Characterization of N1- and N6-Adenosine Adducts andN1-Inosine Adducts Formed by the Reaction of

Butadiene Monoxide with Adenosine: Evidence for theN1-Adenosine Adducts as Major Initial Products

Rebecca R. Selzer and Adnan A. Elfarra*

Department of Comparative Biosciences and Environmental Toxicology Center, University ofWisconsin, Madison, Wisconsin 53706

Received March 1, 1996X

1,3-Butadiene is a known human mutagen and possible human carcinogen; however, themolecular mechanisms of its activity are poorly understood. We have previously shown thatthe primary metabolite, butadiene monoxide (BM), reacts with guanosine to form N1-, N2-,and N7-guanosine adducts. In this study we characterize the reaction of BM with adenosine;ten adducts identified as diastereomeric pairs of N1-(1-hydroxy-3-buten-2-yl)adenosine, N1-(2-hydroxy-3-buten-1-yl)adenosine,N6-(1-hydroxy-3-buten-2-yl)adenosine,N6-(2-hydroxy-3-buten-1-yl)adenosine, and N1-(1-hydroxy-3-buten-2-yl)inosine are characterized. The N6-adenosineandN1-inosine adducts were characterized by their UV spectra, 1H NMR, FAB/MS, and stabilitystudies. The N6-adenosine and N1-inosine adducts were stable for up to 168 h at 37 °C inphosphate buffer (pH 7.4). The N1-adenosine adducts, which were unstable at pH 7.4 at 37 °C(half-life of 7 and 9.5 h for the two regioisomers), were characterized by their UV spectra andtheir ability to undergo the Dimroth rearrangement to yield the corresponding N6-adenosineadducts, or undergo deamination to yield the corresponding N1-inosine adducts. Upon thereaction of BM with adenosine in phosphate buffer (pH 7.4) at 37 °C, theN1-adenosine adductswere the first to be detected, with the N6-adenosine and N1-inosine adducts showing a lag information possibly due to the time needed for rearrangement/deamination. Reaction ofadenosine with an excess of BM in phosphate buffer (pH 7.4) at 37 °C, followed by extractionof the reaction mixture with ethyl ether to remove excess unreacted BM and incubation at 80°C for 1 h, resulted in complete conversion of N1-adenosine adducts to the corresponding N6-adenosine andN1-inosine adducts. Under these conditions, adduct formation exhibited pseudo-first-order kinetics, with the combinedN6-adenosine adducts being formed 3-fold more favorablythan the combined N1-inosine adducts. When incubations were carried out at lower BMconcentrations, the N6-adenosine adducts remained the major detectable adducts at allconcentrations. These results show that adenosine, in addition to guanosine, can lead tomultiple adducts when incubated with BM, and may be useful in development of biomarkersfor exposure to 1,3-butadiene. Characterization of the N1-adenosine adducts and theirrearrangement/deamination products may also contribute to the understanding of mutagenicand carcinogenic mechanisms of 1,3-butadiene.

Introduction

1,3-Butadiene is a petrochemical used in the industrialproduction of synthetic rubber and plastic and is alsofound in cigarette smoke, gasoline formulations, andautomobile exhaust. Rats and mice exposed to 1,3-butadiene form tumors at multiple sites (1, 2). Recentepidemiological studies suggest that current levels ofoccupational exposure may not be adequate to protectworkers from the mutagenic/carcinogenic effects of 1,3-butadiene. One such study demonstrates an increasedincidence of various lymphatic and hematopoietic cancersin industrial workers exposed to 1,3-butadiene (3), whilea second study reports an increase in the hprt mutantfrequency in peripheral lymphocytes of nonsmokingworkers exposed to butadiene (4).

Butadiene monoxide (BM, 3,4-epoxy-1-butene)1 is amajor metabolite of 1,3-butadiene in multiple in vitro andin vivo systems (5-9). Rat liver microsomal incubationshave demonstrated formation of both enantiomers of BMin similar ratios (5). BM has been postulated to beresponsible for 1,3-butadiene mutagenicity in the S9activated Salmonella typhimurium mutagenicity assay(10). BM and 1,3-butadiene were both found to increasea broad range of mutations in vivo in exposed mice (11).We have previously described the in vitro reaction of

BM with guanosine to form eight products which werecharacterized as diastereomeric pairs of N7-(1-hydroxy-3-buten-2-yl)guanosine, N7-(2-hydroxy-3-buten-1-yl)gua-nosine,N1-(1-hydroxy-3-buten-2-yl)guanosine, and N2-(1-hydroxy-3-buten-2-yl)guanosine (12). The N7-adducts

* Corresponding author: Dr. Adnan A. Elfarra, Department ofComparative Biosciences, University of Wisconsin School of VeterinaryMedicine, 2015 Linden Dr. W., Madison, WI 53706; Telephone: (608)262-6518; FAX: (608) 263-3926; Internet: elfarraa@svm.vetmed.wisc.edu.

X Abstract published in Advance ACS Abstracts, June 15, 1996.

1 Abbreviations: BM, butadiene monoxide; ACN, acetonitrile; FAB/MS, fast atom bombardment mass spectrometry; A-1 and A-3 (alsoreferred to as I-3 and I-4, respectively, when they are made frominosine), diastereomeric N1-(1-hydroxy-3-buten-2-yl)inosine; A-2 andA-6, diastereomericN6-(1-hydroxy-3-buten-2-yl)adenosine; A-4 and A-5,diastereomeric N6-(2-hydroxy-3-buten-1-yl)adenosine; A-7, A-8, A-9,and A-10, N1-adenosine-BM adducts; I-1 and I-2, N1-(2-hydroxy-3-buten-1-yl)inosine; 3-NBA, 3-nitrobenzyl alcohol.

875Chem. Res. Toxicol. 1996, 9, 875-881

S0893-228x(96)00039-2 CCC: $12.00 © 1996 American Chemical Society

were the major products, being formed to a 10-foldgreater amount than theN1- andN2-adducts over a rangeof BM concentrations.1,3-Butadiene and BM are capable of inducing muta-

tions at both GC and AT base pairs (11); thus bothguanine and adenine adducts may be involved in buta-diene-induced mutation and carcinogenesis. This studyexamines the reaction of BM with adenosine and identi-fies ten products. Six of the products were characterizedas diastereomeric pairs of N6-(1-hydroxy-3-buten-2-yl)-adenosine,N6-(2-hydroxy-3-buten-1-yl)adenosine, andN1-(1-hydroxy-3-buten-2-yl)inosine. The four remaining prod-ucts were identified as diastereomeric pairs of N1-(1-hydroxy-3-buten-2-yl)adenosine and N1-(2-hydroxy-3-buten-1-yl)adenosine based upon their instability andtheir conversion toN6-adenosine andN1-inosine adducts.The pseudo-first-order kinetic rate constants, stability,and effect of BM concentration on adduct formation wereexamined. Preliminary reports of these results have beenpresented (13, 14).

Experimental Procedures

Materials. Racemic BM, deuterium oxide, [D6]Me2SO, andtrifluoroacetic acid were purchased from Aldrich Chemical Co.(Milwaukee, WI). Adenosine and inosine were purchased fromSigma Chemical Co. (St. Louis, MO). HPLC grade acetonitrile(ACN) was purchased from EM Science (Gibbstown, NJ). NMRsupplies were obtained fromWilmad Glass Co. (Buena, NJ). Allother chemicals were of the highest grade commercially avail-able. Caution: BM is a known mutagen and carcinogen inlaboratory animals and should be handled using proper safetymeasures.Synthesis of BM-N6-Adenosine and BM-N1-Inosine

Adducts. Either adenosine (53.4 mg; 0.2 mmol) or inosine (53.6mg; 0.2 mmol) was dissolved in 3 mL of H2O and heated at 50°C in a 4 mL vial with a screw top Teflon coated cap, untildissolved. An excess of BM (0.161 mL; 2 mmol) was added, andthe reaction was incubated at 50 °C in a Dubnoff shaking waterbath for 24 h. At 24 h, the reaction mixtures were extractedfour times with 4 volumes of ethyl ether to remove unreactedBM. With both adenosine and inosine, four major peaks (A-2,A-4, A-5, A-6 or I-1, I-2, I-3, I-4), in addition to starting material,were separated by HPLC. Minor amounts of the inosine adductsA-1 and A-3, which correspond to adducts I-3 and I-4 whenformed from inosine directly, were formed in the adenosinereaction; however, they proved to be too minor to fractionate.Efforts to obtain the unstable N1-adenosine adducts (A-7, A-8,A-9, and A-10) for characterization were unsuccessful.HPLC Analysis and Purification of BM-N6-Adenosine

and BM-N1-Inosine Adducts. HPLC separations and puri-fication were performed as previously described (12) by reverse-phase HPLC chromatography. Briefly, analytical separationswere performed with a 20 µL injection volume on a Beckmanreverse-phase analytical column using a gradient controlledHPLC system equipped with a diode array detector and UVdetection at 260 nm. HPLC analysis involved use of a lineargradient program starting at 10 min from 0% to 100% pump Bover 8 min [pump A 1% ACN (pH 2.5); pump B 10% ACN (pH2.5)], at a flow rate of 1 mL/min. Semipreparative HPLC wasperformed with a similar gradient system and solvents using a100 µL injection and a rate of 3 mL/min. Separation of syntheticmixtures revealed four products from the adenosine reactionwhose retention times on a typical analytical chromatogramwere 28.24, 29.35, 29.62, and 31.03 min. These peaks werenamed A-2, A-4, A-5, and A-6, respectively. The inosine reactionmix revealed four products whose retention times on a typicalchromatogram were 26.35, 27.01, 27.86, and 28.84 min. Thesepeaks were named I-1, I-2, I-3, and I-4, respectively. All adductswere fractionated 2-4 times by semipreparative HPLC untilthey were isomerically pure by analytical HPLC. The purified

adducts were then used to obtain standard curves. The limitsof detection using the analytical HPLC method were 0.25 µg/mL for all N6-adducts and 0.5 µg/mL for the inosine adducts (rwere greater than 0.999).Identification of BM-N6-Adenosine and BM-N1-Inosine

Adducts. UV spectra at pH 2.5 were performed using aBeckman diode array detector. Positive ion fast atom bombard-ment mass spectra (FAB/MS) were performed using a KratosMS-50 ultrahigh-resolution mass spectrometer fitted with aKratos DS-55 data system (Manchester, U.K.). An Ion TechFAB gun utilizing xenon as the FAB gas was used with a directinsertion FAB probe. Spectra ranged fromm/z 100 to 400 usinga glycerol or 3-nitrobenzyl alcohol (3-NBA) matrix. All spectraare matrix subtracted. Proton NMR spectra were performedfor each adduct both in D2O at 5 °C and in [D6]Me2SO at roomtemperature. Homonuclear decoupling experiments were per-formed on all samples to confirm proton assignments. Chemicalshifts are reported in ppm with the H2O peak as an internalstandard.To characterize the stability of the N6-adenosine and N1-

inosine adducts under alkaline conditions, a mixture of theseadducts was lyophilized to dryness and dissolved in 1 M KOHand incubated at 95 °C for 2 h. Aliquots were withdrawn every30 min, pH adjusted to neutrality with HCl, and immediatelyanalyzed by HPLC for disappearance of the parent adductpeaks.The Reaction of BM with Adenosine under Physiologi-

cal Conditions. Adduct formation and stability under in vitrophysiologic conditions was studied as previously described (12)with some changes. Adenosine (10 mM) and BM (10, 25, 50,and 500 mM) were reacted in 3 mL of 100 mM phosphatereaction buffer (pH 7.4) containing 100 mM potassium chloridein a shaking water bath at 37 °C. The reactions were termi-nated after 60 min by extraction of unreacted BM with ethylether. Samples were analyzed for the formation of the adductsby HPLC. In order to convert the unstable HPLC peakscorresponding to N1-adenosine adducts into the more stableHPLC peaks corresponding to N6-adenosine and N1-inosineadducts for analysis, samples were incubated for 60 min at 80°C. These conditions were found to be the most efficient forcomplete conversion as evidenced by the complete disappearanceof the N1-adenosine adduct peaks and the increase in N6-adenosine and N1-inosine peaks in the absence of BM.In order to calculate the pseudo-first-order rate constants,

adenosine (3.1 or 9.6 mM) and BM (750 mM) were reacted inphosphate buffer (pH 7.4) for 4 h in a 37 °C shaking water bath.Samples were withdrawn at timed intervals, extracted withethyl ether, and incubated at 80 °C to allow conversion ofunstable N1-adenosine adducts to stable N6-adenosine and N1-inosine adducts. The concentration of BM used should leave alarge excess of BM remaining at the end of the experiment,which allows calculation of pseudo-first-order kinetic rateconstants. Analysis for N6-adenosine and N1-inosine adductswas performed by HPLC, as described above.Adduct stability under in vitro physiologic conditions was

examined with purified adduct (A-1, A-2, A-3, A-4, A-5, and A-6)dissolved (0.3 mM) in 100 mM phosphate buffer (pH 7.4)containing 100 mM potassium chloride at 37 °C. Samples werewithdrawn at 24 h intervals and analyzed by HPLC fordisappearance of the parent adduct peak. Stability of A-7, A-8,A-9, and A-10 was determined by similarly incubating an ethylether-extracted mixture of all ten adducts resulting from thereaction of adenosine and BM in phosphate buffer (pH 7.4) at37 °C for 4 h. Samples were withdrawn every 2 h for 24 h andanalyzed for the disappearance of the parent adduct by HPLC.

Results

Reaction of adenosine with BM under in vitro physi-ologic conditions for 24 h resulted in the formation of sixproduct peaks as detected by HPLC. These productswere designated A-1, A-2, A-3, A-4, A-5, and A-6 in orderof their retention times. Shorter incubations (4 h)

876 Chem. Res. Toxicol., Vol. 9, No. 5, 1996 Selzer and Elfarra

resulted in the detection of four additional, earlier elutingpeaks which were designated A-7, A-8, A-9, and A-10(Figure 1). A-7, A-8, A-9, and A-10 were characterizedasN1-adenosine adducts based upon their maximum UVabsorbance (260 nm), which is consistent with themaxima of other N1-adenosine adducts (15), and theirchemical instability at pH 7.4 at 37 °C leading toconversion to the more stable A-1 to A-6 adducts. Previ-ous studies have also found N1-adenosine adducts to beunstable due to the Dimroth rearrangement (16).A-1, A-2, A-3, A-4, A-5, and A-6 were structurally

characterized by analysis of their 1H NMR and FAB/MSspectra. The assignment of protons to NMR signals wasachieved by comparing chemical shifts, multiplicities,integration ratios, and decoupling experiments with thespectra of the characterized BM-guanosine adducts (12),with published spectra of adenosine (17), and with3-butene-1,2-diol (data not shown). Products A-2 andA-6, A-4 and A-5, and A-1 and A-3 have nearly identical1H NMR spectra, respectively, suggesting they are dia-stereomeric pairs. Spectra representing each regioisomerare given in Figure 2. Chemical shifts and J values forall adducts are given in Table 1.A-4 and A-5 were determined to be diastereomers of

N6-(2-hydroxy-3-buten-1-yl)adenosine based on the fol-lowing information. With a maxima of 266 nm, the UVspectra of A-4 and A-5 were consistent with the spectralshape and maxima of other N6-substituted adenosineadducts (15). NMR’s performed in [D6]Me2SO showedintegration of only one of two possible N6 protons (datanot shown). The BM moiety was determined to beattached to the N6 of adenosine at the terminal carbonbased on three pieces of evidence. H8 (Figure 3) is shiftedfurther upfield (4.41, 4.42 ppm) than in A-2 or A-6 (4.65,4.67 ppm), suggesting it is adjacent to the hydroxyl grouprather than the more deshielding N6 of adenosine. Qianand Dipple (18) also found the N6-substituent to be moredeshielding than the hydroxyl moiety in their character-ization of the N6-adenosine adducts of styrene oxide. Inaddition, that study also found little difference in the

â-proton signals of styrene between the regioisomers,which is similar to what is seen with the H12’s in thisstudy (Table 1). H8 does not show any indication ofcoupling with the N6 proton in the [D6]Me2SO spectra(data not shown); however, any coupling between the twoH12 protons and the N6 proton is obscured by the H2Opeak and overlap with H11. Finally, the FAB/MS of A-4and A-5 exhibit anm/z 148 ion, corresponding to adenineplus a methyl group. Specific assignment of protons fromthe NMR’s in D2O were confirmed by homonucleardecoupling experiments. Saturation of H7 identified H3

and H9, while saturation of H8 identified H4 and H12 andsaturation of H10 identified H9 and H11.The FAB/MS fragmentation patterns of A-4 and A-5

were very similar and are consistent with the proposedstructure. The expected molecular ion of m/z 338 (M +1) was present in both spectra (Figure 4B), as was them/z 206 fragment which represents the loss of the sugarat the labile glycosidic linkage. As mentioned above,there was anm/z 148 fragment which corresponds to anadenine moiety plus a methyl group. This fragmentsuggests the loss of three of the BM carbons, which isconsistent with the substitution being on the terminalcarbon of the BM moiety.A-2 and A-6 were determined to be diastereomers of

N6-(1-hydroxy-3-buten-2-yl)adenosine. With a maximumof 266 nm, the UV spectra of A-2 and A-6 were consistentwith the spectral shape and maxima of the A-4 and A-5

Figure 1. HPLC chromatogram of the ten products formed byreaction of racemic BM with adenosine. Peaks A-1 and A-3 wereidentified as diastereomeric pairs of N1-(1-hydroxy-3-buten-2-yl)inosine, A-2 and A-6 as diastereomers of N6-(1-hydroxy-3-buten-2-yl)adenosine, A-4 and A-5 as diastereomers of N6-(2-hydroxy-3-buten-1-yl)adenosine, and A-7, A-8, A-9, and A-10 asdiastereomeric pairs of N1-(1-hydroxy-3-buten-2-yl)adenosineand N1-(2-hydroxy-3-buten-1-yl)adenosine. The retention timesof the minor peaks eluting between A-7 and A-1 are similar tothe retention times of reference I-1 and I-2; however, the UVscans of these peaks appear not to correspond with the inosineadducts I-1 and I-2.

Figure 2. 1H NMR at 5 °C for (A) N6-(2-hydroxy-3-buten-1-yl)adenosine (A-4, 400 MHz), (B) N6-(1-hydroxy-3-buten-2-yl)-adenosine (A-2, 400 MHz), and (C) N1-(1-hydroxy-3-buten-2-yl)inosine (A-1, 500 MHz). Peak numbers correspond toassignments in Figure 3.

Butadiene Monoxide-Adenosine Adducts Chem. Res. Toxicol., Vol. 9, No. 5, 1996 877

regioisomers, and other N6-substituted adenosines (15).As with A-4 and A-5, NMR’s performed in [D6]Me2SOshowed integration of only a single N6 proton (data notshown). The BM moiety was determined to be attachedto the N6 of adenosine at the carbon adjacent to thedouble bond. H8 is shifted downfield of the H8 in A-4 andA-5 (Figure 2). The FAB/MS has anm/z 306 ion (Figure4A) corresponding to the molecular ion minus a methanolgroup, which is consistent with an adduct attached tothe carbon atom next to the double bond of BM. Homo-nuclear decoupling experiments confirmed specific as-signments of protons. Saturation of H4 identified H5 andH6; saturation of H7/8 identified H3, H9, H4, and H12; andsaturation of H10 identified H9 and H11.

Tab

le1.

400an

d500MHz

1 HNMRDataforAden

osinean

dInosineAdductsof

BM

inD2O

(JValues

Rep

ortedin

Hz,Chem

ical

Shifts

Rep

ortedin

ppm)

H1a

H2

H3

H4

H5

H6

H7

H8

H9

H10

H11

H12

A-2

8.34

8.21

6.00

5.86

5.23

5.20

4.65

4.65

4.31

4.16

3.80/3.72c

3.80/3.69c

J)5.2

J)16.8,11.1,4.2

J)14.8

J)10.0

J)NAb

J)NA

J)4.0,4.8

J)3.2,NA

J)12.8,N

A/3.8

dJ

)11.4,N

A/7.2

d

A-4

8.23

8.18

6.00

5.91

5.3

5.21

4.75

4.42

4.37

4.25

3.86/3.78c

3.70/3.54c

J)6.40

J)17.0,10.6,6.2

J)17.2

J)10.8

J)6.0,5.6

J)6.0,NA

J)5.2,3.2

J)2.8

J)13.0,2.6/3.4

dJ

)14.8,N

A

A-5

8.29

8.21

6.00

5.88

5.30

5.20

4.69

4.41

4.35

4.21

3.83/3.76c

3.68/3.54d

J)5.60

J)17.0,10.6,6.2

J)17.2

J)10.4

J)5.6,5.6

J)6.0,5.8

J)5.2,3.6

J)3.2,3.2

J)12.8,2.8/3.6

dJ

)15.0,N

A

A-6

8.35

8.22

6.01

5.87

5.24

5.22

4.67

4.67

4.33

4.17

3.81/3.73c

3.81/3.71c

J)5.6

J)16.7,10.6,4.6

J)17.2

J)10.4

J)NA

J)NA

J)4.8,4.4

J)3.2,NA

J)12.8,3.6/2.8

dJ

)11.2,4.4/6.8

d

I-1

8.42

8.32

6.06

5.96

5.33

5.28

4.74

4.51

4.43

4.27

3.92/3.84c

4.35/4.02c

J)5.5

J)16.5,10.6,5.9

J)17.9

J)10.5

J)5.5,5.0

J)NA

J)4.5,4.5

J)3.5,NA

J)12.5,1.75/3.5d

J)14.0,3.5/8.5

d

I-2

8.36

8.31

6.05

5.96

5.34

5.28

4.74

4.52

4.43

4.27

3.92/3.84c

4.36/4.03c

J)5.5

J)17.0,10.5,6.0

J)17.0

J)10.5

J)5.5,5.0

J)NA

J)4.0,4.5

J)3.5,3.0

J)13.0,2.5/3.5

dJ

)14.0,3.5/8.5

d

I-3

8.43

8.36

6.10

6.13

5.36

5.48

4.80

5.61

4.46

4.29

3.92/3.85c

4.12/4.07c

J)6.0

J)17.0,11.2,5.7

J)17.5

J)11.0

J)5.5,5.5

J)NA

J)4.5,4.0

J)3.0,NA

J)12.5,3.5/2.5

dJ

)12.2,7.5/7.0

d

I-4

8.44

8.36

6.11

6.13

5.35

5.48

4.80

5.61

4.45

4.29

3.92/3.85c

4.13/4.07c

J)5.0

J)17.0,11.0,6.0

J)17.5

J)11.0

J)5.5,5.0

J)NA

J)NA

J)NA

J)12.7NA

J)11.7,7.5/4.5

d

aProtonnumberingrefers

toassignmentsgivenin

Figure

3.Multiplicities

canbe

foundon

thespectra,Figure

2.bNA

)Jvalues

not

availablefrom

spectra.

cTwochem

ical

shiftsaregiven

correspondingto

thetwoprotonsofthemethylenegroup.

dTwoJvalues

aregivencorrespondingto

thetwoprotonsofthemethylenegroup.

Figure 3. Reaction of BM with adenosine forms two N1-adenosine regioisomers which undergo the Dimroth rearrange-ment to form N6-adenosine adducts or deaminate to form asingle N1-inosine regioisomer. Proton numbering is based onchemical shift data from the first spectrum solved (A-4).

Figure 4. FAB/MS of the diastereomers of (A) N6-(1-hydroxy-3-buten-2-yl)adenosine (A-2; 3-NBA matrix), (B) N6-(2-hydroxy-3-buten-1-yl)adenosine (A-5; 3-NBA matrix), and (C) N1-(1-hydroxy-3-buten-2-yl)inosine (A-3; glycerol matrix).

878 Chem. Res. Toxicol., Vol. 9, No. 5, 1996 Selzer and Elfarra

The FAB/MS fragmentation patterns were again verysimilar for A-2 and A-6 and are consistent with theproposed structure (Figures 3 and 4). The expectedmolecular ion of m/z 338 (M + 1) was present in bothspectra, as was, again, the m/z 206 fragment represent-ing the loss of the sugar moiety. In addition, a smallfragment atm/z 306 was evident as described above. Anm/z 136 ion was also observed, which is consistent withthe loss of both the BM and sugar moieties, leaving anadenine fragment.I-3 and I-4, and thus A-1 and A-3, were identified as

diastereomeric pairs of N1-(1-hydroxy-3-buten-2-yl)-inosine, while the remaining two products from thereaction of BM with inosine, I-1 and I-2, were identifiedas diastereomeric pairs of N1-(2-hydroxy-3-buten-1-yl)-inosine. The UV spectra of A-1 and A-3 had a differentshape than theN1- andN6-adenosine adducts. However,the spectra shape and maxima match those of reportedN1-inosine adducts (15) and products formed in thereaction of BM with inosine (I-1, I-2, I-3, and I-4).Additional evidence that A-1 and A-3, formed in thereaction of BM with adenosine, were inosine adducts wastheir coelution with I-3 and I-4, two of four productsproduced by the reaction of BM with inosine. 1H NMR’sperformed in D2O identified I-3 and I-4 as diastereomersofN1-(1-hydroxy-3-buten-2-yl)inosine based on the upfieldshift of H8 of over 1 ppm (to 5.61 ppm) from the shifts ofN1-(2-hydroxy-3-buten-1-yl)inosine (I-1 and I-2; 4.51 ppm),which were also isolated from the reaction of BM withinosine (Table 1). This large downfield shift suggests theallylic carbon of BM is attached to the N1-nitrogen ofinosine which is expected to be strongly deshielding dueto the adjacent carbonyl group. Homonuclear decouplingexperiments confirmed the specific assignment of protons.Saturation of H4 identified H5 and H6; saturation of H8

identified H4 and H12; and saturation of H10 identifiedH9 and H11.The FAB/MS fragmentation patterns of I-3 and I-4

(Figure 4C) were consistent with BM adducts of inosinebased on the molecular ion m/z 339 (M + 1). A secondmajor peak corresponding to the loss of the sugar wasalso present at m/z 207. In addition, the fragment atm/z 115 may correspond to the sugar moiety minus H2O,and the m/z 137 may correspond to the hypoxanthinemoiety plus three protons. The FAB/MS’s for I-1 and I-2are very similar to I-3 and I-4 and contain the major ionsm/z 339, 207, 137, and 115 (data not shown).Incubation of all ten BMsadenosine products in 1 M

KOH at 95 °C resulted in disappearance of all adductsexcept theN6-adenosine adducts (A-2, A-4, A-5, A-6; datanot shown). All other adducts were completely degradedwithin 30 min, while theN6-adenosine adducts remainedpast the 2 h time point. This is additional confirmationthat adducts A-2, A-4, A-5, and A-6 are substituted onan exocyclic amino group of a purine. Alkylation at anyother nucleophilic position of adenosine results in deg-radation under these conditions (19). We have seensimilar results with the BM-guanosine adducts whereN1-guanosine adducts were unstable while the exocyclicN2-guanosine adducts were stable under these alkalineconditions (12).Adenosine (10 mM) was reacted with varying amounts

of BM (10, 25, 50, and 500 mM) for 1 h in 100 mMphosphate buffer (pH 7.4) at 37 °C, and aliquots wereheated at 80 °C for 1 h to convert N1-adenosine adductsto stable N6-adenosine and N1-inosine adducts. All sixadducts (A-1, A-2, A-3, A-4, A-5, and A-6) were detected

at BM concentrations ranging from 2.5-fold to 50-foldover the adenosine concentration and were formed insimilar ratios over all the BM concentrations (Figure 5).At concentrations of BM lower than 25 mM, the inosineadducts were not detectable by this HPLC methodbecause the amounts of adducts were below the limits ofdetection. As indicated earlier, N1-inosine adducts ex-hibited limits of detection that were 2-fold higher thanthe N6-adenosine adducts. Three N6-adenosine adducts(A-2, A-4, A-5) were detected when the BM concentrationwas equimolar with adenosine. Both diastereomers ofN6-(2-hydroxy-3-buten-1-yl)adenosine (A-4 and A-5) weredetectable down to a 0.5-fold BM (5 mM) concentration(data not shown). Concentrations of BM lower than 5mM did not lead to detection of any of the adenosineadducts due to the limits of detection of this method.These data indicate that adenosine adducts can bedetected at BM concentrations equivalent to (10 mM),and in the case of A-4 and A-5 lower than (5 mM), theconcentrations we have previously reported for the N7-and N2-guanosine-BM adducts (12).When adenosine (9.6 mM) was reacted with an unlim-

iting BM concentration (750 mM) at 37 °C (pH 7.4) andaliquots were analyzed directly, theN1-adenosine adductsshowed a rapid initial formation with a plateau at latertime points (Figure 6A) while the N6-adenosine adductsshowed an initial lag in formation at earlier time pointsand a rapid increase starting at 60 min (Figure 6B).These data suggest that the peaks A-7, A-8, A-9, and A-10were serving as precursors for peaks A-2, A-4, A-5, andA-6. When adenosine (3.1 and 9.6 mM) was reacted withBM under the same conditions with the addition of the80 °C heat treatment step, the six stable adducts (A-1,A-2, A-3, A-4, A-5, and A-6) exhibited linear timedependency from 0 to 120 min (Figure 6C,D). Evidencefor the reaction being carried out under pseudo-first-orderkinetics was demonstrated by the amount of adductsformed at the 9.6 mM adenosine concentration being3-fold that formed at 3.1 mM adenosine, and the rateconstant calculated at each concentration was constant.

Figure 5. Effect of BM concentration on the formation of fourN6-adenosine and two N1-inosine adducts from the reaction ofBM with adenosine (10 mM) for 60 min at 37 °C in phosphatebuffer (pH 7.4). Values represent the means ( SD of the resultsobtained from three experiments.

Butadiene Monoxide-Adenosine Adducts Chem. Res. Toxicol., Vol. 9, No. 5, 1996 879

The calculated rate constants during this period for theformation of A-1, A-2, A-3, A-4, A-5, and A-6, were 3.75× 10-3, 2.95 × 10-3, 4.45 × 10-3, 6.75 × 10-3, 1.26 × 10-2,and 3.80 × 10-3, respectively. Due to the Dimrothrearrangement (Figure 3), rates calculated for A-4 andA-5 may be indicative of the rates of formation of thecorresponding N1-adenosine adduct precursors. Rates ofA-2 and A-6 formation are decreased by the competingdeamination reaction forming A-1 and A-3.Studies of the stability of all ten adducts at 37 °C in

100 mM phosphate buffer (pH 7.4) suggest that the N6-adenosine (A-2, A-4, A-5, A-6) and N1-inosine (I-3, I-4)adducts are stable for up to 7 days without decomposition(Figure 7). TheN1-adenosine adducts (A-7, A-8, A-9, andA-10), however, are unstable under these in vitro physi-ologic conditions and rearrange into theN6-adenosine andN1-inosine adducts with a half-life of 7 and 9.5 h for thetwo regioisomers, respectively.

Discussion

This study demonstrates that the reaction of racemicBM with adenosine under in vitro physiologic conditionsproduces ten products which have been characterized asdiastereomeric pairs of N6-(1-hydroxy-3-buten-2-yl)ad-

enosine (A-2 and A-6), N6-(2-hydroxy-3-buten-1-yl)ad-enosine (A-4 and A-5),N1-(1-hydroxy-3-buten-2-yl)inosine(A-1 and A-3), andN1-(1-hydroxy-3-buten-2-yl)adenosineandN1-(2-hydroxy-3-buten-1-yl)adenosine (A-7, A-8, A-9,and A-10). When BM has been extracted out of themixture, incubation at 80 °C for 60 min rearranges N1-substituted adenosines (A-7, A-8, A-9, A-10) to N6-substituted adenosines (A-2, A-4, A-5, A-6) through thewell-known Dimroth rearrangement. In addition, theheat treatment also accelerates the competing deamina-tion reaction of the N1-adenosine adducts containing theterminal hydroxyl group to N1-inosine adducts. N6-(2-Hydroxy-3-buten-1-yl)adenosine (A-4 and A-5) was themajor N6-adenosine adduct formed both before, probablydue to Dimroth rearrangement occurring at 37 °C, andafter the forced Dimroth rearrangement. The sum of theN6-(1-hydroxy-3-buten-2-yl)adenosine andN1-(1-hydroxy-3-buten-2-yl)inosine adduct concentrations, however, wasequal to the sum of the N6-(2-hydroxy-3-buten-1-yl)-adenosine adduct concentrations, suggesting the N1-adenosine regioisomers are formed in similar amounts.Fujii et al. (16) demonstrated thatN1-(1-hydroxyalkyl)-

adenine derivatives could undergo the Dimroth rear-rangement at neutral pH and elevated temperature andalso showed that the terminal hydroxyl group assists inhydrolytic deamination of the substituted adenosine. Thisprocess explains the instability of the N1-adenosineadducts presented in this study, and the appearance ofthe single terminal hydroxyl regioisomer of the BM-inosine adducts. Our in vitro physiologic reactions showthat the N1-adenosine adducts are formed most rapidlyinitially, then are capable of being converted to the N6-adenosine and N1-inosine adducts at neutral pH and atemperature as low as 37 °C.When adenosine was reacted with unlimiting concen-

trations of BM, we were able to calculate pseudo-first-order rate constants. The similar rate constants for theN6-(1-hydroxy-3-buten-2-yl)adenosine andN1-(1-hydroxy-3-buten-2-yl)inosine regioisomers were to be expected, asFujii et al. (16) found the rate constants for deaminationand rearrangement were comparable to each other at pH6.0 and the relative rate of the deamination increases asthe pH decreases. These rates were also found to besimilar to those found with the N1- and N2-adducts of

Figure 6. Rates of formation of N1-adenosine adducts (A) and N6-adenosine adducts (B) at 37 °C in phosphate buffer (pH 7.4)following direct analysis, and rates of formation of N6-adenosine andN1-inosine adducts at 37 °C in phosphate buffer (pH 7.4) followingthe 80 °C heat treatment step (C and D). To guarantee pseudo-first-order kinetic conditions, the BM concentration (750 mM) wasmuch higher than the adenosine concentrations (A, B, D: 9.6 mM; C: 3.1 mM).

Figure 7. Stability of BM-adenosine adducts in 100 mMphosphate buffer containing 100 mM KCl (pH 7.4) at 37 °C.

880 Chem. Res. Toxicol., Vol. 9, No. 5, 1996 Selzer and Elfarra

guanosine, while the rate constants for N6-(2-hydroxy-3-buten-1-yl)adenosine are only slightly lower than therate constants for the N7-adducts of guanosine (12).These results suggest the N6-adenosine adducts may beuseful in the development of a biomarker for exposureto 1,3-butadiene on the basis that they are formed at highrates in the reaction of BM and adenosine and becauseof their increased stability over theN7-guanosine adducts.When adenosine was reacted with varying concentra-

tions of BM, a roughly linear formation of adducts wasfound over the range tested, as was found with thepreviously reported N1-, N2-, and N7-guanosine adducts(12). The ratios of adducts formed over the range of BMtested were consistent with the combined N6-adenosineadducts formed 3-fold more favorably than the combinedN1-inosine adducts. This suggests there is no thresholdBM concentration for adenosine adduct formation nearthe range tested and the adducts may be formed at evenlower concentrations of BM even though this would bebelow the limits of detection of our assay.We found the N6-adenosine adducts and the N1-(1-

hydroxy-3-buten-2-yl)inosine adducts to be completelystable for up to 1 week in phosphate buffer (pH 7.4) at37 °C, suggesting they could be important mutagenicprecursors. N1-Adenosine adducts, however, appear tobe the initial products and convert to N6-adenosineadducts with half-lives of 7 and 9.5 h at in vitrophysiologic conditions. This suggests that the N1-ad-enosine adducts may exist long enough to also bepotentially mutagenic. Because 1,3-butadiene and BMare capable of inducing mutations at AT base pairs (11),the formation of all possible adenosine adducts may beof toxicological significance. Recently, Koivisto et al. (20)synthesized N6-deoxyadenosine and N6-deoxyadenosine3′-monophosphate adducts of BM and used 32P-postla-beling experiments to identifyN6-adenine adducts in calfthymus DNA. NeitherN1-deoxyadenosine norN1-deoxy-inosine adducts were identified in that study, althoughseveral unidentified peaks were observed under neutralaqueous conditions.In conclusion, this study presents strong evidence for

N1-adenosine adducts being initial products of the reac-tion of BM and adenosine. These adducts are relativelyunstable and rearrange to N6-adenosine adducts ordeaminate to N1-inosine adducts. While one N6-adeno-sine regioisomer was formed preferentially to the otherN6-adenosine regioisomer and theN1-inosine regioisomer,all of these adducts were extremely stable, and the N1-adenosine adducts were also long-lived enough to be ofmutagenic significance. Identification of these adductsand characterization of their physicochemical propertiesmay contribute to a better understanding of the molec-ular mechanisms of 1,3-butadiene-induced carcinogenic-ity.

Acknowledgment. This research was supported byNIH Grant ES06841 from the National Institute ofEnvironmental Health Sciences. R.R.S. was supportedby a National Science Foundation Graduate ResearchFellowship. NMR spectra were determined at the Na-tional Magnetic Resonance Facility at Madison, whichis supported in part by NIH Grant RR02301.

References(1) Melnick, R. L., Huff, J., Chou, B. J., and Miller, R. A. (1990)

Carcinogenicity of 1,3-butadiene in C57BL/6 × CH3 F1 mice atlow exposure concentrations. Cancer Res. 50, 6592-6599.

(2) Owen, P. E., Glaister, J. R., Gaunt, I. F., and Pullinger, D. H.(1987) Inhalation toxicity studies with 1,3-butadiene. 3. Two yeartoxicity/carcinogenicity study in rats. Am. Ind. Hyg. Assoc. J. 48,407-413.

(3) Santos-Burgoa, C., Matanoski, G. M., Zeger, S., and Schwartz,L. (1992) Lymphohematopoietic cancer in styrene-butadienepolymerization workers. Am. J. Epidemiol. 136, 843-854.

(4) Ward, J. B., Ammenheuser, M. M., Bechtold, W. E., Whorton, E.B., and Legator, M. S. (1994) hprtmutant lymphocyte frequenciesin workers at a 1,3-butadiene production plant. Environ. HealthPerspect. 102 (Suppl. 9), 79-85.

(5) Bolt, H. M., Schmiedel, G., Gilser, J. G., Rolzhauser, H. P., Lieser,K., Wistuba, D., and Schurig, V. (1983) Biological activation of1,3-butadiene to vinyl oxirane by rat liver microsomes andexpiration of the reactive metabolite by exposed rats. J. CancerRes. Clin. Oncol. 106, 112-116.

(6) Elfarra, A. A., Duescher, R. J., and Pasch, C. M. (1991) Mecha-nisms of 1,3-butadiene oxidations to butadiene monoxide andcrotonaldehyde by mouse liver microsomes and chloroperoxidase.Arch. Biochem. Biophys. 286, 244-251.

(7) Sharer, J. E., Duescher, R. J., and Elfarra, A. A. (1992) Speciesand tissue differences in the microsomal oxidation of 1,3-butadi-ene and the glutathione conjugation of butadiene monoxide inmice and rats. Possible role in 1,3-butadiene-induced toxicity.Drug Metab. Dispos. 20, 658-664.

(8) Duescher, R. J., and Elfarra, A. A. (1992) 1,3-Butadiene oxidationby human myeloperoxidase. Role of chloride ion in catalysis ofdivergent pathways. J. Biol. Chem. 267, 19859-19865.

(9) Duescher, R. J., and Elfarra, A. A. (1994) Human liver microsomesare efficient catalysts of 1,3-butadiene oxidation: evidence formajor roles by cytochromes P450 2A6 and 2E1. Arch. Biochem.Biophys. 311, 342-349.

(10) de Meester, C., Poncelet, F., Roberfroid, M., andMercier, M. (1978)Mutagenicity of butadiene and butadiene monoxide. Biochem.Biophys. Res. Commun. 80, 298-305.

(11) Cochrane, J. E., and Skopek, T. R. (1994) Mutagenicity ofbutadiene and its epoxide metabolites: II. Mutational spectra ofbutadiene, 1,2-epoxybutene, and diepoxybutane at the hprt locusin splenic T cells from exposed B6C3F1 mice. Carcinogenesis 15,719-723.

(12) Selzer, R. R., and Elfarra, A. A. (1996) Synthesis and biochemicalcharacterization of N1, N2-, and N7-guanosine adducts of butadi-ene monoxide. Chem. Res. Toxicol. 9, 126-132.

(13) Selzer, R. R., and Elfarra, A. A. (1995) Characterization ofadenosine and guanosine adducts of butadiene monoxide. Toxi-cologist 15, 76.

(14) Selzer, R. R., and Elfarra, A. A. (1996) Characterization of N1-adenosine and N1-inosine adducts of butadiene monoxide. Fun-dam. Appl. Toxicol. (Suppl.) 30, 233.

(15) Singer, B., and Grunberger, D. (1983) Molecular Biology ofMutagens and Carcinogens, Plenum Press, New York.

(16) Fujii, R., Saito, R., and Terahara, N. (1986) Purines. XXVII.Hydrolytic deamination versus Dimroth rearrangement in the9-substituted purine ring: Effect of an omega-hydroxyalkyl groupat the 1-position. Chem. Pharm. Bull. 34, 1094-1107.

(17) Sadtler Research Laboratories Inc. Standard Proton NMR Col-lection (1978) Philadelphia, PA. Spectra 27540M.

(18) Qian, C., and Dipple, A. (1995) Different mechanisms of aralky-lation of adenosine at the 1- and N6-positions. Chem. Res. Toxicol.8, 289-395.

(19) Moore, P. D., Koreeda, M., Wislocki, P. G., Levin, W., Conney, A.H., Yagi, H., and Jerina, D. M. (1977) In vitro reactions of thediastereomeric 9,10-epoxides of (+) and (-)-trans-7,8-dihydroxy-7,8-dihydrobenzo[a]pyrene with polyguanylic acid and evidencefor formation of an enantiomer of each diastereomeric 9,10-epoxide from benzo[a]pyrene on mouse skin. In Drug MetabolismConcepts (Jerina, D. M., Ed.) ACS Symposium Series No. 44, pp127-154, American Chemical Society, Washington, DC.

(20) Koivisto, R., Kostiainen, R., Kilpelainen, I., Steinby, K., andPeltonen, K. (1995) Preparation, characterization and 32P-post-labelling of butadiene monoepoxide N6-adenine adducts. Carcino-genesis 16, 2999-3007.

TX960039A

Butadiene Monoxide-Adenosine Adducts Chem. Res. Toxicol., Vol. 9, No. 5, 1996 881