneous depurination to generate an abasic site or decomposes by a second route involving base-catalyzed opening of the guanine imidazolium ring, yielding 8,9-dihydro-8-(2,6-diamino-4-oxo-3,4-dihydropyrimid-5-yl-formamido)-9-hydroxy-aflatoxin B₁ (AFB₁-FapyGua).⁸ AFB₁-FapyGua exists as a number of interconverting species, which include E/Z geometric isomers of the formamide group, aR/aS atropisomers about the C5-N⁶-bond, and α/β anomer of the glycosidic bond. Structural studies demonstrated AFB₁-FapyGua prefers the α-anomeric configuration in single-strand DNA but the β-anomer in duplex DNA (Scheme 1).⁹ The AFB₁-FapyGua DNA adducts are more persistent than AFB₁-N7-Gua in vivo and are thought to be the dominant lesions responsible for the genotoxicity of AFB₁.^10,11 AFB₁ DNA adducts induce G → T transversion mutations in Escherichia coli, mammalian cells, and rodent models with the ring-opened Fapy adducts being more mutagenic than AFB₁ – N7-Gua adduct.^12,13 The α-AFB₁-Fapy-dG adduct strongly blocks replication, whereas the minor β-form is primarily responsible for the G →T mutations and 6-fold more mutagenic than AFB₁-N7-Gua.¹² Ensuing studies employing high-fidelity DNA sequencing showed that mutational spectra generated from the liver DNA of male mice following AFB₁ exposure mirrored those of human liver tumors of individuals exposed to this carcinogen, further revealing the critical role of AFB₁ in human liver carcinogenesis.¹⁴

Scheme 1.

The bioactivation of AFB₁ and the reactivity of the AFB₁ exo-8,9 epoxide with DNA to form AFB₁-N7-Gua, and ring-opened AFB₁-FapyGua diastereomers, and abasic site formation.

DNA adduct measurements provide critical information about human exposure to genotoxic chemicals and can be employed to elucidate mechanisms of carcinogen bioactivation, DNA damage, chemoprevention, and repair.¹⁵ DNA adducts can serve as biomarkers for interspecies comparisons of the biologically effective dose of procarcinogens and permit extrapolation of genotoxicity data from animal studies for human risk assessment.^16,17

AFB₁ DNA adducts are commonly assayed following acid treatment of DNA, where AFB₁-N7-Gua and AFB₁-FapyGua are released from the DNA backbone by hydrolysis of the glycosidic bond; the AFB₁-FapyGua base exist as an equilibrium mixture of the aS- (minor) and aR-atropisomers (major) (Scheme 1) which separate by reversed-phase HPLC.⁹ (Footnote: We refer to these atropisomers as aS-AFB₁-FapyGua and aR-AFB1-FapyGua, (Reference 9, and Eliel, E.L.; Wilen, S.H. Stereochemistry of Organic Compounds; John Wiley & Sons, Inc.: New York, 1994; p 1120) instead of cis- (minor) and trans-AFB₁-FapyG (major) previously used to denote these diastereomers). AFB₁-N7-Gua has been measured in human urine by HPLC with UV detection following monoclonal antibody affinity purification,^18,19 and more recently by triple quadrupole mass spectrometry employing ¹⁵N-labeled AFB₁-N7-Gua as an internal standard.²⁰ The ring-opened AFB₁-FapyGua adducts were also detected in the urine of AFB₁-exposed rats.²¹ The AFB₁ DNA adduct measurements in rodent liver were performed following isolation and acid treatment of DNA, recovering AFB₁-N7-Gua and the diastereomeric ring-opened AFB₁-FapyGua adducts (Scheme 1).^11,22 Recent studies reported the simultaneous quantification of AFB₁–N7-Gua, aS-AFB₁-FapyGua and aR-AFB₁-FapyGua in mouse liver DNA by triple quadrupole mass spectrometry employing each isotopically ¹⁵N-labeled internal standard.^23,24

A major challenge in human DNA adduct biomarker research is the paucity of fresh frozen tissue. However, archived, formalin-fixed-paraffin-embedded (FFPE) tissues with clinical diagnoses for the disease are often accessible for chemical analyses. We recently established a method to efficiently reverse formalin-mediated crosslinks of DNA in FFPE tissues under mild conditions, allowing for quantitative measurements of DNA adducts formed with aristolochic acid-I, aromatic amines, heterocyclic aromatic amines, nicotine-derived nitrosamine ketone (NNK), and polycyclic aromatic hydrocarbon carcinogens.^25,26 This breakthrough technology paves the way for advancing DNA adduct biomarker research with human FFPE tissues.^25,26 In this study, we show that ring-opened AFB₁-FapyGua adducts are stable towards the FFPE and DNA retrieval processes in the newborn mouse model.¹¹ We employed isotopically labeled AFB₁-[¹⁵N₅]-FapyGua that were site-specifically synthesized in an 11-mer oligodeoxynucleotide as the internal standards.^9,24

A key advantage of using the 11-mer AFB₁-Fapy-[¹⁵N₅]-dG as the internal standard over the AFB₁-[¹⁵N₅]-FapyGua nucleobase is that the AFB₁-Fapy-[¹⁵N₅]-dG in the oligodeoxynucleotide_undergoes the same acid hydrolysis chemistry as the DNA samples to produce AFB₁-FapyGua and thus, compensates for variations in sample processing conditions, allowing for accurate DNA adduct quantification. AFB₁-FapyGua adducts in FFPE mouse liver DNA are recovered at comparable levels to those in matching fresh frozen samples. With further refinements, our technology can screen for AFB₁ DNA adducts in archived FFPE human specimens, exploring the impact of genetic polymorphisms in xenobiotic metabolism and DNA repair enzymes in AFB₁-DNA adduct formation and carcinogenesis.^11,27

Materials and Methods.

Gentra Puregene DNA isolation kits were purchased from Qiagen (Valencia, California). Z-Fix buffered zinc formalin 10% was from Anatech Ltd. (Battle Creek, MI). HLB 30 mg cartridges were from Waters (Milford, MA). Optima LC-MS formic acid, water, methanol, acetonitrile were purchased from Thermo Fisher Scientific (Waltham, MA). The Z-Fix buffered zinc formalin 10% was from Anatech Ltd. (Battle Creek, MI). The FFPE plus LEV DNA Purification kit was from Promega Maxwell (Madison, WI). AFB₁ was purchased from Sigma-Aldrich Chemical Co. (St. Louis, MO). The AFB₁ epoxide was synthesized as described.¹⁰ Caution: AFB₁ is a potent human liver carcinogen and should be handled in a well-ventilated hood with appropriate lab clothing and gloves. The AFB₁ epoxide should also be handled with the same precautions and lab safety measures. An unmodified 11-mer oligodeoxynucleotide with the sequence 5′-(CCATCGCTACC)-3′, (G = [¹⁵N₅]G or G) was synthesized on a PerSeptive Biosystem model 8909 DNA synthesizer on a 1 μmol scale using Expedite reagents (Glen Research, Sterling, VA) with the standard synthetic protocol for the coupling of the unmodified bases and ¹⁵N₅-deoxyguanosine phosphoramidite as described.⁹ The AFB₁ epoxide was reacted with unlabeled and ¹⁵N-labeled oligodeoxynucleotides to form the AFB₁-FAPY-Gua adducts. The purification, UV spectroscopy, and mass spectral characterization of the unlabeled and ¹⁵N₅-labelled AFB₁-FapyGua oligodeoxynucleotides were reported.^9,24 A molar extinction coefficient of 18,000 at 368 nm was used to determine the concentrations of the AFB₁-Fapy-dG 11-mer oligodeoxynucleotides.^8,28

Animals and dosing.

The breeding and care of Neil1 knockout mice was performed using pre-approved protocols through the Oregon Health & Science University Institutional Animal Care and Use Committee and monitored by the Department of Comparative Medicine. Male and female newborn Neil1^−/− mice backcrossed (19 generations) into a C57Bl6 background were mated. The 6-day-old pups were weighed and then given an intraperitoneal injection of freshly reconstituted AFB₁ in (DMSO) (10 mmol/L) at a 7.5 mg/kg dose or with DMSO only. The pups were returned to their original cages, and after 48 h were euthanized by CO₂ asphyxiation followed by decapitation. Livers were immediately harvested and cut into two sagittal sections. One section was immediately frozen in liquid nitrogen. The second section was immersed in formalin cups containing 20 vol of Z-Fix buffered zinc formalin 10% (20 mL) for 24 h. Thereafter, the tissues were placed in a tissue processor, followed by serial washes of the tissue with increasing percentages of ethanol in water, followed by washing with xylene at 42 °C, before embedding in paraffin, at the Histology Core, University of Minnesota.^29,30 The tissues were stored in the paraffin blocks in the dark at room temperature (up to 2 years) until used for DNA adduct measurements.

DNA isolation from fresh frozen and FFPE mouse liver tissues.

Frozen mouse liver (20 mg) was homogenized in 2 mL of TE buffer (50 mM Tris-HCl pH 8.0 and 10 mM EDTA containing 10 mM βME) using a blade homogenizer (Pro Scientific, Oxford, CT). The homogenates were centrifuged at 3000g at 4 °C for 10 min. The nuclei pellet was resuspended in 50 mM Tris-EDTA buffer, pH 8.0, containing 10 mM EDTA, and digested with RNase A and RNase T₁ for 90 min, followed by proteinase K treatment for 3 h at 37 °C.²⁵ The proteins were removed by the addition of chilled Puregene protein precipitation solution, and DNA was precipitated by adding 0.1 vol of 5 M NaCl followed by 2 vol of isopropanol as described.²⁶

The mouse liver FFPE specimens were dissected from the paraffin block, and the bulk of the paraffin was removed with a scalpel. The tissues were submerged in p-xylene and then underwent serial rehydration with ethanol and increasing percentages of water, followed by homogenization and centrifugation at 3000g to obtain the nuclear pellet.^30,31 The rehydrated mouse liver specimens (the equivalent of 20 mg of rehydrated tissue) were incubated for 16 h at 50 °C in the manufacturer’s lysis buffer with proteinase K to reverse the formalin-mediated crosslinks. The DNA was isolated using Maxwell^® 16 FFPE plus LEV DNA Purification kit with Promega Maxwell 16 MDx system (Madison, WI), following the manufacturer’s protocols with minor modifications as described.^29,30 The DNA concentration was measured by UV absorbance at 260 nm (an absorbance of 1.0 corresponds to 50 μg of double-stranded DNA). The UV spectra ratio of 260 and 280 nm absorbances was ~1.80 to 1.85 and considered high-purity DNA.³¹ The DNA was further characterized by HPLC with UV detection following nuclease digestion.²⁵ All four canonical 2′-deoxynucleosides were recovered in high yield with no detection of incompletely digested oligodeoxynucleotides, and monitoring for guanosine revealed that RNA contamination was <1% (Figure S1). Thus, the recovered DNA was largely devoid of formalin-mediated crosslinks and high purity.

DNA hydrolysis and recovery of AFB₁-FapyGua adducts.

The DNA samples (5 μg) were dissolved in 100 μL of 0.1 N HCl in Eppendorf safe-lock tubes. The equivalent of 5 adducts per 10⁸ nts of total AFB₁-[¹⁵N₅]-FapyGua 11-mer oligodeoxynucleotide containing aR/aS-AFB₁-Fapy-dG forms were added to the samples and heated at 95 °C for 1 h. After cooling, the DNA hydrolysate was neutralized with 100 μL of 0.1 N NaOH and processed by solid phase extraction (SPE) with a Waters HLB 30 mg resin. The HLB resin was prewashed with LC-MS grade MeOH with 0.1% HCO₂H (2 × 1 mL), then with LC-MS grade 90% H₂O/10% CH₃OH/0.1% HCO₂H (2 × 1 mL). The DNA hydrolysates were applied to the resin, washed with 90% H₂O/10% MeOH/0.1% HCO₂H (2 × 1 mL), and manually eluted with 1 mL of CH₃OH/0.1% HCO₂H and vacuum centrifuged to dryness in 1.5 mL Eppendorf tubes. The DNA hydrolysate extracts were reconstituted in 95% 0.1% HCO₂H/5% CH₃CN (25 μL).

Calibration curves.

The calibration curves were constructed for aS-AFB₁-FapyGua (minor) and aR-AFB₁-FapyGua (major) isomers in calf thymus (CT) DNA, serving as the background DNA matrix. The single-stranded 11-mer containing AFB₁-Fapy-¹⁵N₅-labeled-dG diastereomers was added to the CT DNA (5 μg) at 5 adducts per 10⁸ DNA bases. The relative amount of the aS-AFB1-FapyGu to aR-AFB₁-FapyGua was 15.1 ± 0.2% in the unlabeled 11-mer and 15.3 ± 0.1% in the labeled oligodeoxynucleotides following acid hydrolysis (vide infra). The single-stranded 11-mer containing AFB₁-Fapy-dG was added to the CT DNA at eight calibrant levels ranging from 0 up to per 1.54 adducts per 10⁶ bases for the minor aS-AFB₁-FapyGua, and up to 8.46 adducts per 10⁶ bases for the major aR-AFB₁-FapyGua adduct. The equivalent of 0.8 μg equivalent of DNA hydrolysate was assayed, and each calibration level was done in triplicate. The linear regression was performed using ordinary least squares with equal weightings. The regression curves were plotted as the observed AFB₁-FapyGua adduct versus the calculated levels.

Nanoflow Liquid Chromatographic and Mass Spectrometric Analysis of DNA Adducts by Velos Ion Trap MS³.

Ion trap MS³ with the Velos Pro (Thermo Fisher Scientific, San Jose, CA) was conducted with an UltiMate 3000 RSLC nano UHPLC System and Easy-Spray^™ Flex ion source (Thermo Fisher Scientific, San Jose, CA). The chromatography was done by direct injection. The analytical column was a Prontosil C18AQ reversed-phase column (0.1 × 150 mm, 3 μm particle size, 100 Å pore size) from Michrom, Auburn, CA. The A and B solvents were (A) 0.01% HCO₂H in H₂O and (B) 0.01% HCO₂H in 95% CH₃CN. DNA hydrolysate (0.8 μg equivalent of DNA hydrolysate, 4 μL) was assayed by direct injection. The chromatography was done at a flow rate of 1 μL/min and commenced at 5% B for 4 min. Thereafter, the gradient increased to 50% at 12 min and then to 95% B at 13 min and held at 95% B for 3 min, followed by a 1.0 min gradient to 5% B with a 7 min equilibration before commencing a new analysis. The source parameters were as follows: ion transfer tube temperature, 250 °C; spray voltage, 2250 V (positive ion mode); S-lens RF level%, 70. The analysis was conducted at the MS³ scan stage. The maximum injection time was 50 ms. The isolation widths were m/z 4.0 and 1.0, respectively, for the MS² and MS³ scan events. The MS transitions were m/z 498.1 → 480.1 → for AFB₁-FapyGua and 503.1 → 485.1 → for AFB₁-[¹⁵N₅]-FapyGua. The normalized collision energies were 25% and 30%, and the activation Q was 0.25 and 0.4 for MS² and MS³ scan events, respectively. The activation times were 10 ms for both scan events. The data were acquired in centroid mode by Xcalibur^™ version 3.0.

Nanoflow Liquid Chromatographic and Mass Spectrometric Analysis of DNA Adducts by Orbitrap MS².

AFB₁-FapyGua analysis was performed using the same UltiMate 3000 RSLC nano UHPLC systems interfaced with an Orbitrap Fusion Tribrid MS (Thermo Fisher Scientific, San Jose, CA). Chromatography was performed with a Premier LC-MS Magic3 C18AQ column (50 μm × 200 mm, 3 μm particle size, 100 Å) interfaced to a nano Easy-Spray^™ Flex ion source. The LC solvents were (A) 0.01% HCO₂H in H₂O and (B) 0.01% HCO₂H in 95% CH₃CN. The DNA hydrolysate extract (0.8 μg equivalent of DNA, 4 μL) was injected onto a Thermo Acclaim PepMap trap cartridge RP C18 (0.3 × 5 mm, 5 μm particle size, 100 Å, Thermo Fisher Scientific, Waltham, MA), which was then washed with solvent A at a 12 μL/min flow rate for 2 min. After trapping, the adducts were back-flushed onto the analytical column at a flow rate of 0.5 μL/min and run for 4 min at 1% B. The gradient linearly increased to 40% B at 10 min, and then the flow rate decreased to 0.15 μL/min at 11.0 min. The gradient increased linearly from 40% at 11 min to 99% B at 19.0 min. The column was held for 1.5 min at 99% B at a flow rate of 0.5 μL/min, followed by a 1.5 min gradient to 1% B with a 6 min equilibration before commencing a new analysis. The source parameters were as follows: ion transfer tube temperature, 275 °C; spray voltage, 2200 V (positive ion mode); RF-lens, 59%; quadrupole isolation; m/z isolation width (m/z 3); normalized AGC 600%; injection time 246 ms; normalized CID collision energy 20%; CID activation time 10 ms; activation Q 0.25; Orbitrap detection, 30,000 resolution full width at half maximum (FWHM) at m/z 200. Extracted ion chromatograms (EICs) for AFB₁-FapyGua [M+H]⁺ m/z 498.1 → 480.1150 and 452.1201 and AFB₁-[¹⁵N₅]-FapyGua [M+H]⁺ m/z 503.1 → 485.1002 and 457.1052, attributed to the losses [M+H–H₂O]⁺ and 457.1052 [M+H–H₂O-CO]⁺ were plotted with a 10 ppm mass tolerance. The data were acquired in profile mode by Xcalibur^™ version 4.4.

Results and Discussion.

This study aimed to determine if ring-opened AFB₁-Fapy-dG adducts were stable towards the formalin-fixation-paraffin-embedding and the de-crosslinking procedures and if the DNA recovered from FFPE specimens could be used for quantitative AFB₁-DNA adduct measurements. The cationic AFB₁-N7-Gua was not detected and likely underwent depurination or hydrolysis to the AFB₁-FapyGua adducts during the FFPE or the DNA isolation procedure, which required proteinase K treatment at 50 °C for 16 h to reverse the DNA crosslinks fully. Therefore, we focused on the AFB₁-FapyGua adducts, which are more persistent lesions than the cationic AFB₁-N7-Gua adduct, as seen in rodents.^11,22

Internal standards, Limits of Quantification, Calibration Curves, and Method Performance.

Labeled AFB₁-[¹⁵N₅]Fapy-dG and unlabeled AFB₁-Fapy-dG 11-mer oligodeoxynucleotides (250 pg oligodeoxynucleotide spiked in 5 μg CT DNA) underwent acid hydrolysis at 95 °C for 1 h, followed by neutralization and SPE (vide supra). The relative amounts of aS- and aR-AFB₁-FapyGua diastereomers were determined with the Orbitrap at the LC/MS² scan stage as described above. The EIC signals were integrated for the unlabeled and ¹⁵N₅-labeled adduct AFB₁-FapyGua adducts using the transitions m/z 498.1 → 480.1150 and 452.1201 and m/z 503.1 → 485.1002 and 457.1052, respectively. The relative amount of the aS-AFB₁-FapyGua to aR-AFB₁-FapyGua was 15.1 ± 0.2% in the unlabeled 11-mer and 15.3 ± 0.1% in the labeled oligodeoxynucleotides after acid hydrolysis (4 independent measurements); these differences were not statistically significant. The 11-mer internal standard was added to DNA at 5 AFB₁-[¹⁵N₅]FapyGua adducts per 10⁶ bases. Thus, the level for the minor aS-AFB₁-Fapy[¹⁵N₅]Gua internal standard was 0.77 adducts per 10⁶ bases, and the major aR-AFB₁-Fapy[¹⁵N₅]Gua form was 4.23 adducts per 10⁶ bases.

The calibration curves had a goodness-of-fit regression value of r² ≥ 0.998 for the Velos and the Lumos MS instruments (Supporting information, Figures S2 and S3). The limit of detection (LOD) and quantitation (LOQ) values for the Velos ion trap were set at 3σ and 10σ SD units above the mean background level signal by plotting the data as absolute adduct values compared to the background signals in the fortified CT DNA hydrolysates.³² The LOQ values of the aS-AFB₁-FapyGua and aR-AFB₁-FapyGua adducts were ~8 adducts per 10⁸ DNA bases at the MS³ scan stage on the Velos when assaying 0.8 μg of DNA hydrolysate obtained from fresh frozen and FFPE tissues. The LOD and LOQ levels with the high-resolution Orbitrap MS-based instrument at the MS² scan stage are impacted by the number of scans acquired across the peak, and often, there are no measurable or sporadic background signals approaching the LOD. Therefore, the LOQ values for both AFB₁-FapyGua diastereomers were set at a signal ten-fold above the background with a minimum of 10 scans across the peaks. The LOQ values of the aS-AFB₁-FapyGua and aR-AFB₁-FapyGua adducts were estimated at 4.0 adducts per 10⁸ bases when assaying 0.8 μg of DNA hydrolysate on the column on the Orbitrap. Increasing the amount of DNA assayed by 2-fold lowered the quantification limits of the Orbitrap LC/MS² assay to ~2.5 adducts per 10⁸ bases with 1.6 μg DNA assayed. (unpublished observations, M. Bellamri and R. Turesky). Further improvements in sensitivity might be achieved by optimizing the SPE conditions to remove ion suppressive components or by chemical derivatization of AFB₁-FapyGua with xylene-quaternary ammonium cationic-based tagging reagents, which significantly increase signals of response of some adducts.³³

The performance of the analytical method is summarized in Table 1. Mouse liver DNA samples were pooled from fresh frozen and FFPE samples to have enough DNA for three independent hydrolyses and AFB₁-FapyGua measurements over 3 days. The intraday % coefficient of variations (% CV) were ≤ 10.0%, and the inter-day % CV was ≤ 15.4% for both adducts on both MS instruments. The mean DNA adduct levels of the fresh frozen DNA samples were about ~20% higher than for FFPE samples in this set of pooled DNA samples, signifying that the ring-opened AFB₁-FapyGua adducts are stable towards the formalin fixation and de-crosslinking processes. Our previous data reported for DNA adducts of the 2′-deoxynucleoside DNA adducts aristolochic acid, 4-aminobiphenyl, 2-amino-1-methyl-6-phenylimidazo[4,5-b]pyridine, and benzo[a]pyrene recovered from matching freshly frozen and FFPE rodent or human tissues were within 20% of each other on average.^25,26,34

Table 1.

AFB₁-FapyGua DNA adduct measurement reproducibility in fresh frozen and FFPE rat liver.^b

Analyte	Tissue	MS Instrument	Statistics	Day 1	Day 2	Day 3	Intra-day CV (%)	Inter-day CV (%)
aS-AFB1-FapyGua (minor)	Fresh	Velos	mean	0.55	0.44	0.54	10.0	15.1
			SD	0.07	0.04	0.04
	FFPE	Velos	mean	0.42	0.43	0.42	7.7	7.1
			SD	0.04	0.03	0.02
aS-AFB1-FapyGua (minor)	Fresh	Lumos	mean	0.69	0.60	0.64	6.0	8.7
			SD	0.04	0.03	0.04
	FFPE	Lumos	mean	0.58	0.47	0.54	8.7	12.6
			SD	0.03	0.07	0.05
aR-AFB1-FapyGua (major)	Fresh	Velos	mean	3.69	3.14	2.98	8.0	14.3
			SD	0.16	0.10	0.37
	FFPE	Velos	mean	3.05	2.60	2.54	6.6	11.0
			SD	0.12	0.31	0.10
aR-AFB1-FapyGua (major)	Fresh	Lumos	mean	4.06	3.23	3.26	6.7	15.4
			SD	0.27	0.09	0.26
	FFPE	Lumos	mean	3.31	2.63	2.71	5.3	14.1
			SD	0.13	0.19	0.12

^aDNA adducts are reported as adducts per 10⁶ bases, (mean and standard deviation (SD) with three independent replicates assayed per day. The Velos ion trap data were acquired at the MS³ scan stage, and the Orbitrap data were acquired at the MS² scan stage.

Analysis of AFB₁-FapyGua in fresh frozen and FFPE mouse liver DNA.

The measured AFB₁-FapyGua adduct levels in the liver of 7 newborn mice employing the Velos ion trap with multistage MS³ scanning and Orbitrap at the MS² scan stage are reported in Figure 1. Similar to previous studies, the aR-AFB₁-FapyGua was the predominant isomer, occurring at ~5 to 7-fold higher levels than the aS-AFB₁-FapyGua.^23,24 The levels of adducts recovered from fresh frozen and FFPE specimens were generally within 20% for most liver samples. Some minor differences in adduct levels between the Velos ion trap and Orbitrap MS were observed, particularly for the aS-AFB₁-FapyGua in mouse livers ML2 and ML3, which were about 3-fold above LOQ values on the Velos. The adduct level measurements of ML2 and ML3 were about 1.8-folder lower on the Orbitrap. These differences may be attributed to the Orbitrap’s higher mass accuracy and precision, where measurements were conducted at 30,000 resolution with a 10 ppm mass tolerance compared to the nominal mass resolution of the Velos ion trap MS, where background isobaric product ions may have inflated the adduct levels. We note that the adduct measurements acquired on the Orbitrap were done three months after those obtained on the Velos. Overall, the adduct level measures are in good agreement between both MS instruments. Representative Velos ion trap LC/MS³ EIC and MS³ product ion spectra of the ring-opened AFB₁-FapyGua diastereomers acquired from FFPE mouse liver DNA from an untreated control mouse and AFB₁-treated mouse DNA are shown in Figure 2. A representative Lumos Orbitrap LC/MS² EIC and MS² and MS³ product ion spectra AFB₁-FapyGua diastereomers are shown in Figure 3. The proposed fragmentation of AFB₁-FapyGua using CID at the MS² and MS³ scan stages with the Lumos Orbtrap are shown in Figure S4.

Figure 1.

aS-AFB₁-FapyGua and aR-AFB₁-FapyGua adduct measurements in AFB₁-treated newborn mice 48 h post-treatment with the (A) Velos Ion Trap and (B) Lumos Orbitrap MS. Data are reported as the mean and standard deviation (n = 3 independent analyses per mouse). ML1 is an untreated control mouse liver DNA sample. *P < 0.05 (unpaired t-test).

Figure 2.

(A) LC/MS³ EIC traces of an untreated FFPE negative control and AFB₁-treated mouse liver DNA samples. The AFB₁-[¹⁵N₅]Fapy-dG 11-mer was added at a level of 5 adducts per 10⁶ DNA bases, with the minor aS-AFB₁-[¹⁵N₅]FapyGua recovered at a level of 0.77 adducts per 10⁶ bases and the major aR-AFB₁-[¹⁵N₅]FapyGua recovered at 4.23 adducts per 10⁶ bases, following acid hydrolysis. The elution times are aS-AFB₁-FapyGua (t_R: 7.1 min) and aR-AFB₁-FapyGua (t_R: 7.4 min). The adduct signals are normalized to the base peak of the aR-AFB₁-[¹⁵N₅]FapyGua. (B) MS³ product ion spectra of aS-AFB₁-FapyGua, aR-AFB₁-FapyGua, and their internal standards (lower panel). The equivalent of 0.80 μg DNA hydrolysate was assayed by the Velos Ion Trap MS. NL is the normalized ion counts.

Figure 3.

(A) LC/MS² EIC traces of an untreated FFPE negative control and AFB₁-treated mouse liver DNA samples. The AFB₁-[¹⁵N₅]Fapy-dG 11-mer was added at a level of 5 adducts per 10⁶ DNA bases, with minor aS-AFB₁-[¹⁵N₅]FapyGua recovered at a level of at 0.77 adducts per 10⁶ bases and the major aR-AFB₁-[¹⁵N₅]FapyGua recovered at 4.23 adducts per 10⁶ bases, following acid hydrolysis.. The elution times are aS-AFB₁-FapyGua (t_R:12.4 min) and aR-AFB₁-FapyGua (t_R: 13.4 min). (B) MS² and MS³ product ion spectra of aR-AFB₁-FapyGua, and its internal standard. The equivalent of 0.80 μg DNA hydrolysate was assayed by the Orbitrap Lumos MS. NL is the normalized ion counts.

The cationic AFB₁-N7-Gua undergoes rapid deglycosylation from rodent liver DNA, whereas the ring-opened AFB₁-FapyGua adduct is more persistent.^10,11,22 Brown and coworkers predicted that the β-anomeric AFB₁-FapyGua should be the principal AFB₁ lesion in human genomic DNA since the major α-anomer form is cytotoxic and blocks cell replication.^9,12 The long-term persistence and biochemical fate of AFB₁-FapyGua diastereomers in rodent or human liver tumors has not been investigated using specific and quantitative LC-MS-based methods. Putative AFB₁-FapyGua adducts have been detected in liver biopsy samples of patients with HCC by immunohistochemistry (IHC),^35–37 and putative AFB₁ adducts were detected in rat liver tumors more than 1 year post-AFB₁-dosing by IHC,³⁸ suggesting AFB₁-FapyGua adducts are highly persistent. However, the chemical identities of these lesions are unknown; they could be altered chemical structures of the AFB₁-FapyGua adducts or possibly other components in the liver that cross-react with the antibody.

As far as we are aware, there are no existing data in the literature reporting quantitative mass spectrometry measurements of AFB₁-FapyGua adducts in human liver tissue. The IHC studies imply the presence of AFB₁-FapyGua at levels exceeding one adduct per 10⁶ nucleotides;^35–37 however, the accuracy of these findings is uncertain due to potential issues with antibody specificity. Consequently, it is plausible that the actual level of DNA adducts may be lower than indicated by these studies. The sensitivity and quantification limits of our nanoflow LC high-resolution accurate Orbitrap MS² method are 50 to 100-fold lower than that reported by IHC. It is vital to cross-characterize AFB₁ DNA adducts by IHC and specific LC-MS-based methods to confirm DNA adduct identity in human tissues. Our technology established and validated in this study provides a quantitative method for measuring AFB₁-FapyGua adducts in human FFPE liver tissue.

In previous studies, we successfully recovered high-quality, fully nuclease-digestible DNA, ranging from 2.5 to 8 μg DNA, from two human kidney paraffin-embedded block section cuts (1 – 1.8 cm² in surface area and 10 μm in thickness).²⁵ Using this method, we screened non-tumor adjacent FFPE rodent and human tissues for DNA adducts of various carcinogens, including the renal carcinogen aristolochic acid-I (AA-I), the bladder carcinogen 4-aminobiphenyl (4-ABP), the proposed human prostate carcinogen 2-amino-1-methylimidazo[4,5-b]pyridine (PhIP), and other adducts formed with carcinogens present in tobacco smoke and grilled meats.^25,26,30,34 These DNA adducts were detected as the modified 2′-deoxynucleosides following nuclease digestion by ion trap or Orbitrap MS. Based on these findings, we anticipate that approximately five section cuts of human liver FFPE tissue would provide sufficient DNA for measuring AFB₁-FapyGua adducts. However, for AFB₁ DNA adduct measurements, we employ acid hydrolysis instead of nuclease digestion. This is because the ring-opened AFB₁-Fapy-dG obtained through nuclease digestion undergoes spontaneous equilibration of the α- and β-anomers and the furanose and pyranose forms of the deoxyribose ring, resulting in a complex mixture of Fapy species unsuitable for LC-MS analysis.⁹ The data reported in our current study demonstrate that our DNA retrieval technology from FFPE newborn mouse liver effectively recovers the ring-opened diastereomeric AFB₁-Fapy-dG adducts in high yield, which can be successfully screened as the modified AFB₁-FapyGua nucleobase following acid hydrolysis of DNA.

The screening of AFB₁-N7-Gua adducts in urine samples is a well-established and widely used method for measuring current exposure to AFB₁, while also serving as a valuable biomarker to assess chemoprotective efficacy.¹ However, the availability of fresh frozen biopsy tissue for human biomonitoring studies is often limited, impeding the advancement of DNA adduct biomarker research on cancer-causing agents. Employing archived liver tissues from biobanks enables retrospective studies, offering an opportunity to explore potential associations between AFB₁-FapyGua adduct levels with phase I and II enzymes involved in AFB₁ bioactivation and detoxification, as well as DNA repair enzymes. The use of FFPE specimens can significantly advance our understanding of genetic polymorphisms associated with the risk of AFB₁-related cancers.^1,11,39 Importantly, this approach is not feasible with urinary DNA adducts.

Conclusions

We successfully developed a robust method to recover ring-opened AFB₁-FapyGua DNA adducts from the FFPE blocks of newborn mouse liver tissue. The DNA adduct levels were comparable to those found in matching fresh-frozen tissue samples. The current quantification limit of the Orbitrap LC/MS² is 4 AFB₁-FapyGua adducts per 10⁸ bases with 0.8 μg DNA assayed. However, opportunities for improvement exist. Refinements in the SPE method, aimed at removing ion suppressive components, or the exploration of alternative approaches such as chemical derivatization of the AFB₁-FapyGua adducts with quaternary ammonium cationic functional groups, have the potential to enhance the adduct response during electrospray ionization, resulting in lowering the limit of assay quantification.