Methemoglobin Formation and Characterization Hemoglobin Adducts of Carcinogenic Aromatic Amines and Heterocyclic Aromatic Amines

Abstract

Arylamines (AA) and heterocyclic aromatic amines (HAA) are structurally related carcinogens formed during combustion of tobacco or cooking of meat. They undergo cytochrome P450 mediated N-hydroxylation to form metabolites which bind to DNA and lead to mutations. The N-hydroxylated metabolites of many AA also can undergo a co-oxidation reaction with oxy-hemolgobin (HbO₂) to form methemoglobin (met-Hb) and the arylnitroso intermediates, which react with the β-Cys⁹³ chain of Hb to form Hb-arylsulfinamide adducts. The biochemistry of arylamine metabolism has been exploited to biomonitor certain AAs through their Hb arylsulfinamide adducts in humans. We examined the reactivity of HbO₂ with the N-hydroxylated metabolites of 4-aminobiphenyl (ABP, HONH-ABP), aniline (ANL, HONH-ANL), and the HAA 2-amino-9H-pyrido[2,3-b]indole (AαC, HONH-AαC), 2-amino-1-methyl-6-phenylimidazo[4,5-b]pyridine (PhIP, HONH-PhIP) and 2-amino-3,8-dimethylimidazo[4,5-f]quinoxaline (MeIQx, HONH-MeIQx). HONH-ABP, HO-ANL, and HONH-AαC induced methemoglobinemia and formed Hb sulfinamide adducts. However, HONH-MeIQx and HONH-PhIP did not react with the oxy-heme complex, and met-Hb formation and chemical modification of β-Cys⁹³ residue was negligible. Molecular modeling studies showed that the distances between the H-ON-AA or H-ON-HAA substrates and the oxy-heme complex of HbO₂ were too far away to induce methemoglobinemia. Different conformational changes in flexible helical and loop regions around the heme pocket induced by the H-ON-AA or H-ON-HAAs may explain the different proclivities of these chemicals to induce methemoglobinemia. Hb-Cys^93β sulfinamide and sulfonamide adducts of ABP, ANL, and AαC were identified, by Orbitrap MS, following proteolysis of Hb with trypsin, Glu-C, or Lys-C. Hb sulfinamide and sulfonamide adducts of ABP were identified in blood of mice exposed to ABP, by Orbitrap MS. This is the first report of the identification of intact Hb sulfinamide adducts of carcinogenic AAs in vivo. The high reactivity of HONH-AαC with HbO₂ suggests that the Hb sulfinamide adduct of AαC may be a promising biomarker of exposure to this HAA in humans.

Graphical Abstract

INTRODUCTION

Aromatic amines (AA) and heterocyclic aromatic amines (HAA) are structurally related classes of environmental carcinogens.^1,2 Occupational exposures to AAs can occur in the rubber, textile, leather, dye, and cosmetic industries.^3–5 4-Aminobiphenyl (ABP), 2-naphthylamine (2-NA), and o-toluidine are also present in tobacco smoke and contribute to the pathogenesis of bladder cancer in smokers.⁶ Several HAAs, 2-amino-9H-pyrido[2,3-b]indole (AαC) in particular, occur as atmospheric pollutants⁷ and also arise in tobacco smoke.^8,9 Other prominent HAAs, including 2-amino-1-methyl-6-phenylimidazo[4,5-b]pyridine (PhIP) and 2-amino-3,8-dimethylimidazo[4,5-f]quinoxaline (MeIQx) are formed in during the cooking of meat.^1,10

AA and HAA undergo metabolic activation by cytochrome P450 (P450) oxidation to form genotoxic N-hydroxylated intermediates.^11,12 These metabolites can react with DNA or undergo further biotransformation by conjugating enzymes to form unstable esters.^13,14 The esters undergo heterolytic cleavage to form the presumed nitrenium ions,¹⁵ which readily form DNA adducts that can lead to mutations.¹⁶

DNA adducts of carcinogens have been employed to assess chemical exposures. However, the measurement of DNA adducts in human specimens remains an analytical challenge because many DNA lesions are rapidly repaired, and the trace levels of DNA adducts present in tissues can be difficult to measure, even with current MS instruments.^2,17 Reactive metabolites of AAs, HAAs, and other procarcinogens also bind to nucleophilic sites in proteins to form covalent adducts.^18,19 Stable carcinogen protein adducts do not undergo repair and are expected to follow the kinetics of the lifetime of the protein. Thus, adduct formation will accumulate during chronic exposure.^20,21 Hemoglobin (Hb) and serum albumin are the two major proteins in blood: 1 mL of blood contains about 150 mg Hb, and 30 mg albumin as compared to 0.003–0.008 mg of DNA.²¹ Therefore, carcinogen adducts of Hb or albumin may be more abundant and facile to detect than DNA adducts in blood samples. Both Hb and albumin have served as sources of protein to biomonitor the exposure to a number of genotoxicants and reactive electrophiles.^19,22–25

Hemoglobin is an α2β2 tetramer protein.²⁶ The N-terminal valine of the β-chain of Hb has served as a nucleophile to trap alkylating agents.²¹ Hb contains six reduced Cys residues. The two β-Cys⁹³ residues are surface-exposed to solvent and capable of adducting to electrophiles,^26,27 whereas the β-Cys¹¹² and α-Cys¹⁰⁴ reside in the interior portion of Hb and react more slowly with electrophiles.^26,28,29 The Cys³⁴ residue of human albumin is also accessible to the solvent and a strong nucleophile, which reacts with many electrophiles of diverse structure.²⁵ Adducts formed at these sites of Hb and albumin have been employed for biomonitoring exposures to hazardous chemicals.^19,22,23,25

The biochemistry of arylamine-induced toxicity and methemoglobinemia is well documented.^30,31 The arylhydroxylamine metabolites, largely produced by cytochrome P450s,^2,32,33 can penetrate the erythrocyte and undergo a co-oxidation reaction with HbO₂ to form met-Hb and the arylnitroso intermediates.^31,34 The arylnitroso compounds undergo enzymatic redox cycling by NADPH reductase within the erythrocyte to reform the aryhydroxylamine and commence another round of co-oxidation with HbO₂, leading to methemoglobinemia (Scheme 1). Arylnitroso intermediates are electrophiles, some of which react selectively with the β-Cys⁹³ residue of Hb chain to form sulfinamide adducts.^31,35,36 Hb-arylamine sulfinamide adducts are fairly stable in vivo,^20,35 but upon acid or base treatment in vitro, the adducts undergo hydrolysis to yield the parent amine and Hb-Cys^93β sulfinic acid.^34,37 The released aromatic amine (AA) can be readily measured by mass spectrometry (MS) methods. The Hb sulfinamide adduct of 4-aminobiphenyl (ABP) has been employed extensively as a biomarker to monitor occupational and environmental exposures to ABP and to assess the risk of urinary bladder cancer of smokers.^38–41 The elevated levels of Hb adducts of ABP, and other primary arylamines, such as 3-aminobiphenyl, 2-NA, o- and p-toluidine, 2,4-dimethylaninline, and 2-ethylaniline are associated with cigarette smoking.³⁹

Scheme 1.

Co-oxidation of HbO₂ by N-hydroxylated aromatic amines and heterocyclic aromatic amines in RBC

PhIP and MeIQx also undergo extensive N-oxidation in humans;^2,42-44 however, the reactivity of HONH-PhIP and HONH-MeIQx with Hb is very low, and Hb adducts of these procarcinogens are not promising biomarkers for exposure assessment.^45–49 To our knowledge, the reactivity of HONH-AαC with Hb has not been investigated. With our interest to develop biomarkers of HAAs with macromolecules for human risk assessment, we sought to understand the reactivity of N-hydroxylated intermediates of ABP, aniline (ANL), AαC, MeIQx, and PhIP (Figure 1) with HbO₂ in red blood cells (RBC), and compare the capacity of these N-hydroxylated species to induce methemoglobinemia and form adducts with Hb. AαC was found to be a more potent inducer of met-Hb formation in RBC than any of the other AAs or HAAs tested. The structures of the Hb adducts were elucidated following proteolysis, by ultraperformance liquid chromatography/tandem mass spectrometry (UPLC/MS²). Sulfinamide and sulfonamide adducts of ABP, ANL, and AαC were identified at the β-Cys⁹³ residue, whereas HONH-MeIQx and HONH-PhIP failed to react with HbO₂ and did not form adducts.

Figure 1.

Chemical structures of N-hydroxylated metabolites of ABP, ANL, MeIQx, PhIP, and AαC

MATERIALS AND METHODS

Caution

Many AAs, HAAs and, their derivatives are carcinogens and, must be handled in well ventilated hood with appropriate clothing.

Chemicals and Materials

Phenylhydroxylamine (HONH-ANL), trypsin, iodoacetamide (IAM) and, Pd/C were purchased from Sigma-Aldrich Chemical Co. (St. Louis, MO). AαC, MeIQx and, PhIP were procured from the Toronto Research Chemicals (Toronto, ON, Canda). [4b,5,6,7,8,8a-¹³C₆]-AαC was a gift of Dr. Daniel Doerge, National Center for Toxicological Research (Jefferson, AR). 4-Nitrobiphenyl was obtained from Apolo Scientific, Stockport, UK. MS grade Pierce^™Lys-C endoproteinase was purchased from Thermo Scientific (Grand Island, NY). Glu-C (endoproteinase) was purchased from Worthington Biochemical Corporation (Lackwood, NJ). Amicon Ultra Centrifugal Filters (10 kDa cut off) were purchased from Milllipore (Billerica, MA). LC-MS grade solvents were obtained from Fisher Scientific (Pittsburg, PA). 4-Chloromercuribenzoic acid (PMB) and tetrahdyrofuran (THF) were purchased from Alfa Aeser (Ward Hill, MA). Human whole blood was purchased from Bioreclamation LLC (Hicksville, NY). Isolute C18 solid phase extraction (SPE) columns (25 mg) were obtained from Biotage (Charlotte, NC). All other chemicals and solvents were ACS grade and procured from Sigma-Aldrich.

Synthesis of N-Oxidized Metabolites of AA and HAA

2-Nitro-1-methyl-6-phenylimidazo[4,5-b]pyridine (NO₂-PhIP) and 2-nitro-3,8-dimethylimidazo[4,5-f]quinoxaline (NO₂-MeIQx) were prepared by diazotization of PhIP and MeIQx.⁵⁰ 2-Nitro-9H-pyrido[2,3-b] indole (NO₂-AαC) and [¹³C₆]-NO₂-AαC were synthesized by the oxidation of AαC with dimethyldioxirane.⁵¹ The synthesis of 2-hydroxyamino-9H-pyrido[2,3-b]indole (HONH-AαC), [¹³C₆]-HONH-AαC, 2-hydroxyamino-1-methyl-6-phenylimidazo[4,5-b]pyridine (HONH-PhIP) and 4-hydroxyaminobiphenyl (HONH-ABP) was carried out by reduction of their respective nitro derivatives in THF, using hydrazine with Pd/C as a catalyst.^50,51 2-Hydroxyamino-3,8-dimethylimidazo[4,5-f]quinoxaline (HONH-MeIQx) was prepared by reduction of nitro derivative of MeIQx with ascorbic acid.⁵⁰ The hydroxyamino derivatives of PhIP and MeIQx were enriched by C-18 SPE, whereas the hydroxyamino derivatives of ABP and AαC were concentrated by vacuum centrifugation. All compounds were stored in ethanol except HONH-MeIQx and HONH-ABP, which were stored in DMSO/ethanol (3/1) in liquid nitrogen.

RBC and Hb Preparation

The RBC were prepared from whole blood as described previously.³⁵ Briefly, whole blood from healthy donors was centrifuged (800 × g for 10) at 4 °C to obtain packed RBC. The RBC were washed 3 times with cold Ringer’s buffer. To determine Hb concentration, RBC suspension was lysed by adding distilled water followed by sonication with a Branson 3800 ultra sonicator for 5 min. There after cell lysate was centrifuged at 25,000 × g at 4 °C. The HbO₂ in solution was measured at 542 nm using ε (M⁻¹ cm⁻¹) = 57.1.⁵²

Met-Hb Formation and Hb Adduct Formation with N-Hydroxy Derivatives of AAs and HAAs

HONH-AαC, HONH-ABP, HONH-PhIP, HONH-ANL or HONH-MeIQx (0.09 μM) was reacted with RBC containing 10 molar excess HbO₂ tetramer (0.9 μM) diluted in 0.5 mL of Ringer’s buffer (pH 7.4). The reaction mixture was incubated at 37 °C for 4 h. At various time points, an aliquot of RBC was lysed and met-Hb formation was estimated at λ_631nm using ε (M⁻¹ cm⁻¹) = 15.2.⁵² Met-Hb formation also was determined by incubating N-hydroxy derivatives of AAs and HAAs (0.09μM) with a 10 molar excess of purified HbO₂ (0.9 μM).

More concentrated solutions were employed to examine Hb adduct formation with AAs and HAAs. The RBC suspension contained tetrameric HbO₂ (1 nmol) in 0.25 mL of Ringer’s buffer (pH 7.4) and a 3-fold molar excess of HONH-ABP (3 nmol), HONH-ANL (3 nmol), or HONH-AαC (3 nmol). Incubations were conducted at 37 °C for a time period ranging from 15 to 120 min. The RBC were lysed and unreacted metabolites were removed by centrifugation with Amicon Ultra 10k cut-off filters. The Hb was alkylated with IAM (24 nmol) at 37 °C for 1 h. The excess of IAM was removed by Amicon Ultra centrifugal filters. The Hb was resuspended in 0.5 or 1 mL of either 100 mM potassium phosphate buffer (pH 7.4) or 50 mM ammonium bicarbonate containing 1 mM CaCl₂ (pH 8.5). In some studies, the modification of Hb was performed at 1/10^th and 1/100^th fold lower concentrations of HONH-AαC or HONH-ABP in Ringer’s buffer (1 mL).

For mass tag-data dependent mass spectrometry analysis of adducts of AαC, the RBC (1 nmol) was reacted with a 3-fold molar excess of a 1:1 ratio of HONH-AαC (1.5 nmol): HNOH-[¹³C₆] AαC (1.5 nmol) in 100 mM potassium phosphate buffer (pH 7.4, 0.25 mL) at 37 °C for 30 min.

Titration of Free Hb-β-Cys⁹³

Unmodified Hb and Hb modified with N-hydroxylated AAs and HAAs were titrated with p-chloromercuribenzoic acid (PMB) according to Boyer,²⁷ by measuring the increase in absorbance at 250 nm attributed to the mercaptide adduct formed at Cys³⁴.

Optimization of Proteolytic Digestion

The unmodified and modified Hb (5μg) were digested using trypsin or Lys-C at protease:protein ratio of 1:50 (w/w) or 1:25 (w/w) in 50 mM ammonium bicarbonate buffer (pH 8.5) containing CaCl₂ (1 mM) in a volume of 1 mL. For Glu-C digestion, Hb (5 μg) was digested with 1:50 (w/w) or 1:25 (w/w) Glu-C in 100 mM potassium phosphate buffer (pH 7.4, 1 mL). All the digestions were carried out at 37 °C for 18 h. After digestion, peptide adducts were desalted and enriched by C-18 SPE as described previously.⁵³ The eluted peptide adducts were concentrated to 100 μL under gentle stream of nitrogen in presence of 50 μL of DMSO.

Animal Study

The guidelines established by the National Institutes of Health Office of Laboratory Animal Welfare were adhered to for the use of animals with housing conditions. All protocols were reviewed and approved by the University of Minnesota Institutional Animal Care and Use Committee. Eleven weeks old adult male B6C3F1/J mice (Jackson Laboratory, Bar Harbor, ME) were used in this study. The animals (N = 5) were given intraperitoneal injection of ABP at a dose of 40 mg/kg body weight (100 μL of 80% DMSO : 20% H₂O). The blood was collected from the abdominal aorta into vacutainers containing potassium EDTA (Becton Dickinson), 24 h following dosing. The Hb was processed as described.³⁵

UPLC-Mass Spectrometry

A Thermo Dionex Ultimate 3000 Nano/Cap UPLC system (Thermo Scientific, San Jose, CA) with an Atlantis C18 nanoACQUITY column (0.3 × 150 mm, 3μm particle size, 100 Å, Waters Corp., Milford, MA) was used for separation of proteolytic digests of Hb. Solvent A contained 5% CH₃CN, 94.99% H₂O, and 0.01% HCO₂H, and solvent B was comprised of 95% CH₃CN, 4.99% H₂O, and 0.01% HCO₂H. For data dependent acquisition (DDA), a 50 min linear gradient was employed and commenced at 99% solvent A, and increased to 60% solvent B at 40 min, and then increased to 90 % solvent B at 50 min. For targeted measurements, a 20 min linear gradient was used and commenced at 99% solvent A and reached 60% solvent B for 15 min, and then arrived at 90 % solvent B at 20 min. The flow rate was set at 5 μL/min. Mass spectral data were acquired on an Orbitrap Elite^™ hydrid ion trap-Orbitrap mass spectrometer (Thermo Fisher, San Jose, CA) with a Michrom Advance CaptiveSpray^™ source (Auburn, CA). Typical tuning parameters were as follows: capillary tube temperature, 270 °C, spray voltage, 2.0 kV for positive ion mode; S-lens RF level, 67.5%; in-source fragmentation, 5V; normalized collision energy, 35 eV; activation Q, 0.3; for studies with HCD, the collision energy used was 20 eV. The full scan MS spectral data acquisition was carried out at a resolving power of 60,000 (at 400 m/z) over a mass range of 100–1500 m/z whereas, MS/MS data were acquired either with the iontrap or Orbitrap (resolving power 30,000) with an isolation width m/z = 2. All manipulations were conducted with Xcalibur version 3.0.63 software.

Data Analysis of Hb Adducts

The identification of Hb adducts was conducted with the BumberDash platform (Version 1.4.115) with Myrimatch search algorithm (Version 2.1.138) containing 32 protein subset of RefSeq human protein database, version 37.3 with forward and reverse protein sequences.^54,55 The dynamic modifications used for AαC-Hb adducts (AαC, +181.0640 on Cys, Ser, Thr, Val, His, Glu; AαC sulfinamide (AαCSO, +197.0589), AαC sulfonamide (AαCSO₂, +213.0538) and, Hb oxidation (sulfinic acid, +31.9898 and sulfonic acid, +47.9982) on Cys were previously described.⁵³ Dynamic modifications for ANL-Hb adducts; Aniline (ANL, +91.0422) on Cys, Ser, Thr, Val, His, Glu; Aniline sulfinamide (SOANL, +107.0371) and Aniline sulfonamide (SO₂ANL, +123.0320) on Cys whereas, for ABP-Hb adducts; ABP (+167.0735) on Cys, Ser, Thr, Val, His, Glu; ABP sulfinamide (SOABP, 183.0684) and ABP sulfonamide (SO₂ABP, 199.2059) on Cys were used in Myrimatch configuration file. For Trypsin, Lys-C digests, two missed cleavages were allowed with semi specificity and, mass tolerances 1.25 m/z (precursor ions) and 0.5 m/z (product ions). In Myrimatch algorithm, Glu-C enzyme was not present in the library; therefore, enzyme specificity was changed to nonspecific. The spectral identity was filtered through IDPicker algorithm with 1% false discovery rate.

Molecular Modeling

The molecular modeling work presented in this paper was carried out using MOE 2013.08 (Chemical Computing Group, Inc., Montreal, Canada.) Protein superposition was done using the “Protein Superpose” application,⁵⁶ generating pairwise root-mean-square deviation (RMSD) values.⁵⁷ The molecular docking simulation was conducted using the induced fit protocol within the MOE Docking application.^58,59 This protocol allows the active site residue side chains to be included in the final minimization stage of docking. PyMOL (Schrödinger, LLC) was used to color code the secondary structure of human hemoglobin (2DN1.pdb) according to a crystallographic temperature factor (B-factor) spectrum.

RESULTS

N-Oxidized metabolites of AAs undergo co-oxidation with HbO₂ to form met-Hb and the arylnitroso intermediate, which can react with the β-Cys⁹³ chain of Hb to form Hb-arylsulfinamide adducts (Scheme 1). The reactivities of HONH-ABP and HONH-ANL with HbO₂ and formation of the Hb-arylsulfinamide adducts are well documented.^31,37 We compared the reactivity of HbO₂ with these arylhydroxylamines and the N-hydroxylated intermediates of PhIP, MeIQx, and AαC, which are three prominent HAAs formed in cooked meat or tobacco smoke.¹ Met-Hb formation and titration of available β-Cys⁹³ were examined as a preliminary measure of reactivity of N-hydroxy-HAAs with Hb and adduct formation.

Kinetics of Met-Hb Formation in RBC Treated with N-Hydroxy-AAs or N-Hydroxy-HAAs

The incubation of RBC with 0.1 mol of HONH-ABP, HONH-ANL or HONH-AαC per mol HbO₂ tetramer resulted in a rapid oxidation of HbO₂, based on the increase in absorption at λ_631nm attributed to met-Hb formation (Figure 2A). In contrast, the reaction of HONH-MeIQx or HONH-PhIP with RBC did not show appreciable change in visible spectrum of HbO₂ (Figure 2A). The amount of met-Hb formation induced by HONH-AαC was ~1.4 – 1.6-fold higher than that level induced by HONH-ABP and HONH-ANL, whereas the formation of met-Hb was negligible for HONH-PhIP and HONH-MeIQx (Figure 2B).

Figure 2.

Met-Hb formation in RBC (0.9 μM HbO₂ tetramer in Ringer’s phosphate buffer, pH 7.4) incubated with N-hydroxy metabolites of AαC, ABP, ANL, MeIQx, PhIP (0.09 μM), or the control (C₂H₅OH as solvent vehicle). (A) Visible spectra of HbO₂ and met-Hb after 120 min. B) Kinetics of met-Hb formation in RBC over 120 min.

Free Thiols in Hb in RBC Treated with N-Hydroxy-AAs or N-Hydroxy-HAAs

The Hb in RBC was treated with a 30-fold higher concentration of N-hydroxy-AA or N-hydroxy-HAA (3 mol N-hydroxy substrate per mol HbO₂ tetramer). Following lysis of the RBC, the titration of untreated Hb with PMB revealed the presence of ~2 thiols per Hb tetramer, a value consistent to those previously reported in the literature (Table l).^31,60 These two thiols are attributed to the β-Cys⁹³ chain residues, which are surface-exposed and rapidly react with PMB, whereas the β-Cys¹¹² and α-Cys¹⁰⁴ residues are partially shielded and react far more slowly with PMB.^28–30,35 The treatment of RBC with HONH-ABP or HONH-AαC resulted in conversion of ~39 – 46% of HbO₂ to met-Hb, and the titration of the Hb with PMB revealed that the available β-Cys⁹³ residues were reduced to less than 0.1 mol “SH” per mol Hb tetramer (Table 1). These findings reveal that HONH-AαC, like primary arylhydroxylamines, undergoes co-oxidation with HbO₂ in the erythrocyte to form 2-nitroso-9H-pyrido[2,3-b]indole (NO-AαC). The NO-AαC undergoes redox cycling with NADPH reductase in the erythrocyte to reform HONH-AαC, resulting in a new round of co-oxidation with HbO_2, leading to methemoglobinemia.³⁰ The oxidation of HbO₂ and the diminution of the available β-Cys⁹³ residue of Hb treated with HNOH-MeIQx and HNOH-PhIP were negligible (Table 1).

Table 1.

Determination of Free Cys and met-Hb Formation by N-Hydroxy Metabolites of AαC, ABP, PhIP, and MeIQx

Substrate	Reduced β-Cys93	met-Hb (%)	mol met-Hb formed per mol of HONH-Substrate or H2Oa
Hb + H2O	2.00 ± 0.10	2.5 ± 0.14	0.03
Hb +HONH-MeIQx	1.96 ± 0.10	3.7 ± 0.18	0.05
Hb +HONH-PhIP	1.81 ± 0.10	4.7 ± 0.20	0.06
Hb +HONH-ABP	<0.10 ± 0.01	41.4 ± 0.33	0.55
Hb+HONH-AαC	<0.10 ± 0.02	48.8 ± 0.31	0.64

HbO₂ (tetramer) content of RBC was adjusted to 2 nmol and RBC were incubated with 6 nmol of HONH-AA or HONH-HAA for 30 min, followed by titration with PMB.

^amet-Hb formation in the presence or absence of HONH-AA or HONH-HAA (mol/mol); mol of met-Hb formation in the H₂O control was normalized to the ratio of the HONH-substrate treated cells by dividing amount of met-Hb formed by 6. Each value represents the mean and standard deviation of three experiments.

The poor reactivity of HONH-MeIQx and HONH-PhIP with HbO₂ could be due to the inability of the metabolites to penetrate the RBC membrane and react with the oxy-heme complex. Alternatively, the HONH-HAAs may be unstable in the RBC environment and react with small molecular weight nucleophiles, such as thiols, or the HONH-HAAs may undergo rapid enzymatic reduction to the parent amines.⁶¹ Therefore, we examined the reactivity of N-hydroxy metabolites of PhIP, MeIQx, as well as AαC, ANL, and ABP with purified HbO₂. We observed the same pattern of reactivity: HONH-AαC, HONH-ANL, and HONH-ABP oxidized HbO₂ to the met-Hb form, and the free thiol content was decreased to less than 0.1 mol per mol of HbO₂ tetramer. In contrast, HONH-PhIP and HONH-MeIQx did not induce oxidation of HbO₂, and the number of titratable thiols remained unchanged from the control (Supporting Information, Table S-1). Thus, HONH-PhIP and HONH-MeIQx do not efficiently react with the oxy-heme complex of HbO₂ and this explains their poor binding to Hb.

Mass Spectrometric Characterization of Hb Adducts of AαC

We sought to determine if the reduction of titratable β-Cys⁹³ residues in RBC treated with HNOH-AαC was attributed to Hb adduct formation or oxidation of the thiol group to the Cys-sulfinic acid (CysSO₂H) or Cys-sulfonic acid (CysSO₃H), by the H₂O₂ generated during the co-oxidation reaction of HONH-AαC with HbO₂.³⁰ UPLC/MS² was employed to characterize the Hb β-Cys⁹³ chain containing the presumed adducts of ABP, ANL, and AαC, following digestion with different proteases. Trypsin, Lys-C, chymotrypsin, and Glu-C were examined to determine optimal digestion conditions for recovery of the β-Cys⁹³ arylsulfinamide linkages. Trypsin cleaves peptide chain at C-terminal of Lys and or Arg, whereas Lys-C specifically cleaves at C-terminal of Lys.⁶² Glu-C is an endoprotease that cleaves peptide chain at the C-terminal acidic amino acids Glu and Asp, with higher specificity at Glu.⁶³ The peptide sequences recovered from the β-chain of Hb, by the different proteases, are depicted in Scheme 2.

Scheme 2.

Schematic of Hb-β chain digestion with trypsin, Lys-C, chymotrypsin and Glu-C and peptide sequence with β-Cys⁹³.

Trypsin and Lys-C digestion

The MS-tag DDA was performed Hb modified with 3-fold molar excess of 1:1 HONH-AαC and HONH-[¹³C₆]AαC, following digestion of Hb at a ratio of trypsin to Hb of 1:10. The peptide adducts identified by the Myrimatch algorithm are summarized in Table 2. Two tryptic peptides encompassing β-Cys⁹³, the single missed-cleavage peptide A) ⁸⁴GTFATLH*CDKLHVDPENFR¹⁰⁵ (M.W: 2529.8) and fully digested peptide B) ⁸⁴GTFATLH*CDK⁹⁶, (M.W: 1420.6) containing AαC sulfinamide (SOAαC) and AαC sulfonamide (SO₂AαC) adducts were identified (Scheme 2). These were the sole adducts of AαC detected by Myrimatch. This search engine also identified sulfinic (-SO₂H) and sulfonic acid (-SO₃H) at β-Cys⁹³, β-Cys¹¹² and α-Cys¹⁰⁴ residues of Hb; however, these thiol oxidation products were identified both in untreated and AαC-treated Hb (Table 2)

Table 2.

Peptide adducts and Oxidized Peptides Identified by Mass-tag Data Dependent Acquisition

Peptide	Observed precursor ions (m/z)	Z	M.W.	tR (min)	AαC-peptide adduct	Site of modification	Enzyme
α-Chain
α	754.7	4	3014.6	19.8	100LLSHC[SO3H]LLVTAAHLPAFTPAVHASLDK126	α-Cys104	Trypsin
	750.7	4	2998.6	20.5	100LLSHC[SO2H]LLVTAAHLPAFTPAVHASLDK126
β-Chain
A	686.3	4	2741.2	19.12	83GTFATLSELHC[SO2AαC]DKLHVDPENFR104a	β-Cys93	Trypsin
	682.3	4	2725.2	19.54	83GTFATLSELHC[SOAαC]DKLHVDPENFR104a
	641.1	4	2560.2	14.95	83GTFATLSELHC[SO2H]DKLHVDPENFR104a
	645.1	4	2577.2	15.59	83GTFATLSELHC[SO3H]DKLHVDPENFR104a
B	540.2	3	1617.7	16.91	83GTFATLSELHC[SOAαC]DK95a	β-Cys93	Trypsin or Lys-C
	545.6	3	1633.7	16.55	83GTFATLSELHC[SO2AαC]DK95a
	485.2	3	1452.7	11.40	83GTFATLSELHC[SO2H]DK95a
	490.6	3	1468.7	11.41	83GTFATLSELHC[SO3H]DK956a
D	501.6	3	1501.7	13.26	91LHC[SOAαC]DKLHCDPE101a	β-Cys93	Glu-C
	506.9	3	1517.7	14.10	91LHC[SO2AαC]DKLHCDPE101a
	496.9	3	1487.7	14.94	91LHC[SOABP]DKLHCDPE101b
	502.2	3	1503.7	15.14	91LHC[SO2ABP]DKLHCDPE101b
	471.6	3	1411.7	12.26	91LHC[SOANL]DKLHCDPE101c
	476.9	3	1427.7	12.59	91LHC[SO2ANL]DKLHCDPE101c
	446.5	3	1337.6	08.39	912LHC[SO2H]DKLHCDPE101a.b.c
	451.9	3	1352.6	08.43	91LHC[SO3H]DKLHCDPE1012a.b.c
E	584.7	3	1767.0	14.90	105LLGNVLVC[SO3H]VLAHHFGK120a	β-Cys112	Trypsin
	590.3	3	1751.0	15.41	105LLGNVLVC[SO2H]VLAHHFGK120a

Z – Charge state; t_R- Retention time; M.W- Molecular weight;

^aHONH-AαC-Hb modified sample;

^bHONH-ABP-Hb modified sample;

^cHONH-ANL-Hb modified sample.

The product ion spectra of the β-Cys⁹³SOAαC ([M+4H]⁴⁺ at m/z 682.8, upper panel) and β-Cys⁹³SO[¹³C₆]AαC (M+4H]⁴⁺ at m/z 684.3, lower panel) of peptide A, ⁸⁴GTFATLH*CDKLHVDPENFR¹⁰⁵, contain one major common product ion as the base peak at m/z 849.1 (Figure 3A). This fragment ion is the triply charged radical cation species GTFATLHC^[SO]DKLHVDPENFR, which is formed by the cleavage of the sulfinamide linkage ([M+3H-AαC^•]³⁺ and [M+3H-[¹³C₆]AαC^•]³⁺). The AαC^•+ isotopomers are not observed in the product ion spectra because of the low mass cutoff or the so-called “1/3 Rule”.⁶⁴ The product ion spectra of β-Cys⁹³SOAαC ([M+3H]³⁺ at m/z 540.2, upper panel) and β-Cys⁹³SO[¹³C₆]AαC ([M+3H]³⁺ at m/z 542.3, lower panel) of the fully digested tryptic peptide B, ⁸⁴GTFATLH*CDK⁹⁶, are shown in Figure 3C. In this case, product ions of AαC^•+ and [¹³C₆]AαC^•+ are observed, respectively, at m/z 183.1 and 189.1, along with the common ion at m/z 718.9, which is the doubly charged radical cation species GTFATLHC^[SO]DK.

Figure 3.

TIC and CID-MS² spectra of AαC-peptide A (GTFATLSELH*CDKLHVDPENFR) adducts (A) [M+4H]⁴⁺ of β⁹³Cys-SO-AαC at m/z 682.3 (upper panel) and ⁹³Cys-SO-[¹³C₆]AαC at m/z 683.8 (lower panel) at t_R 19.2 min; (B) [M+4H]⁴⁺ of β⁹³Cys-SO₂-AαC at m/z 686.3 (upper panel) and ⁹³Cys-SO₂-[¹³C₆]AαC at m/z 687.8 (lower panel) at t_R 19.5 min. TIC and CID-MS² spectra of AαC-peptide B (GTFATLSELH*CDK) adducts (C) [M+3H]³⁺ of β⁹³Cys-SO-AαC (upper panel) at m/z 540.2 and ⁹³Cys-SO-[¹³C₆]AαC (lower panel) at m/z 542.2 at t_R 16.55 min; (D) [M+3H]³⁺ of β⁹³Cys-SO₂-AαC (upper panel) at m/z 545.6 and ⁹³Cys-SO₂-[¹³C₆]AαC (lower panel) at m/z 547.6 at t_R 16.91 min of trypsin digest of Hb modified with 3 molar excess of HONH-AαC and HONH-[¹³C₆]AαC. Mass spectral data are analyzed using front end ion trap MS.

The Hb sulfonamide linkage of AαC is considerably more stable towards CID than the Hb sulfinamide linkage. The product ion spectra of the β-Cys⁹³SO₂AαC and β-Cys⁹³SO₂[¹³C₆]AαC adducts ([M+4H]⁴⁺ at m/z 686.3 and 687.8 of peptide A) in Figure 3B, and the product ion spectra Cys⁹³SO₂AαC and β-Cys⁹³SO₂[¹³C₆]AαC ([M+3H]⁺³ at m/z 545.6 and 547.6 of peptide B) in Figure 3D display a series of -y_n ions of the respective GTFATLHC^[SO₂AαC]DKLHVDPENFR and GTFATLHC^[SO₂AαC]DK structures. Characteristic -y′_n product ion series corresponding to the loss of SO₂AαC or SO₂[¹³C₆]AαC ion from -y_n ion series are also observed in the product ion spectra of the AαC sulfonamide adducts of B peptide (Figure 3D).

The AαC adducts of tryptic peptide A, based on ion counts, was 5 to 8 -fold higher than the fully digested AαC adducts of tryptic peptide B (Figure 1S1 in Supporting Information). However, peptide B with an optimal length (13 amino acid residues) for chromatography may be the preferred biomarker candidate peptide over larger sized single missed-cleavage peptide A with 21amino acid residues. The digestion of Hb-AαC adducts with a trypsin to Hb ratio of 1:10 yielded highest amount of adducted peptides A and B, but the large amount of protease also resulted in a large recovery of AαC, presumably as a result of hydrolysis of the sulfinamide adducts (Figure S1A–S1B in Supporting Information). Thus, we investigated the efficacy of other proteases to recover β-Cys⁹³ containing peptides of AαC in high yield, while minimizing the hydrolysis of sulfinamide linkage.

Lys-C, an endoproteinase, which cuts primarily at C-terminal Lys residues,⁶² produced B peptides containing SOAαC, SO₂AαC adducts, or -SO₂H and -SO₃H modifications at β-Cys⁹³ (Table 2, Scheme 2). Employing Lys-C for proteolysis at protease to Hb ratio of 1:50, increased the recovery of β-Cys⁹³SOAαC and β-Cys⁹³SO₂AαC adducts of peptide B by 7 and 8.4-fold in comparison to the digestion done with trypsin (Figure 4A). In contrast, the amount of AαC recovered from Hb digested with Lys-C was also 3-fold lower than that amount recovered from digestion of Hb with trypsin (Figure 4B).

Figure 4.

Ion counts of (A) β⁹³Cys-SO-AαC and β⁹³Cys-SO₂-AαC adducts of peptide B (GTFATLSELH*CDK) and peptide D (LH*CDKLHVDPE), (B) free AαC recovered from 1/50 Trypsin, 1/50 Lys-C or 1/25 Glu-C (enzyme to protein ratios) digests of Hb modified with 3 molar excess of HONH-AαC. Ion counts of β⁹³Cys-SO-AαC and β⁹³Cys-SO₂-AαC of peptide D ( adducts recovered from (C) 1/25 Glu-C and 1/50 Glu-C digests (enzyme to protein ratio) and (D) 1/25 Glu-C digestion at timer intervals of 4, 8,17 and 24 h of Hb modified with 3 molar excess of HONH-AαC. Data are plotted as mean and standard deviation of ion counts (N = 3). The level of adducts were determined from the area of total counts of CID-MS² spectral data analyzed using front end iontrap MS

Chymotrypsin Digestion

Chymotrypsin generates smaller sized peptides and frequently with one or two missed-cleavages, which restricts its application for this study (Scheme 2). Proteolysis of Hb with a chymotrypsin to Hb ratio of 1:25 or 1:50 protein did not result in the recovery of peptide adducts of AαC, and the level of free AαC was 5-fold higher than those amounts recovered by digestion with trypsin (1:10 trypsin to Hb). Thus, the majority of the sulfinamide adduct was hydrolyzed during digestion with chymotrypsin (data not shown). Multiple missed-cleaved peptides, including, HCDKL, SELHCDKL, and ATLSELHCDKL were also identified in chymotryptic digest of unmodified Hb (data not shown).

Glu-C Digestion

Glu-C is an endoproteinase that cleaves peptide bonds at the carboxy terminus of either Glu or Asp, with higher specificity towards Glu.⁶³ Two Glu residues are situated at positions 91 and 102 of the β-chain of Hb. Both SOAαC and SO₂AαC adducts within the peptide sequence ⁹²LH*CDKLHVDPE¹⁰² (M.W. 1304.6) were recovered by digestion with Glu-C (Table 2). The optimal amount of Glu-C employed for digestion of Hb used a ratio of Glu-C to Hb of 1:25, and proteolysis was conducted for 17 h There was no evidence of single missed-cleavage peptides, when assayed by MS-tag DDA and screened with the Myrimatch algorithm. The β-Cys⁹³SO₂H and β-Cys⁹³SO₃H containing modifications also were identified (Table 2, Figure S2 in Supporting Information). The signals of Hb β-Cys⁹³ adducts were greatest with Glu-C, followed by Lys-C, and lastly for trypsin (Figure 4A). Optimal digestion conditions, Glu-C:Hb ratio of 1:25 and a 17 h (Fig. 4C–D) were used for mass spectral characterization of Hb adducts of ABP and ANL described below.

Mass Spectral Characterization of Hb-AαC, Hb-ABP, Hb-ANL Adducts Following Digestion of Hb with Glu-C

Triply charged precursor ion species were the most abundant ions for all the peptide adducts. The reaction of RBC with a 3 mol excess of HONH-AαC, HONH-ABP or HONH-ANL to HbO₂ tetramer resulted in formation of both the sulfinamide and sulfonamide adducts at β-Cys⁹³, designated as peptide D, LH*CDKLHVDPE (Table 2). The product ion spectra are described below.

LH*CDKLHVDPE Sulfinamide Adducts of AαC, ABP, and ANL

The [M+3H]³⁺ precursor ions at m/z 501.5664 (t_R:13.26), 496.9029 (t_R:14.94) and 471.5591 (t_R:12.26) correspond, respectively, to β-Cys⁹³SOAαC, β-Cys⁹³SOABP, and β-Cys⁹³SOANL adducts of Peptide D. The observed m/z values for precursor ions of the adducts were within 3 ppm of the calculated m/z values. The adducted peptides were subjected to CID- or HCD fragmentation and accurate mass measurements were acquired with the Orbitrap. The product ion spectrum of β-Cys⁹³SOAαC ([M+3H]³⁺ at m/z 501.5664) obtained by CID, displays two fragment ions at m/z 183.0788 and 660.8107, corresponding to AαC^•+ and the doubly charged sulfoxide radical LHC^[SO]DKLHVDPE (Figure 5A1). The ion at m/z 660.8107 was also the base peak in the product ion spectra of β-Cys⁹³SOABP and β-Cys⁹³SOANL adducts (Figure 5B1, 5C1). In the case of Cys⁹³SOABP, the product ion at m/z 169.0887 (ABP^•+) was observed, whereas the radical cation of ANL at m/z 93.0570, arising by bond cleavage between SO and amine group of ANL was not seen due to low mass cutoff of the CID product ion spectra (Figure 5B1, 5C1). HCD product ion spectra were acquired on the triply charged precursor ions of β-Cys⁹³SOAαC (m/z 501.5664), β-Cys⁹³SOAαC (m/z 496.9029), and β-Cys⁹³SOANL (m/z 471.5591). The lower mass range of the spectra show the corresponding product ions of the amines at m/z 183.0787 (AαC^•+), 169.0888 (ABP^•+), and 93.0570 (ANL^•+) (Figure 5A2,B2, and C2), and the –b₂ and -y₂ -y₃ type product ions were present in the HCD product ion spectra. Consecutive reaction monitoring was done by CID at the MS³ scan stage of the product ion at m/z 660.8100 ([M′]^2•+), which is in common to all three arylsulfinamde adducts. A complete series of -b₂-b₉ and -y₂-y₁₀ ions confirmed the sulfoxide structure (LHC^[SO]DKLHVDPE) and respective LH*CDKLHVPE sulfinamide adducts of AαC, ABP and, ANL (Figure 5D). All the product ions were observed within 3 ppm of the calculated values.

Figure 5.

TIC and MS² spectra of [M+3H]³⁺ of peptide D (LH*CDKLHVDPE) adducts (A1) CID, (A2) HCD of β⁹³Cys-SO-AαC at m/z 501.5664 at t_R 13.26 min; (B1) CID, (B2) HCD of β⁹³Cys-SO-ABP at m/z 496.9029 at t_R 14.94 min; (C1) CID, (C2) HCD of β⁹³Cys-SO-ANL at m/z 471.5591 at t_R 12.26 min and (D) CID-MS³ of LHC^[SO]DKLHVDPE [M′^•]²⁺ at m/z 471.5591 >660.8101 > recovered from the Glu-C (1/25^th enzyme to protein ratio) digest of Hb modified with 3 molar excess of respective HONH-AαC, HONH-ABP or HONH-ANL. MS/MS spectral data were acquired on Orbitrap MS at 30,000 resolving power.

LH*CDKLHVDPE Sulfonamide Adducts of AαC, ABP and ANL

The CID product ion spectra of [M+3H]³⁺ precursor ions of β-Cys⁹³SO₂AαC (m/z 506.8980), β-Cys⁹³SO₂ABP (m/z 502.2345), and β-Cys⁹³SO₂ANL (m/z 476.8907), show the presence of typical –b_n (-b₂-b₁₀) and –y_n (-y₂-y₃ and-y₉-y₁₀) series of ions, which confirm the structures of the adducts of LH*CDKLHVDPE (Peptide D), (Figure 6A1, B1 and C1). The precursor ion m/z values were observed within 3 ppm of calculated m/z values. The C-terminal of Peptide D is an acidic Glu residue, which resulted in high abundance and sequence coverage of –b_n ions. The ions at m/z 909.4035, 1024.4306 and, 1121.4838 in the product ion spectrum of precursor ion at m/z 506.8980 (SO₂AαC) were assigned as putative -b type internal ions as a result of loss of N-terminal dipeptide Leu-His from -b8-Leu-His (*CDKLHV), -b₉-Leu-His (*C^[SO₂AαC]DKLHVD), and -b₁₀-Leu-His (*C^[SO₂AαC]DKLHVDP) ions (Figure 6A1).⁶⁵ Similarly, -b type internal ions as a result of loss of N-terminal dipeptide Leu-His from -b_n ions are observed in the product ion spectra of SO₂ABP adduct (m/z 502.2345) and SO₂ANL (m/z 476.8907) (Figure 6B1, C1).

Figure 6.

TIC and MS² spectra of [M+3H]³⁺ of peptide D (LH*CDKLHVDPE) adducts (A1) CID, (A2) HCD of β⁹³Cys-SO₂-AαC at m/z 506.8980 at t_R 14.1 min; (B1) CID, (B2) HCD of β⁹³Cys-SO₂-ABP at m/z 502.2345 at t_R 15.14 min; (C1) CID, (C2) HCD of β⁹³Cys-SO₂-ANL at m/z 476.8907 at t_R 12.59 min recovered from the Glu-C (1/25^th enzyme to protein ratio) digest of Hb modified with 3 molar excess of respective HONH-AαC, HONH-ABP or HONH-ANL.MS/MS spectral data were acquired on Orbitrap MS at 30,000 resolving power.

The HCD product ion spectrum of β-Cys⁹³SO₂AαC, ([M+3H]³⁺ at m/z 506.8980), shows fragment ions at m/z 184.0869 and 183.0791, which are attributed to protonated AαC and the radical cation of AαC^•+ (Figure 6A2). Similarly, the HCD spectra of β-Cys⁹³SO₂ABP and β-Cys⁹³SO₂ANL ([M+3H]³⁺ at m/z 502.2345, and m/z 476.8907, respectively) display product ions at m/z 169.0888 and 93.0570, corresponding to ABP^•+ and ANL^•+ (Figures 6B2, 6C2).

Concentration-dependent Adduct Formation of AαC and ABP in RBC

The reactivity of HONH-AαC and HONH-ABP with Hb was carried out over a 300-fold concentration range (0, 3, 0.1 and 0.01 mol of HONH substrate to HbO₂ tetramer). The Hb adducts were monitored by measuring ion counts of sulfinamide and sulfonamide of Peptide D adducts. The adduct formation and recovery of AαC and ABP showed an approximate linear-concentration response relationship (Figure 7A,B).

Figure 7.

Ion counts of peptide D (LH*CDKLHVDPE) adducts, (A) β⁹³Cys-SO-AαC and β⁹³Cys-SO₂-AαC adducts and, (B) β⁹³Cys-SO-ABP and β⁹³Cys-SO₂-ABP adducts recovered from the Glu-C digests of 1 mole Hb modified with 3, 0.1 and 0.01 mol equivalents of HONH-AαC or HONH-ABP. The level of adducts were determined from the area of total counts of CID-MS² spectral data analyzed using orbitrap MS analyzer at 30,000 resolving power. (C) Ion counts of peptide D (LH*CDKLHVDPE) of β⁹³Cys-SO₂H and β⁹³Cys-SO₃H recovered from the Glu-C digests of 1molar Hb modified with 3, 0.1 and 0.01molar excess of HONH-AαC or HONH-ABP. Data are plotted as mean and standard deviation of ion counts (N = 3).

ABP-Hb Adduct Formation in Mouse

The β-chain of mouse and human Hb share 80.3% sequence homology, and the amino acid sequence is identical in both species from residues 74–108 containing the β-Cys^93.66 Thus, we sought to determine if the intact sulfinamide and/or sulfonamide adduct of ABP could be recovered in vivo from mice dose with ABP (40 mg/kg body weight). Hb from ABP-treated and untreated mice was collected and digested with Glu-C as described above. Mouse Hb digests (200 ng Hb) were used for targeted UPLC-MS/MS analysis in the Orbitrap, by assaying for β-Cys⁹³SOABP ([M+3H]³⁺ at m/z 496.9029) and β-Cys⁹³SO₂ABP ([M+3H]³⁺ at m/z 502.2345) adducts of Peptide D (LH*CDKLHVDPE), and free ABP ([M+H]⁺ at m/z 170.0964) released during proteolysis of mouse Hb-ABP. The HbCys^93β-ABP adducts and free ABP are present in the ion chromatograms of Hb obtained from ABP-treated mice (Figure 8), but the signals are absent in the chromatograms of Hb from the control mice. The CID product ion spectra of β-Cys⁹³SOABP and β-Cys⁹³SO₂ABP adducts of Peptide D (LH*CDKLHVDPE) from mouse Hb (Figure S3A and S3B in Supporting Information) are identical to those observed for human Hb treated with HONH-ABP (Figure 5B1 and 6B1). The levels of β-Cys⁹³SOABP and β-Cys⁹³SO₂ABP and, ABP in ABP treated mice are summarized in Figure 9.

Figure 8.

Ion chromatograms of peptide D (LH*CDKLHVDPE) β⁹³Cys-SO-ABP, β⁹³Cys-SO₂-ABP adducts and free ABP recovered from Glu-C digest of Hb from mouse treated with ABP (40 mg/kg body weight) and untreated mouse.

Figure 9.

Ion counts of β⁹³Cys-SO-ABP, β⁹³Cys-SO₂-ABP adducts and free ABP recovered from Glu-C digests of Hb from mouse treated with ABP (40 mg/kg body weight) and human Hb (2nmol) in vitro modified with 3 molar excess of HONH-ABP (6 nmol). The Glu-C digestion was carried out with Hb (5 μg of Hb from in vivo mice treated with ABP and Hb of modified in vitro with HONH-ABP). The protein digest (100 ng) was used for the LC-MS/MS based measurement of ABP-Hb adducts and free ABP. Data are plotted as mean and standard deviation of ion counts of adducts recovered from five ABP-treated mice and three independent experiments conducted in vitro. The level of adducts were determined from the area of total counts of CID-MS² spectral data analyzed using orbitrap MS analyzer at 30,000 resolving power.

The relative abundance of mouse Hb-ABP sulfinamide adduct and free ABP in treated mice was comparable to those obtained from the incubation of a 3-fold molar excess of HONH-ABP with Hb in human RBC; however, the level of the Hb-ABP sulfonamide adduct relative to the Hb-ABP sulfinamide in mouse was 13-fold lower than level obtained from the in vitro studies with human RBC (Figure 9). Considering the average mouse Hb content (~268 mg mouse Hb/1.8 mL of total blood volume/30 g body weight, 4.1 μmol),⁶⁷ the dose of 40 mg ABP/kg body weight (1.2 mg/30 g body weight, 7.1 μmol) was equivalent to ~2-fold molar excess of ABP/mol mouse Hb tetramer. Although, the metabolism of ABP in vivo is a far more complex biological process than the in vitro reaction of HbO₂ with HONH-ABP, the amount of the ABP-Hb adduct formed in vivo is comparable to that obtained from the in vitro study.

Molecular Modeling

We sought to determine if variations in met-Hb formation among different HONH-AA and HONH-HAA substrates could be attributed to differences in their proximity of binding to the oxy-heme complex. The β chains of Hb from four crystal structures: 2DN1,⁶⁸ 4NI0,⁶⁹ 1YZI,⁷⁰ and 1GZX⁷¹ were examined to identify a suitable overall hemoglobin structure into which to dock HONH-ANL, HONH-ABP, HONH-AαC, HONH-PhIP, and HONH-MeIQx. All four β chains were close to identical in configuration, including the proximity of their heme pockets; RMSD values between the four B chains were 0.77–0.79Å. Given these slight variations, 2DN1.pdb was chosen as the docking target for all five AAs and HAAs, chiefly because 2DN1.pdb exhibits the HbO₂ in its relaxed state.^26,68 The heme pocket is narrow and there is very little room for the compounds addressed here to interact with the heme oxygen, a finding which is supported by our docking results. Figure 10A illustrates the highest-scoring docked pose of HONH-ABP with 2DN1.pdb. The measured distance between H-ONH-ABP and the oxy-heme moiety of 2DN1.pdb was predicted to be located far from the heme oxygen (10.22 Å). The measured distances (Table 3) between the H-ONH-substrates of other AA and HAAs were also predicted to be located too far from the heme oxygen to undergo co-oxidation and induce methemoglobinemia.⁷² To explore the flexibility of the heme pocket, the secondary structure of human hemoglobin (2DN1.pdb) was color coded according to a crystallographic temperature factor (B-factor) spectrum using PyMOL (Schrödinger, LLC), ranging from red (high B-factor) to blue (low B-factor), with white representing average values over the range of B-factors in this system. As shown in Figure 10B, helix D, the end of helix E, E-F loop, and the initial residues of helix F are more flexible regions according to experimental temperature factors. Therefore, we hypothesize that significant conformational changes are likely to take place in the flexible helix D near the heme pocket. Such changes in configuration would expose the heme pocket to interactions between the heme oxygen and small molecules.

Figure 10.

(A) The highest-scoring docked pose of HONH-ABP with HbO₂ β chain (2DN1.pdb). The H-ONH-ABP substrate and the oxy-heme moiety of 2DN1.pdb shows the predicted bond length between the H-ON-ABP substrate and the heme oxygen is 10.22 Å. (B) 2DN1.pdb colored according to crystallographic temperature factors (B-factors). Areas with high B-factors are colored red (hot) while areas with low B-factors are colored blue (cold). Flexible helix D is near the heme pocket.

Table 3.

Measured distances between the H-ONH-AA and H-ONH-HAA substrates and oxy-heme moiety of 2DN1.pdb in the highest-scoring docked poses of the five AAs and HAAs

HONH-AA or HONH-HAA	Measured distance between the H-ON-substrate and oxy-heme moiety of 2DN1.pdb in the highest-scoring docked pose
HONH-ANL	7.15 Å
HONH-ABP	10.22 Å
HONH-AαC	10.01 Å
HONH-PhIP	7.52 Å
HONH-MeIQx	9.38 Å

DISCUSSION

Hb adducts of AAs have been used in human biomonitoring studies.^39,73,74 The adducts have been measured indirectly, following acid or base treatment of Hb in vitro, to hydrolyze the presumed sulfinamide adducts formed at the β-Cys⁹³ of Hb, and liberate the parent amines, which are assayed by MS.^34,37 Because of their purported instability during proteolysis, mass spectral structural characterization of Hb arylsulfinamide adducts have not been reported,^35,36 although we recently characterized, by UPLC/MS², sulfinamide and sulfonamide peptide adducts of ABP, PhIP, and AαC formed at the Cys³⁴ of human albumin, following proteolysis with trypsin/chymotrypsin.^53,75 We have shown that the primary structure of Hb-peptide containing β-Cys⁹³ adducts of ABP, ANL, and AαC formed in vitro can be characterized by UPLC-MS/MS, employing bottom-up proteomic approaches following proteolytic digestion of the adducted Hb with trypsin, chymotrypsin, Lys-C or Glu-C. The Glu-C was an ideal protease to recover the peptide, ⁹²LH*CDKLHVDPE¹⁰² with minimal hydrolysis of the arylsulfinamide linkage. The pattern of CID fragmentation of arylsufinamide and arylsulfonamide adducts at β-Cys⁹³ of Hb was similar to the previously characterized sulfur-nitrogen linked albumin adducts of ABP and AαC.^53,75 As a proof of principle, we demonstrate that Hb sulfinamide and sulfonamide adducts of ABP formed in mice dosed with ABP (40 mg/kg) also can be detected by bottom-up proteomic approaches. At this dose, the levels of ABP adducts of Hb were comparable to the in vitro studies using a 3-fold molar excess of ABP/Hb tetramer. The identification of arylsulfinamide and arylsulfonamide adducts at ⁹²LH*CDKLHVDPE¹⁰² suggests that application of Glu-C and an UPLC-MS/MS based bottom-up proteomic approach may be achieved for the characterization of other reactive electrophiles that adduct to the β-Cys⁹³ of Hb. Indeed, data-independent-acquisition mass spectrometry has been employed to identify peptide site-specific modifications at the β-Cys⁹³ of Hb.⁷⁶ However, because of the challenges in the enrichment of the peptide adduct from the vast excess of non-modified protein digest, the biomonitoring of AαC in humans, like other arylamine Hb adducts, is expected to be more facile by acid or base hydrolysis of the sulfinamide linked adduct followed by the direct measurement of AαC.

We confirmed that the N-hydroxylated intermediates of the primary arylamines ANL and ABP induce methemoglobinemia and form sulfinamide adducts with Hb in RBC. Unexpectedly, the N-hydroxylated metabolite of AαC, a larger, linear tricyclic ring structure, was discovered to be more potent than HONH-ABP and HONH-ANL in inducing methemoglobinemia, and HONH-AαC also formed a sulfinamide adduct with Hb. However, HONH-PhIP and HONH-MeIQx, HAAs with angular tricyclic ring structures, did not interact with the oxy-heme complex and failed to induce met-Hb formation. Furthermore, the titratable β-Cys⁹³ residues were relatively unchanged following treatment of HbO₂ with HONH-PhIP or HONH-MeIQx. The binding proclivities of N-hydroxylated AAs to Hb are thought to depend upon their ability to undergo co-oxidation of HbO₂ to form met-Hb and the arylnitroso intermediates, which selectively bind to the β-Cys⁹³ residue of Hb (Scheme 1).^35,36 N-Hydroxy-AA and N-hydroxy-HAA undergo facile aerial oxidation to form their nitroso intermediates.^77,78 Moreover, we showed that HONH-ABP, HONH-PhIP, and HONH-AαC incubated with human serum albumin under similar buffer and pH conditions to those employed for HbO₂ resulted in the formation of AA and HAA sulfinamide adducts at the Cys³⁴ residue of albumin.⁷⁵ These findings signify that the nitroso intermediates were produced by aerial oxidation of the HONH-AA and HONH-HAA substrates. Thus, in addition to the poor reactivity of HONH-PhIP and HOHN-MeIQx with the oxy-heme complex of Hb, we surmise that the nitroso intermediates of PhIP and MeIQx may not reach and efficiently react with the β-Cys⁹³ Hb residue, which is on the distal side of the oxy-heme complex.

A study on the structure-activity relationship of arylamine binding to Hb conducted by Sabbioni reported that the binding of ortho-substituted alkyl or halogen derivatives of ANL to Hb was 5-fold or lower than the binding of ANL in the rat;⁷⁹ however, the reactivity of the N-hydroxylated ANL derivatives with HbO₂ was not examined. The decrease in the level of binding of ortho-substituted ANL derivatives to HbO₂ may have been attributed to lower amounts of the N-hydroxylated metabolites produced in vivo, or by hindered reactivity of the N-hydroxylated metabolites with the oxy-heme complex, resulting in diminished levels of met-Hb and the arylnitroso intermediates.⁷⁹ Another study reported that the binding of 2-aminofluorene, 2-napthylamine, and 2-aminopyrene to Hb in the rat were, respectively, 2.7%, 0.1%, and < 0.005% of the level of binding of 4-ABP.⁸⁰ Thus, certain arylamines with fused bicyclic or tricyclic ring structures can form sulfinamide adducts with Hb. The level of Hb adduct formation is dependent upon the extent of metabolism the arylamines to their N-hydroxy intermediates, the ability of the N-hydroxy metabolites to react with oxy-heme complex to form the arylnitroso species, and their ensuing reactivity with β-Cys⁹³ residue of Hb. The N-methyl substituents ortho to the hydroxylamine groups of HONH-PhIP and HONH-MeIQx may contribute to the poor reactivity of these HAA with HbO₂ in vitro. The low level of binding of PhIP and MeIQx to Hb in vitro is consistent with that observed in vivo in rodents and humans.^47,49

The β-Cys⁹³ is located between heme-linked proximal His92(F8) and Asp94(FG1), and near the functionally important α1 and β2 interface. In the native form of tetrameric Hb, the two β-Cys⁹³ residues are accessible to solvent and can react with low molecular weight electrophiles,²⁷ whereas the α-Cys¹⁰⁴ and β-Cys¹¹² are shielded and their reactivity with electrophiles is minor.^60,81 However, under high concentrations of PMB and prolonged incubation, a partial disassembly of the Hb subunits occurs, and the α-Cys¹⁰⁴ and β-Cys¹¹² become more accessible to solvent and can react with PMB and other electrophiles.^29,82,83 The reactivity of β-Cys⁹³ with sulfhydryl agents has been employed to separate the Hb tetramer into separate monomer chains.⁸⁴ The β-Cys⁹³ also binds low molecular weight alkylating agents such as styrene oxide,⁸⁵ benzene oxide,²⁵ nitric oxide,⁸⁶ and acrylate contact allergens present in paint.⁸⁷ The induced-fit docking program of MOE 2013.08 (Chemical Computing Group, Inc.) allowed only the active site residue side chains to be included in the final minimization stage of docking.^58,59 Based on these docking results (Figure 10 and Table 3), variations in sidechain configuration were too small to allow any of the five AA and HAA substrates to approach the oxy-heme complex closely enough to engage in hydrogen bonding and/or electron donation. Therefore, significant conformational changes are likely to occur in the flexible helical and loop regions near the heme pocket upon ligand binding, resulting in enlargement of the cavity to permit interactions between the heme oxygen and HONH-AA or HONH-AαC substrates. With this in mind, our future work will incorporate additional experimental studies, including redox potential determination, other electronic properties and chemical descriptors, to elucidate the chemical reactivity of the various N-hydroxy AA and HAA substrates with HbO₂ and the role that protein conformational changes may play in reactivity of these genotoxicants with HbO₂.^79,88

We had reported the crystal structure of the single stable Hb adduct of ABP, which confirmed that adduct formation had occurred at the β-Cys⁹³ chain of Hb.²⁸ The biphenyl group was wedged between the F and H helices close to the section where the two helices cross. The predominant interactions between adduct and protein occurred along the backbone of residues Leu141(H19) Ala142(H20) and the side chain of Lys144 (HC1) along one side of the phenyl rings, and residues Ala86(F2) - Cys93(F9) along the other side. The second phenyl ring was exposed at the surface of the heme pocket. It has been largely inferred that Hb adducts of other primary arylamines exist as sulfinamide linkages formed at the β-Cys⁹³ chain.^34,37 Our findings demonstrate that AαC also binds to the β-Cys⁹³ to form the sulfinamide linkage. This results reinforces the findings that certain HONH-substrates with considerable diversity of shape can induce structural changes in Hb which allow them to interact with the oxy-heme complex and burrow into mobile regions surrounding the heme pocket and adduct to the distal the β-Cys⁹³ group (Fig. 10). The high reactivity of HONH-AαC with HbO₂ suggests that the Hb sulfinamide adduct of AαC may be a promising biomarker of exposure to this HAA in humans.

ABBREVIATIONS

AA

arylamines

AαC

2-Amino-9H-pyrido[2,3-b]indole

HONH-AαC

2-hydroxyamino-9H-pyrido[2,3-b]indole

NO-AαC

2-nitroso-9H-pyrido[2,3-b]indole

NO₂-AαC

2-nitro-9H-pyrido[2,3-b]indole

ABP

4-aminobiphenyl

HONH-AA

N-hydroxylated derivatives of arylamines

HONH-ABP

4-hydroxyaminobiphenyl

HONH-ANL

phenylhydroxylamine

HAA

heterocyclic aromatic amines

Hb

hemoglobin

HbO₂

Oxyhemoglobin

RBC

red blood cell

PhIP

2-amino-1-methyl-6-phenylimidazo[4,5-b]pyridine

HONH-PhIP

2-hydroxyamino-1-methyl-6-phenylimidazo[4,5-b]pyridine

NO₂-PhIP

2-nitro-1-methyl-6-phenylimidazo[4,5-b]pyridine

MeIQx

2-amino-3,8-dimethylimidaxo[4,5-f]quinoxaline (MeIQx)

HONH-MeIQx

2-hydroxyamino-3,8-dimethylimidaxo[4,5-f]quinoxaline

NO₂-MeIQx

2-nitro-3,8-dimethylimidaxo[4,5-f]quinoxaline

IAM

iodoacetamide

SPE

solid phase extraction

THF

tetrahydrofuran

DDA

data dependent acquisition

MS/MS

tandem mass spectrometry

CID

collision induced dissociation

HCD

high-energy collision dissociation

RMSD

root-mean-square deviation

UPLC

ultraperformance liquid chromatography