Quantification of N-(deoxyguanosin-8-yl)-4-aminobiphenyl adducts in human lymphoblastoid TK6 cells dosed with N-hydroxy-4-acetylaminobiphenyl and their relationship to mutation, toxicity, and gene expression profiling

Article abstract of DOI:10.1021/ac0607360

Gene expression profiles that are anchored to phenotypic endpoints may lead to the identification of signatures that predict mutagenicity or carcinogenicity. The study presented here describes the analysis of DNA adducts in the human TK6 lymphoblastoid ce

Full text of DOI:10.1021/ac0607360

Anal. Chem. 2006, 78, 6422-6432
Quantification of
N-(Deoxyguanosin-8-yl)-4-aminobiphenyl Adducts in
Human Lymphoblastoid TK6 Cells Dosed with
N-hydroxy-4-acetylaminobiphenyl and Their
Relationship to Mutation, Toxicity, and Gene
Expression Profiling
Elaine M. Ricicki,† Wen Luo,†,‡ Wenhong Fan,‡ Lue Ping Zhao,‡ Helmut Zarbl,*,‡ and Paul Vouros*,†
1The Barnett Institute and Department of Chemistry and Chemical Biology, Northeastern University,
Boston, Massachusetts 02115, and 2Fred Hutchinson Cancer Research Center, Seattle, Washington 98109
Gene expression profiles that are anchored to phenotypic
endpoints may lead to the identification of signatures that
predict mutagenicity or carcinogenicity. The study pre-
sented here describes the analysis of DNA adducts in the
human TK6 lymphoblastoid cell line after exposure to
N-hydroxy-4-aminobiphenyl, a mutagenic metabolite of
4-aminobiphenyl. A validated nano-LC microelectrospray
mass spectrometry assay is reported for the detection and
quantification of N-(deoxyguanosin-8-yl)-4-aminobiphenyl
(dG-C8-ABP), the principal DNA adduct of 4-aminobi-
phenyl. Limits of quantification, based on a signal-to-noise
ratio of 10:1, are determined to correspond to ∼27 fg of
dG-C8-ABP injected on-column. The assay has been used
to measure the steady-state levels of the adduct in the
human TK6 lymphoblastoid cell line as a function of dose
(0.5, 1.0, and 10.0 µM) and time (2, 6, and 27 h) after
exposure to N-hydroxy-4-aminobiphenyl. The levels of dG-
C8-ABP adducts in the cells, ranging from 18 to 500
adducts in 109 nucleotides, were then correlated to cell
toxicity, induced mutation at the TK (thymidine kinase)
and HPRT loci, and gene expression profiling through
microarray analysis. Cell cultures were evaluated for
toxicity by growth curve extrapolation, mutation assays
were performed on the HPRT and TK loci, and gene
expression profiles were generated by analyses using
microarray technology. In the mutation assay analysis, as
the toxicant concentration increased, there was an in-
crease in mutation fraction, indicating a direct correlation
to metabolite dosing level and mutations occurring at
these two loci. Statistical analysis of the gene expression
data determined that a total of 2250 genes exhibited
statistically significant changes in expression after treat-
ment with N-OH-AABP (P < 0.05). Among the genes
identified, 2245 were up-regulated, whereas 5 genes that
had functions in cell survival and cell growth and, hence,
could be indicators of toxicity, were down-regulated rela-
tive to controls. The results demonstrate the value of
anchoring gene expression patterns to phenotypic mark-
ers, such as DNA adduct levels, toxicity, and mutagenic-
ity.
Many aromatic amines, in particular, arylamines and nitro-
amines, are potent mutagens and have been implicated in chemical
carcinogenesis. Some of the chemicals in this classification include
2-chloroaniline (2-CA), 4-chloroaniline (4-CA), 2-methylaniline (2-
MA), 4-methylaniline (4-MA), 2,4-dimethylaniline (2,4-DMA), 2,6-
dimethylaniline (2,6-DMA), 2-aminobiphenyl (2-ABP), 3-aminobi-
phenyl (3-ABP), and 4-aminobiphenyl (4-ABP), the latter being
the primary focus of several investigations of mutagenesis and
tumorigenesis.1-3 4-Aminobiphenyl is an environmental contami-
nant found in cigarette smoke, paints, food colors, hair dyes, and
fumes from heated oils and fuels.4-7 The mutagenic activity of
aromatic amines has been demonstrated using a variety of
approaches. Mutation assays in both cell lines and animal models
have concluded that these compounds, especially 4-aminobiphenyl,
have significant mutagenic activity. For example, Phillipson and
Ioannides first investigated hepatic microsomal preparations
derived from mice, hamsters, rats, pigs, and humans for the
metabolic activation of the aromatic amines to mutagens.8 More
recently, Lasko et al. investigated the mutagenesis of a reactive
* Corresponding author address: Department of Chemistry and Chemical
Biology, Northeastern University, 120 Hurtig Hall, 360 Huntington Ave., Boston,
MA 02115. Phone: (617) 373-2840. Fax: (617) 373-2693. E-mail: p.vouros@neu.edu.
† Northeastern University.
‡ Fred Hutchinson Cancer Research Center.
(1) Flamini, G.; Romano, G.; Curigliano, G.; Chiominto, A.; Capelli, G.; Bonin-
segna, A.; Signorelli, C.; Ventura, L.; Santella, R. M.; Sgambato, A.; Cittadini,
A. Carcinogenesis 1998, 19, 353-357.
(2) Otteneder, M.; Lutz, W. K. Mutat. Res. 1999, 424, 237-247.
(3) Schieferstein, G. J.; Littlefield, N. A.; Gaylor, D. W.; Sheldon, W. G.; Burger,
G. T. Eur. J. Cancer Clin. Oncol. 1985, 21, 865-873.
(4) Oh, S. W.; Kang, M. N.; Cho, C. W.; Lee, M. W. Dyes Pigm. 1997, 33,
119-135.
(5) Garrigos, M. C.; Reche, F.; Pernias, K.; Sanchez, A.; Jimenez, A. J.
Chromatogr., A 1998, 819, 259-266.
(6) Tokiwa, H.; Nakagawa, R.; Horikawa, K. Mutat. Res. 1985, 157, 39-47.
(7) Turesky, R. J.; Freeman, J. P.; Holland, R. D.; Nestorick, D. M.; Miller, D.
W.; Ratnasinghe, D. L.; Kadlubar, F. F. Chem. Res. Toxicol. 2003, 16, 1162-
1173.
(8) Phillipson, C. E.; Ioannides, C. Mutat. Res. 1983, 124, 325-336.
6422 Analytical Chemistry, Vol. 78, No. 18, September 15, 2006
10.1021/ac0607360 CCC: $33.50 © 2006 American Chemical Society
Published on Web 07/22/2006

form of 4-aminobiphenyl in the Escherichia coli virus, M13mp10,
and Levine et al. researched frameshift mutations in 4-aminobi-
phenyl-induced mutations in Salmonella.9,10 Studies of DNA
binding spectra and biopsies of human tissues have also implicated
4-aminobiphenyl in human bladder carcinogenesis.11,12
Aromatic amines, including 4-aminobiphenyl, can bind co-
valently to DNA bases, including various sites on guanosine and
adenosine bases; however, 4-aminobiphenyl has shown preferen-
tial adduction to deoxyguanosine at the C8 position forming
N-(deoxyguanosin-8-yl)-4-aminobiphenyl (dG-C8-ABP). The latter
adducts have been shown to be premutagenic in both in vitro and
in vivo assays, suggesting that these adducts are also the
precursors for carcinogenic lesions.13,14 As a result, the detection
and measurement of 4-aminobiphenyl adduct levels in tissues of
exposed individuals has significant implications for risk assess-
ment.
for the analysis of dG-C8-ABP adducts using a capillary (320-µm-
i.d.) LC-microESI-MS/MS technique for the analysis of DNA
from human pancreas tissue and used this method to compare
adduct levels in a small set of smokers and nonsmokers. Although
this technique allowed for the detection of the adduct in human
samples, there was no correlation between adduct levels and
smoking.21
Although measuring steady-state adduct levels in human
subjects provides valuable information for risk assessment, their
mere detection provides little insight into the mechanisms of
toxicity and mutagenesis. One approach to understanding the
implications of exposure is to determine how DNA adduct levels
alter gene expression patterns within the cells and how these
changes impact cellular processes, such as DNA repair and cell
cycle control, and then to relate them to biological endpoints, such
as cell growth, apoptosis, and mutagenesis.22-24
It has long been known that upon exposure to DNA-damaging
agents, cells alter gene expression in various organisms, including
yeast and human cells.23-28 For example, exposure to DNA-
damaging agents can increase the transcription of DNA repair
genes and activate cell cycle checkpoints. Historically, gene
expression profiling has employed a variety of analytical methods,
including Northern blotting, dot blots, RNAase protection assays,
or S1 nuclease analysis.29 Typically, the techniques analyze
changes in expression for one or several genes at a time; however,
the advent of high-throughput methods such as DNA microarrays
has allowed for expression profiling in response to toxicants on a
genome-wide scale. Accumulating evidence indicates that compar-
ing global changes in expression patterns as a function of dose
and time after exposure can lead to the identification of gene
signatures that are indicative of toxicity.30 In support of this
hypothesis, two recent dose response studies using DNA micro-
arrays demonstrated that, at least for estrogenic compounds, there
is a dose below which there is no observable transcript effect level
(NOTEL).31,32
Mutagenic adducts of 4-aminobiphenyl, including dG-C8-ABP,
have been detected in human urinary bladder and lung tissue at
levels ranging from <1 to 50 in 108 nucleotides.15 To detect and
measure adducts at these low levels, researchers have developed
a variety of sensitive analytical techniques. Historically, the
prominent analytical technique for DNA adducts has been the 32P-
postlabeling method developed by Randerath and Randerath.16
Kadlubar and co-workers used this technique for the identification
of dG-C8-ABP in biopsy samples of human urinary bladder,
whereas Swaminathan and Hatcher used 32P-postlabeling for the
identification of novel DNA adducts in bladder epithelial cell lines
exposed to a metabolite of 4-aminobiphenyl.17,18
Within the past two decades, liquid chromatography-mass
spectrometry has become a fundamental technique in the analysis
of DNA adducts. Beland and co-workers first synthesized, char-
acterized, and quantified a standard solution of the predominant
aminobiphenyl adduct, N-(deoxyguanosin-8-yl)-4-aminobiphenyl,
by HPLC-ESI-MS.19 These initial studies were followed by a study
reporting the detection and quantification of the dG-C8-ABP adduct
in calf thymus DNA reacted with the N-hydroxy-4-ABP and in
hepatic DNA isolated from mice treated with 4-ABP. For the calf
thymus DNA, the reported levels of dG-C8-ABP were between
1.8 and 430 adducts in 107 normal nucleotides, and for the exposed
mouse liver DNA, dG-C8-ABP levels ranged from 4.9 to 30 in 107
nucleotides.20 Most recently, Ricicki et al. developed a method
It is generally believed that in addition to providing insights
into the mechanisms of toxicity, expression signatures can serve
as biomarkers of toxicity and, hence, can be used for toxicant
classification, exposure monitoring, and risk assessment.33 How-
(20) Doerge, D. R.; Churchwell, M. I.; Marques, M. M.; Beland, F. A. Carcino-
genesis 1999, 20, 1055-1061.
(21) Ricicki, E. M.; Soglia, J. R.; Teitel, C.; Kane, R.; Kadlubar, F.; Vouros, P.
(9) Lasko, D. D.; Harvey, S. C.; Malaikal, S. B.; Kadlubar, F. F.; Essigmann, J.
M. J. Biol. Chem. 1988, 263, 15429-15435.
(10) Levine, J. G.; Knasmuller, S.; Shelton, M. L.; DeMarini, D. M. Environ. Mol.
Chem. Res. Toxicol. 2005, 18, 692-699.
(22) Luo, W.; Fan, W.; Xie, H.; Jing, L.; Ricicki, E.; Vouros, P.; Zhao, L. P.; Zarbl,
H. Chem. Res. Toxicol. 2005, 18, 619-629.
Mutagen. 1994, 24, 11-22.
(23) Guo, Y.; Breeden, L. L.; Preston, B. D.; Zarbl, H.; Eaton, D. L. Mol. Cell.
(11) Feng, Z.; Hu, W.; Rom, W. N.; Beland, F. A.; Tang, M. S. Carcinogenesis
2002, 23, 1721-1727.
(12) Airoldi, L.; Orsi, F.; Magagnotti, C.; Coda, R.; Randone, D.; Casetta, G.;
Peluso, M.; Hautefeuille, A.; Malaveille, C.; Vineis, P. Carcinogenesis 2002,
23, 861-866.
Biol. 2005, 25, 5823-5833.
(24) Guo, Y.; Breeden, L. L.; Kelley, E.; Eaton, D. L.; Zarbl, H. Mutat. Res. 2005,
in press.
(25) Jelinsky, S. A.; Samson, L. D. Proc. Natl. Acad. Sci. U.S.A. 1999, 96, 1486-
1491.
(13) Verghis, S. B.; Essigmann, J. M.; Kadlubar, F. F.; Morningstar, M. L.; Lasko,
(26) Jelinsky, S. A.; Estep, P.; Church, G. M.; Samson, L. D. Mol. Cell. Biol. 2000,
20, 8157-8167.
D. D. Carcinogenesis 1997, 18, 2403-2414.
(14) Torino, J. L.; Burger, M. S.; Reznikoff, C. A.; Swaminathan, S. Carcinogenesis
2001, 22, 147-154.
(27) Zhan, Q.; Carrier, F.; Fornace, A. J., Jr. Mol. Cell. Biol. 1993, 13, 4242-
4250.
(15) Lin, D.; Lay, J. O., Jr.; Bryant, M. S.; Malaveille, C.; Friesen, M.; Bartsch,
H.; Lang, N. P.; Kadlubar, F. F. Environ. Health Perspect. 1994, 102 (Suppl
6), 11-16.
(28) Fornace, A. J., Jr.; Alamo, I., Jr.; Hollander, M. C. Proc. Natl. Acad. Sci.
U.S.A. 1988, 85, 8800-8804.
(29) Nuwaysir, E. F.; Bittner, M.; Trent, J.; Barrett, J. C.; Afshari, C. A. Mol.
(16) Randerath, E.; Randerath, K. IARC Sci. Publ. 1993, 305-314.
(17) Talaska, G.; al-Juburi, A. Z.; Kadlubar, F. F. Proc. Natl. Acad. Sci. U.S.A.
1991, 88, 5350-5354.
Carcinog. 1999, 24, 153-159.
(30) Skopek, T. R.; Liber, H. L.; Penman, B. W.; Thilly, W. G. Biochem. Biophys.
Res. Commun. 1978, 84, 411-416.
(18) Swaminathan, S.; Hatcher, J. F. Chem.-Biol. Interact. 2002, 139, 199-213.
(19) Beland, F. A.; Doerge, D. R.; Churchwell, M. I.; Poirier, M. C.; Schoket, B.;
Marques, M. M. Chem. Res. Toxicol. 1999, 12, 68-77.
(31) Crespi, C. L.; Thilly, W. G. Mutat. Res. 1984, 128, 221-230.
(32) Zarbl, H. Current Protocols in Toxicology; John Wiley & Sons: New York,
2001; Vol. 9, pp 1.4.2-1.4.16.
Analytical Chemistry, Vol. 78, No. 18, September 15, 2006 6423

ever, to identify signatures of toxicity, mutagenicity or carcino-
genicity, gene expression signatures must be anchored to phe-
notypic changes that are associated with a particular endpoint.
For example, a gene expression signature that is associated with
cellular toxicity may predict toxicity at exposure levels that do
not result in overt toxicity, that is, doses that are below the so-
called no observable adverse effect limit (NOAEL). However, the
external dose is not always an effective measure of biologically
effective exposure, because the absorption, distribution, metabo-
lism, and excretion of toxicants can be modulated by both genetic
and environmental factors. In contrast, DNA adduct levels are a
more proximal measure of the biologically effective dose for
genotoxicants, integrating exposure, metabolic activation, DNA
adduction, and repair. Thus, gene expression profiles that are
anchored to DNA adduct formation are more likely to lead to the
identification of signatures that predict mutagenicity or carcino-
genicity. Although an increasing number of studies are attempting
to anchor expression profiles to phenotypic endpoints, only a few
studies of 4-aminobiphenyl have attempted to correlate expression
profiles to phenotypic endpoints, and none attempted to anchor
gene expression profiles to 4-aminobiphenyl-DNA adduct levels.22-24
In the present study, we developed a highly sensitive mass
spectrometry-based assay for detection of the mutagenic dG-C8-
ABP adduct. This assay was then used to measure the steady state
levels of the adduct in the human TK6 lymphoblastoid cell line
as a function of dose and time after exposure to N-hydroxy-4-
acetylaminobiphenyl, the mutagenic metabolite of 4-aminobiphe-
nyl. Cell cultures dosed at various concentrations and harvested
at several time points after exposure were independently but
simultaneously evaluated for toxicity, mutation levels, gene
expression profiles, and DNA adduct quantity. Integration of these
data allows for a comprehensive assessment of the cellular
responses as a function of exposure. Therefore, the focus of this
report is 3-fold: First, the method developed herein provides a
novel, validated assay for the analysis of dG-C8-ABP DNA adducts
by microcapillary liquid chromatography-microelectrospray mass
spectrometry. Second, the assay is utilized for the quantification
of dG-C8-ABP adducts present in human TK6 cells from both a
dose and time response study. Third, the quantification of dG-
C8-ABP adducts is correlated to cell toxicity, induced mutation at
the TK and HPRT loci, and gene expression profiling through
microarray analysis.
from Amersham Pharmacia Biotech (Piscataway, NJ). N-Hydroxy-
4-acetylaminobiphenyl (N-OH-AABP) was purchased from
National Cancer Institute Chemical Carcinogen Reference Stan-
dard Repository (Kansas City, MO). Blood and Cell Culture DNA
Maxi Kits and RNA Maxi Kits were purchased from Qiagen Inc.
(Valencia, CA). RPMI medium 1640 with L-glutamine was obtained
from Life Technologies, Inc. (Grand Island, NY). Solvents were
obtained from Fischer Scientific (Pittsburgh, PA) and were HPLC
grade unless otherwise noted. All water was prefiltered through
a Millipore Milli-Q plus system (Millipore Co., Bedford, MA) and
had a minimum resistance of 18 MΩ.
Adduct Synthesis and HPLC Purification. As described
previously,21 the N-(deoxyguanosine-8-yl)-4-aminobiphenyl stan-
dard was synthesized by hydrazine reduction of 4-nitrobiphenyl
to N-hydroxy-4-aminobiphenyl following addition of pyruvonitrile
for conversion to N-acetoxy-4-aminobiphenyl. Upon completion of
the reaction, molecular weight of the product was confirmed by
mass spectrometry. The product was then reacted with 2′-
deoxyguanosine in a 10:1 mole ratio to produce dG-C8-ABP. The
deuterated internal standard, N-(deoxyguanosine-8-yl)-4-aminobi-
phenyl-d9 (dG-C8-ABP-d9) was prepared using 4-nitrobiphenyl-d9
in a similar manner, except on a smaller scale due to limited
availability of 4-nitrobiphenyl-d9.
Purification of both standards was performed by reversed-
phase HPLC (column: Supelco, 15 cm × 4.6, C18 column) using
a 20-min linear gradient of 20-70% methanol (aqueous portion:
10 mM ammonium acetate). Standard and internal standard
identification was performed by infusing effluent from the HPLC
diode array detector to a TSQ 700 triple quadrupole mass
spectrometer for analysis in the product ion scanning mode.
Preparation of dG-C8-ABP Standard Curves. A 200-µg
portion of calf thymus DNA was added to each of six vials
containing 250 µL of 5 mM TRIS/10 mM MgCl2 (pH 7.2), then
18 fmol of dG-C8-ABP-d9 was added to the solution (20 µL working
solution) along with 30 µL of a previously prepared serial dilution
solution of dG-C8-ABP, such that the mass range spanned 2 orders
of magnitude. Specifically, the masses of dG-C8-ABP added were
350, 68.3, 54.7, 34.1, 20.1, 10.0, and 5.00 fmol. A solution of DNase
I (dissolved at 10 mg/mL in 5 mM TRIS/10 mM MgCl2 (pH 7.4)
was added at 770 units/mL and incubated at 37 °C for 5 h. Alkaline
phosphatase (straight, 4.0 units/mL) and 0.3 units/mL phosphodi-
esterase (1.0 mg/mL in 5 mM TRIS/10 mM MgCl2, pH 7.40) were
introduced to the reaction solution, and the digest was then
incubated for 18 h. The digestion was terminated with 3 volumes
of ice cold ethanol (-20 °C), following which the mixture was
centrifuged for 15 min at 5000g (Fisher Scientific, model Marathon
21000R) to pellet and discard any insoluble material. The super-
natant was then evaporated to ∼100 µL, and the digestion mixtures
were diluted with 1.5 mL of 5 mM potassium phosphate, monoba-
sic (pH 3.4) for further sample enrichment by solid-phase
extraction.
MATERIALS AND METHODS
Chemicals. Caution: 4-Aminobiphenyl and its derivatives are
carcinogenic to humans and should be handled carefully. The
following chemicals were obtained from Sigma Chemical Co. (St.
Louis, MO): calf thymus DNA, potassium phosphate, monobasic
(anhydrous, min. 99%), deoxyribonuclease I (type 2 from bovine
pancreas), alkaline phosphatase (type IIIs), tris (hydroxymethyl)-
aminomethane (TRIZMA), magnesium chloride, 6-thioguanine
(6TG), triflurothymidine (F3TdR), deoxycytidine, hypoxanthine,
aminopterin, thymidine, and horse serum. Acetic acid (glacial,
99.99+%) and anhydrous DMSO (99.997% purity) sealed with N2
were obtained from Aldrich Chemical Co. (Milwaukee, WI).
Phosphodiesterase 1 (Crotalus adamanteous venom) was received
Adduct Enrichment and Sample Purification Using Solid-
Phase Extraction. Solid-phase extraction was accomplished
utilizing a vacuum manifold (Alltech Associates, Deerfield, IL).
Disposable, Teflon manifold needles (Alltech Associates, Deerfield,
IL) replaced standard valves to limit cross-contamination. Isolute
C18 (EC), 100-mg sorbent cartridges (Argonaut Technologies,
Foster City, CA) were conditioned with 3 mL of methanol followed
(33) Iwata, M.; Graf, L.; Awaya, N.; Torok-Storb, B. Blood 2002, 100, 1318-
1325.
6424 Analytical Chemistry, Vol. 78, No. 18, September 15, 2006

by 3 mL of 5 mM potassium phosphate, monobasic (pH 3.40,
adjusted with glacial acetic acid). Digestion mixtures were then
applied to the conditioned cartridges and permitted to flow
through the cartridge at ∼1 mL/min. Washes of 3 mL of 10:90
(v/v%) methanol/5 mM potassium phosphate, monobasic (pH
3.40) and 3 mL of water were used to elute any extraneous
material, including unadducted nucleosides and other polar
constituents. Adducted DNA nucleosides were then eluted from
the sorbent with 1 mL of methanol and evaporated to dryness
under vacuum. Prior to analysis, the sample was reconstituted in
40 µL of water.
Cell Culture. Human lymphoblastoid line TK6 30 was derived
from the parental human lymphoblastoid line HH4 and heterozy-
gous for thymidine kinase, an enzyme which phosphorylates
thymidine and its toxic analogues in an ATP-dependent reaction.
The cells were grown in spinner flasks to allow for gentle stirring.
Cells were maintained in exponential growth by daily dilution in
RPMI 1640 medium supplemented with 5% donor horse serum
and maintained in 37 °C incubators with a 5% CO2 atmosphere.
Cell counts were taken daily using a Coulter Counter, and cultures
were diluted to ∼4 to 5 × 105 cells/mL. Detailed growth records
were maintained and used to ascertain any effects of the treatment
on growth rate.
Chemical Treatment and Cellular DNA Extraction. N-OH-
AABP stock solutions were prepared in anhydrous DMSO
(99.997% purity) and used immediately. The final DMSO concen-
tration in the culture was <0.1%. To reduce the background HPRT
mutant fraction prior to mutagen treatment, TK6 cells were first
cultured in “CHAT” culture medium for 3 days, followed by a 2-day
recovery period in “THC” medium.31 Cells were then treated with
three concentrations of N-OH-AAB: 0.5, 1.0, 10.0 µM for up to
27 h. Duplicate cultures were completed at each condition.
Untreated cells were used as negative controls. At 2, 6, and 27 h
after the treatment, two 100-mL aliquots of cells (∼5 × 105 cells/
mL) were removed, and cells were collected by centrifugation,
immediately frozen down in liquid nitrogen, and stored at -80
°C. DNA was extracted from human cells using Qiagen Blood
and Cell Culture DNA Maxi Kits according to the manufacturer’s
instructions. Once DNA was isolated from tissue, 200-µg aliquots
were added to siliconized microcentrifuge tubes. A solution of 5
mM TRIS/10 mM MgCl2 (pH 7.2) was added to produce final
sample volumes of 300 µL. The genomic DNA was then digested
to nucleosides with DNase I at 37 °C for 5 h as described above.
Mutagenesis, Survival Measurements, and Global Gene
Expression. The measurement of mutant fractions at HPRT or
TK locus, cellular toxicity, and global gene expression after the
N-OH-AABP treatment were described in detail previously.22
Briefly, to determine mutant fraction at HPRT or TK locus, cells
were plated in the presence and absence of 6-thioguanine or
triflurothymidine, respectively. Survival of treated cultures was
calculated from back-extrapolation of growth curves by compari-
son of the growth of treated cultures to control cultures.
Human cDNA arrays were generated in the Genomics Facility
at Fred Hutchinson Cancer Research Center.32,33 Microarrays were
constructed employing the first 17 569 clones from the sequence-
verified ResGen Human Unigene Set comprising 31 105 clones
(Research Genetics, Huntsville, AL; now Invitrogen). Clones are
representative of over 20 different human tissues/organs. The
average length of the probe was 1200 ( 364 bases, with the
minimum probe length being 251 bases and the maximum probe
length being 2000+ bases.
RNA from the cells stored at -80 °C was extracted using
Qiagen RNeasyMaxi Kits according to the manufacturer’s instruc-
tions (Qiagen, Valencia, CA), followed by a quality test using
Agilent BioAnalyzerchips. Synthesis of the labeled first strand
cDNA was conducted using InVitrogen Superscript II reverse
transcript system with starting material of 30 µg total RNA plus
oligo-dT. The aminoallyl labeled dUTP was added to the reaction
to generate aminoallyl-labeled second strand cDNA. Following the
hydrolysis reaction, single-stranded cDNA probes were purified
using a Microcon-30 concentrator. Probe mixtures were evapo-
rated in a vacuum centrifuge, and the cDNA pellet was resus-
pended in 4.5 µL of water. The dye coupling reactions were
performed by mixing the cDNA samples with Cy3 or Cy5 dyes
and incubating for 1 h in the dark. After the reactions were
quenched with hydroxylamine, Cy3 and Cy5 samples were
combined for hybridization. To remove unincorporated/quenched
Cy dye, a Qia-quick PCR purification kit (Qiagen) was used. The
reactions were purified with Qiaquik PCR purification kits to
remove the unincorporated/quenched dyes. The mixture of Cy3-
and Cy5-labeled samples was dried down in a vacuum centrifuge
and then brought to a volume of 20 µL with water. Millipore 0.45
µM spin columns were used to clean up after adding 20X SSC
and Poly A to the mixture. Finally, the samples were stored at
-20 °C until ready for hybridization.
The labeled cDNAs were co-hybridized to cDNA microarrays
in triplicate, including one dye swap. Slides were scanned on the
GenePix 4000A Microarray Scanner at the optimal wavelength for
the Cy3 and Cy5 using lasers. The fluorescence signals of each
spot on the slide were analyzed and extracted with GenePix
software to generate.gpr file, which was further formatted to a
plain.txt file for further data analysis. All the data-mining steps,
including transformation, cleaning-up, and normalization, were
performed using the SAS statistical software.
The mean log2 ratios of fluorescence intensities relative to
untreated controls for each gene and each treatment (0.5, 1, or
10 µM N-OH-AABP) were clustered using the J-Express
Software (University of Bergen, Norway) unsupervised algorithm.
Briefly, the fluorescence intensity of each spot was calculated
using local median background subtraction. The relative fluores-
cent units (RFUs) were then normalized to the median signal (Cy3
and Cy5) for that slide. If both Cy3 and Cy5 signals were under
the signal intensity of negative controls, genes were considered
as not being expressed in either sample. The normalized microar-
ray data were then analyzed using two approaches, visualization
and statistical analysis. The first of these involved a comparison
of the normalized fluorescence intensity ratios and was used only
for the purpose of visualizing the data. In this approach, the fold
change in gene expression for each array feature was determined
as a ratio of normalized fluorescence intensities relative to the
reference sample (i.e., RFU treated/RFU untreated). We then
selected a value of 2.0 in fold change in gene expression as the
arbitrary cutoff for inclusion of data in clustering algorithms. The
log2 gene expression ratios for all genes showing a 2-fold or
greater change in gene expression at any concentration were used
in unsupervised cluster analysis to generate the dendrogram
Analytical Chemistry, Vol. 78, No. 18, September 15, 2006 6425

Figure 1. Schematic of capillary liquid chromatography system in line with Finnigan TSQ 700 mass spectrometer.
shown in Figure 8. The arbitrary 2-fold cutoff was selected because
it is the most frequently used value for preliminary comparison
of expression profiles. As illustrated in Figure 8, this approach
revealed significant differences in the expression profiles induced
by N-OH-AABP. Nonetheless, the data generated using this
arbitrary cutoff value do not provide any indication as to the
statistical significance of the differences in expression profiles.
The latter requires that the data be subjected to rigorous statistical
analyses, as described next.
Capillary Liquid Chromatography and µESI Mass Spec-
trometry. Chromatographic separations were performed on a
PicoFrit Aquasil packed C18 capillary column (75-µm i.d., 5-cm
length, 3-µm particle size) (New Objective, Woburn, MA). Solvent
was delivered using an Agilent 1100 Capillary HPLC System
(Agilent Technologies, Palo Alto, CA) at a flow rate of 300 nL/
min. A four-port microbore (0.15 mm) valve equipped with a
stainless steel stator and 0.5-µL PEEK internal sample loop was
placed in-line and used for sample introduction. The desired
analyte was separated from other components using a binary
mobile phase composed of 1% acetic acid in water and methanol.
The mobile phase was initially set at 30% methanol (1% acetic acid).
To allow for sample concentration at the head of the column, the
initial conditions were held for 1 min after introduction. The
percent methanol was subsequently increased to 70% using a step
gradient at 1 min and held for 6 min. This was followed by a step
gradient down to 30%, which was held for 13 min to allow for
reequilibration of the column.
All mass spectrometric analyses were performed on a Finnigan
TSQ 700 triple quadrupole mass spectrometer (Thermo, San Jose,
CA) equipped with an in-house-built micro-ESI source, as il-
lustrated in Figure 1. The mass spectrometer was calibrated with
a 1,3,6 polytyrosine solution. Capillary temperature was set to 175
°C, and the capillary voltage was set at 1.8 kV; due to the low
flow rate, a sheath gas was unnecessary. All sample analyses were
performed in the positive mode utilizing selected reaction moni-
toring (SRM). In the specific applications, the first quadrupole
was set to transmit [M + H]+, which corresponded to 435.10 and
444.10 for dG-C8-ABP and dG-C8-ABP-d9, respectively. The ions
then entered the second quadrupole, where they were fragmented
using a collision energy of 30 V. The third quadrupole transmitted
only the aglycon fragments [M + H - 116]+. For the dG-C8-ABP
and dG-C8-ABP-d9 ions, these masses corresponded to 319.0 and
328.1.
To make statistical inference, we tested a null hypothesis for
every gene: gene expression does not change in N-OH-AABP-
treated samples, as compared to the untreated control samples.
On a cDNA array, for each gene, the log ratio of signal from
treated sample in one channel over the signal from control sample
in the other channel was used as raw data, thereby eliminating
some systemic variation. Thus, the null hypothesis becomes the
following: the log ratio is equal to zero. After testing the
hypotheses for every gene, the genes that rejected the hypothesis
at a certain significance level were selected as candidate genes;
i.e., N-OH-AABP has an effect on the expression profiles of
these genes. For this analysis, we used the software GenePlus
(Enodar Biologic Corporation, http://www.enodar.com, Seattle,
WA), which implements a regression model using the estimating
equation technique.34-36 Normalization was carried out simulta-
neously in the regression procedure. The regression coefficient
and its standard error were estimated from normalized data. The
standardized coefficient, Z, defined as the regression coefficient
divided by its standard error, was used to make the statistical
inference. The higher the absolute value of standardized coef-
ficient, the more significant the gene expression change is. The
sign of the standardized coefficients reflects the directionality of
change. To control false positives, the type I error was set equal
to 5% in analyses. Since multiple hypothesis testing was carried
out simultaneously for multiple genes, Bonferroni’s correction was
used to control the inflated type I error.
Quantification of dG-C8-ABP in Human TK6 Cells. The
standard curve for dG-C8-ABP was prepared in triplicate by spiking
synthetic preparations of the adduct and a deuterium-labeled
internal standard into a solution of calf thymus DNA, followed by
the previously described enzymatic digestion, and analyzed in
(34) Zhao, L. P.; Prentice, R.; Breeden, L. Proc. Natl. Acad. Sci. U.S.A. 2001,
98, 5631-5636.
(35) Thomas, J. G.; Olson, J. M.; Tapscott, S. J.; Zhao, L. P. Genome Res. 2001,
11, 1227-1236.
(36) Xu, X. L.; Olson, J. M.; Zhao, L. P. Hum. Mol. Genet. 2002, 11, 1977-1985.
6426 Analytical Chemistry, Vol. 78, No. 18, September 15, 2006

Figure 2. Infusion analysis of dG-C8-ABP (A) and dG-C8-ABP-d9 (B) standard and internal standard. Product ion spectra of the [M + H]+ ion
of the standard and internal standard with a collision offset voltage of -70. Structures illustrate proposed fragmentation patterns.
triplicate. The curve was then constructed by plotting the analyte-
to-internal standard peak area ratio at each mass level versus the
mass of dG-C8-ABP in each 200 µg of DNA digested. Linear
regression data derived from the plot (m ) 0.0177, R2 ) 0.9987)
enabled the quantification of dG-C8-ABP.
RESULTS
Mass Spectrometric Analysis of ABP-DNA Adduct Syn-
thetic Standard and Internal Standard. The identities of the
synthetic dG-C8-ABP standard and its deuterium-labeled analogue
(95% isotopic purity) were confirmed by direct infusion of their
respective solutions (0.330 µg/mL and 1.17 µg/mL) into a TSQ
700 mass spectrometer at a rate of 0.3 µL/min. The ESI/MS/MS
mass spectra of the unlabeled and labeled adducts utilizing a
Figure 3. (A) Extracted ion chromatogram (435 f 319) of undi-
gested 6 fg/µL (3 fg on-column) solution of dG-C8-ABP; (B) extracted
ion chromatogram (444 f 328) of undigested 15 fg/µL (7.5 fg on-
collision offset voltage of -20V verified the respective protonated
[M + H]+ molecular masses of 435 and 444 for dG-C8-ABP and
dG-C8-ABP-d9 as well as the facile formation of the aglycon
fragment ion [M + H - 116]+ at m/z 319 and 328, respectively,
for both standards.21 The ESI/MS/MS mass spectra of the [M +
H]+ ions subjected to an increased collision energy (-70 V)
produced a wide variety of additional fragment ions (Figure 2,
panels A and B). Inset in each spectrum are the structures of dG-
C8-ABP and dG-C8-ABP-d9. Within each structure is the proposed
fragmentation pattern producing the predominant mass-to-charge
ratios.
LC-MS Analysis of ABP-DNA Adduct Standard and
Internal Standard. Although nucleosides often exhibit limited
retention using traditional reversed-phase C18 liquid chromatog-
raphy, the aminobiphenyl adduct showed greater retention than
normal nucleosides because of the hydrophobic aromatic amino
moiety. After trial of several different liquid chromatography
methods, optimal liquid chromatography conditions were deter-
mined to be a step gradient with initial conditions set at 30%
methanol (1% acetic acid) and increased immediately to 70%
methanol (1% acetic acid) 1 min after injection. Standard dG-C8-
ABP and dG-C8-ABP-d9 were analyzed using this methodology to
determine limits of detection using the 75-µm C18 column and the
column) solution of dG-C8-ABP-d9.
in-house built mass spectrometry interface. Figure 3 illustrates
the results from 0.5-µL injections of a 6 fg/µL dG-C8-ABP and 15
fg/µL dG-C8-ABP-d9 solution. Panel A is the extracted ion
chromatogram for dG-C8-ABP (m/z 435 f 319); Panel B is the
extracted ion chromatogram for dG-C8-ABP-d9. This example
demonstrates some important observations with regard to the
mass spectrometry of the analytes of interest. Foremost, 3 fg on-
column is close to the limit of quantification on the basis of the
13:1 signal-to-noise ratio. Additionally, the elution pattern is of
importance because although not completely resolved, as seen
previously, the deuterium-labeled internal standard elutes slightly
earlier than the nonlabeled analyte, dG-C8-ABP. Furthermore, this
sample was used as a check for system suitability, with the sample
analyzed each day prior to unknown analyses. After a week of
inter- and intraday analyses, the RSD of the standard and internal
standard retention time was <2%, and the RSD of the analyte to
internal standard peak area ratio was 4.88%.
As discussed earlier, a calibration curve was prepared by
spiking dG-C8-ABP as well as a deuterium-labeled internal
Analytical Chemistry, Vol. 78, No. 18, September 15, 2006 6427

Table 1. Quantification of DG-C8-ABP in Cell Dosing
Study
µM
anal/IS %RSD
av
% RSD adducts/
concn
ratio
ratioa
fmol
fmola
109 nt
control
2 h
0.5
1.0
10.0
0.5
3.84
1.52
2.63
3.76
3.80
2.00
1.10
1.51
207
17.7
11.3
323
17.3
53.2
274
1.60
4.17
7.23
3.93
3.20
1.31
1.57
345
30.0
18.3
539
28.4
88.4
457
6 h
0.496
0.382
5.90
1.0
10.0
0.5
27 h
0.490
1.12
5.04
1.0
10.0
a n ) 3.
LC-MS/MS Detection of dG-C8-ABP in Dosed TK6 Cells.
After validation of the capillary liquid chromatography-mass
spectrometry methodology, human TK6 cells dosed with N-OH-
AABP were analyzed for the presence of dG-C8-ABP. The dosing
and harvesting schedules were as described in detail in the
Experimental Section. Briefly, the cells were dosed with 0.5, 1.0,
or 10.0 µM N-OH-AABP and harvested at 2, 6, or 27 h. A
complete study of these samples demonstrated DNA adduct
presence as a function of dose and time (See Table 1). Figure 5
shows representative chromatograms based on variation of dose
levels. Panel 1 shows extracted ion chromatograms (m/z 435 f
319, dG-C8-ABP) from the analyses of cells harvested at 6 h; the
chromatograms in panel 2 resulted from the analyses of cells
harvested at 27 h. As labeled in the Figure, chromatogram “a”
represents the analysis of the blank human TK6 cells used as the
control. There was a small peak in the blank that was consistently
observed throughout all the analyses of human cells. The
reproducible retention time of 15.29 min demonstrated that this
impurity did not interfere with and was not related to the analyte
of interest, dG-C8-ABP, which eluted at ∼14.0 min.
Figure 4. (A) Extracted ion chromatogram (435 f 319) of limit of
quantification 5-fmol dG-C8-ABP standard (∼54 fg/µL solution; 27 fg
on column) (B) Extracted ion chromatogram (444 f 328) of corre-
sponding 18-fmol internal standard, dG-C8-ABP-d9 (∼200 fg/µL; 100
fg on-column).
standard into calf thymus DNA, followed by digestion and sample
enrichment. Figure 4 is a representative chromatogram of the
lowest standard (5 fmol of dG-C8-ABP spiked into calf thymus
DNA). Panel A is the extracted ion chromatogram of m/z 435 f
319 (dG-C8-ABP); Panel B is the extracted ion chromatogram of
m/z 444 f 328 (dG-C8-ABP-d9). As demonstrated in Figure 4, the
samples processed through digestion and solid-phase extraction
did not exhibit the same sensitivity as unprocessed standards.
The lowest calibration curve standard represents an injection of
∼27 fg on-column, and on the basis of a signal-to-noise ratio of
10:1, it was determined to be the limit of quantification. As would
be expected, in comparison with Figure 3, the amount injected
after sample preparation is ∼10 times more than that prior to
digestion and sample enrichment. The analysis of the low standard
was determined to be reproducible by triplicate analysis of the
standard curve in which the retention time had a limited variation
(<1.9%) and the analyte-to-internal standard peak area ratio did
not vary by more than 7.7%.
To visually compare the peaks presented, the y-axes of all
chromatograms were normalized on the basis of the peak of
greatest intensity of the figure (chromatogram “d” of panel 1,
Figure 5). Due to the decreased peak height in the lower doses,
Figure 5. Extracted ion chromatograms of dG-C8-ABP (435 f 319) from the dose response study. “a” in 1 and 2 is the chromatogram from
the analysis of blank human TK6 cells. Panel 1 (left) chromatograms are from harvesting 6 h after dosing; panel 2 (right), 27 h after dosing. (b)
0.5, (c) 1.0, and (d) 10.0 µM dosing.
6428 Analytical Chemistry, Vol. 78, No. 18, September 15, 2006

Figure 6. Extracted ion chromatograms of dG-C8-ABP (435 f 319) from time response study of 1.0 µM dosing. “a” is the chromatogram from
the analysis of blank human TK6 cells, “b” is analysis of cells harvested 2 h after dosing, “c” is 6 h after dosing, and “d” is 27 h after dosing.
Figure 7. (A) Survival of TK6 cells as a function of OH-AABP dose 27 h after treatment. Percent determined by growth curve extrapolation
relative to vehicle-treated control cells. (B) Mutant fractions induced at the TK locus of TK6 cells as a function of carcinogen dose. Induced
mutant fractions were determined by selection in medium containing trifluorothymidine. (C) Mutant fractions induced at the HPRT locus of TK6
cells as a function of carcinogen dose. Induced mutant fractions were determined by selection in medium containing 6TG.
the scale had to be increased, which is indicated by the magnifica-
tion inset on the chromatogram. At both time points, there was a
significant increase in both area and signal-to-noise ratio when
comparing the 1.0 µM dose to the 10.0 µM dose. In the same
respect, 10.0 µM doses of the 6- and 27-h harvestings demonstrate
relatively similar peak areas and signal-to-noise ratios.
Whereas Figure 5 demonstrated a comparison based on dose
levels, the chromatograms in Figure 6 illustrate a comparison
based on length of exposure. As in Figure 5, chromatogram “a”
displays the analysis of the control human TK6 cells. Rows b, c,
and d are the extracted ion chromatograms (m/z 435 f 319) from
cells harvested at 2, 6, and 27 h, respectively. As in Figure 5, the
y-axes are normalized to the peak with the greatest intensity
(chromatogram “d”). In the analysis of the cells harvested after 2
h, no adduct-specific peak was detectable. However, after 6 h, the
presence of a small peak corresponding to the analyte of interest,
dG-C8-ABP, was clearly visible. The analyte peak area of the cell
sample harvested 27 h after dosing was significantly greater than
that of the 6-h sample. Additionally, closer investigation of the
peak illustrated a shoulder, suggesting the possible presence of
an isomeric adduct.37
On the basis of the equation produced from the calibration
curve, the analyte-to internal standard peak area ratio from the
samples was utilized to calculate the amount of adduct in each
sample. Table 1 displays a summary of these calculations. In
addition, the relative standard deviations (n ) 3) are displayed
(37) Swaminathan, S.; Hatcher, J. F. Chem.-Biol. Interact. 2002, 139 (2), 199-
213.
Analytical Chemistry, Vol. 78, No. 18, September 15, 2006 6429

for both the ratio and the average calculated mass of dG-C8-ABP
adduct. The final column of the table also demonstrates the
calculation of the number of adducts in 109 nucleotides. The
standard deviations indicate sufficient reproducibility with no more
than 3.8% variation for analyte-to-internal standard peak area ratios.
Further, it should be noted that 2 h after dosing at the 0.5 and
1.0 µM levels, no adducts were detected using this methodology;
however, 6 and 27 h after dosing, dG-C8-ABP adducts were present
at all three dosing levels. Finally, the 10 µM metabolite dose was
sufficient for the rapid formation of adducts considering all time
points collected after dosing. Moreover, at this dose, the detected
levels of the dG-C8-ABP adduct were similar and were at least
3.5 times greater than in any sample dosed with either 0.5 or 1.0
µm N-OH-AABP. These finding suggest that at the higher dose,
adduct levels reached a steady-state level by 6 h after the cells
were treated with N-OH-AABP.
Toxicity and Mutation Results. The first parameter observed
in the cell-dosing experiment was the toxicity of the 4-aminobi-
phenyl metabolite. Figure 7A illustrates a summary of the cell
survival results for the samples harvested after 27 h of dosing.
The 95, 85, and 60% designate the cell survival rate at concentra-
tions of the 0.5, 1.0, and 10.0 µM, respectively. It should be noted
that cell survival was essentially 100% after 2 and 6 h of exposure,
and cell death was observed only after 27 h.
We next determined the mutagenicity of given exposures using
mutation assays for the TK and HPRT loci (Figure 7B and C,
respectively). The mutation fraction (x-axis) is plotted as a function
of concentrations (y-axis) for both loci. For both assays, as the
toxicant concentration increased, there was an increase in muta-
tion fraction, indicating a direct correlation to metabolite dosing
level and mutations occurring at these two loci. For example, the
mutation levels at the HPRT locus following exposure with 10.0
µM N-OH-AABP increased from a background level of 2.31 ×
10-6 to 40.7 × 10-6, an increase of about 20-fold. Similarly, the
levels on the TK locus increased approximately 40-fold from 1.63
Figure 8. Clustering of the average fold change in gene expression
levels induced in TK6 cells as a function of carcinogen-induced
toxicity. TK6 cells were treated with N-OH-AABP at 0.5, 1.0, or 10
µM to result in 5 (L), 15 (M), and 40% (H) toxicity (see Figure 7A).
Experimental conditions are on the horizontal axis and are specified
at the top of the dendogram. Individual genes are represented along
the vertical axis. The log2 ratio expression for each of these genes
under each treatment is represented by a red and green scale, where
saturated red denotes log2 value of 3.784, and saturated green
represents a log2 value of -3.784.
× 10-5 to 63.4 × 10-5
.
Gene Expression Profiling. The dendrogram in Figure 8,
shows gene expression profiles obtained by the microarray
analysis of the mRNA isolated from the cells harvested 27 h after
dosing with three different doses of N-OH-AABP at 0.5, 1.0,
and 10 µM. Each horizontal bar represents a different gene
analyzed by the microarray experiments, and columns represent
experiments performed at different doses. For each gene, the
mean log ratio of the gene expression values between treated and
untreated control was calculated using the six independent
microarrays from triplicates. After subtraction of local median
background and normalization to the median signal for a given
microarray, the relative fold change in expression for each feature
was calculated as the mean log2 ratio of fluorescence intensities
relative to untreated controls. A total of 521 array features were
found to change at least 2-fold under one treatment condition and
were subsequently subjected to hierarchical clustering using the
unsupervised algorithm in J-Express Software. Examination of the
dendrogram in Figure 8 clearly indicates that not all genes show
comparable responses to increasing doses of N-OH-AABP.
Whereas some genes show a clear dose response, the response
of other genes was clearly saturated, even at the lowest dose.
Although a detailed analysis of the data is beyond the scope of
the present paper due to a lack of statistical power, the combina-
tion of DNA adduct and gene expression data as a function of
dose in the same cells affords the opportunity to directly correlate
expression profiles with adduct levels in the same cells.
Statistical regression was next used to test if the log ratio values
were significantly different from zero. A log ratio of zero indicated
no change in expression. If the log ratio was not zero, expression
changes were sorted on the basis of direction of change, either
up- or down-regulated relative to controls. Full statistical analysis
of the data, complete with a description of data analysis methods
and results was published in our previous study.22 Statistical
analysis determined that a total of 2250 genes exhibited statistically
significant changes in expression after treatment with N-OH-
AABP (P < 0.05). Among the genes identified, 2245 were
up-regulated, and 5 were down-regulated relative to controls. For
example, five genes that were down-regulated had functions in
cell survival cell growth and, hence, could be indicators of
toxicity.22
Table 2 consists of partial lists of genes with known ontogonies
whose expression levels were significantly up-regulated in re-
sponse to N-OH-AABP. Among these are genes whose ontogo-
6430 Analytical Chemistry, Vol. 78, No. 18, September 15, 2006

Table 2. Subset of Genes Whose Expression Ratios Were Identified as Significantly Increased by Treatment with
Both N-OH-AABP as Determined by Statistical Analysis with 95% Confidence
gene
gene bank
gene ontology gene name
symbol
accession ID
fold change
Za
Stress Response
HSPA8
HSPC125
DNAJB9
HSP105B
MTF1
heat shock 70 kD protein 8
AA629567
1.7
1.4
1.5
1.6
1.5
5.0
5.4
8.0
5.4
8.9
HSPC125 protein
W92994
DnaJ (Hsp40) homolog
heat shock 105 kD
metal-regulatory transcription factor 1
AA045793
AA485151
AA448256
Inflammary
SCYA4
small inducible cytokine A4
interleukin 18 receptor 1
interleukin 13 receptor
H62864
AA482489, AI821652
AA478570
1.5
2.1
1.5
4.5
5.5
5.0
IL18R1
IL13RA1
DNA Repair
CRA
cisplatin resistance-associated
W72697
W81192
N52195
H49455
AA453300
1.85
1.28
1.57
3.44
1.4
3.80
6.39
5.74
9.81
4.7
glutathione transferase omega
GSTTLp28
HRMT1L1
APXL
HMT1 (methyltransferase)-like 1
apical protein, Xenopus laevis-like
xeroderma pigmentosum, complementation group A
XPA
Tumor Suppressor and Oncogenes
TNF receptor-associated factor 3
RAB18, member RAS oncogene family
p53-induced protein
tumor protein p53-binding protein, 2
RAB23, member RAS oncogene family
TRAF3
RAB18
PIG11
H48096
1.8
1.4
1.5
1.4
2.0
7.1
7.9
3.8
4.5
8.3
AA156821
AA132070
H69077
TP53BP2
RAB23
AA134570
Transcription
SFRS5
POLR2C
TCEB3
splicing factor, arginine/serine-rich 5
polymerase (RNA) II (DNA-directed)
transcription elongation factor B (SIII)
R73672
AA430656
AA133129
1.4
1.2
1.2
3.8
4.2
4.2
Metabolism
HSPC051
HCCS
ubiquinol-cytochrome c reductase complex
holocytochrome c synthase
sphingosine-1-phosphate lyase 1
fatty acid amide hydrolase
paraoxonase 2
dolichyl-P-Glc:Man9GlcNAc2-PP-
dolichylglucosyltransferase
duodenal cytochrome b
AA025006
T89379
AA459381
AA431988
AA446028
N63143
1.3
1.5
1.3
1.3
1.2
1.2
5.1
6.3
6.1
5.1
3.8
4.0
SGPL1
FAAH
PON2
ALG6
FLJ23462
AA457501
1.7
10.9
Cell Growth and Cell Death
inhibitor of DNA binding 2
ID2
H82442
R20666
1.4
1.3
4.4
4.7
endothelial differentiation, sphingolipid
G-protein-coupled receptor, 1
EDG1
cell growth regulatory with ring finger domain
RAN binding protein 6
RAN binding protein 2
transducer of ERBB2, 2
H2B histone family, member Q
kinesin-like 5 (mitotic kinesin-like protein 1)
HMBA-inducible
CGR19
RANBP6
RANBP2
TOB2
H2BFQ
KNSL5
HIS1
IGF2
MCM5
W89211
AA489790
R27450
1.4
1.7
1.2
1.9
1.4
1.5
1.3
1.7
1.2
9.2
5.7
3.9
9.2
7.6
4.0
7.3
6.5
4.6
AA486088
AA456298
AA454098
W68585
N54596
insulin-like growth factor 2 (somatomedin A)
minichromosome maintenance deficient
(S. cerevisiae) 5 (cell division cycle 46)
peroxisome proliferative activated receptor, alpha-like
cyclin-dependent kinase 9 (CDC2-related kinase)
CDC16 (cell division cycle 16, S. cerevisiae, homolog)
cyclin T2
W94630
PPARAL
CDK9
CDC16
CCNT2
CAPN7
MAP2K4
W96146
N80054
1.5
1.3
1.3
1.3
1.4
1.3
6.8
5.0
AA410604
R63702
4.0
4.5
calpain 7
N46420
AA293365
10.9
5.4
mitogen-activated protein kinase kinase 4
a Standardized coefficient, Z, defined as regression coefficient divided by its standard error, was used to make the statistical inference. The
higher the absolute value Z, the more significant the gene expression change is. The sign of the standardized coefficients reflects the directionality
of change. In this case, Z > 0 represent the increased gene expression, as compared to the untreated control.
nies suggest a role in DNA repair, mutagenesis, and apoptosis.
Among the genes whose expression levels were consistently up-
regulated was the XPA gene that encodes a DNA damage
recognition protein. Other genes up-regulated included cisplatin
resistance-associated protein that affect membrane permeability
of chemicals; antioxidant enzyme, apical protein (APXL); glu-
tathione transferase omega (GSTTLp28); and detoxification en-
zyme, HMT1 (methyltransferase)-like 1.
DISCUSSION
In recent years, the use of capillary liquid chromatography-
mass spectrometry for the analysis of DNA adducts has introduced
Analytical Chemistry, Vol. 78, No. 18, September 15, 2006 6431

a highly reliable methodology for their quantification. As these
methodologies continue to mature, new opportunities are emerg-
ing to further explore the broader implications of adduct formation
and their relationship to cell processes. In this paper, we report
on an improved method, based on the use of nanocapillary (75-
µm i.d., flow rates of 200 nL/min) liquid chromatography
combined with microelectrospray ionization mass spectrometry,
for the trace level analysis of dG-C8-4-ABP in human cells dosed
with N-OH-AABP. The present method has the advantage of
requiring a smaller injection volume and consumes the equivalent
of 2.50, as opposed to 13.3, µg of DNA per analysis when using
320-µm-i.d. columns (flow rate of 20 µL/min).21
continued to increase with time of exposure, suggesting that the
repair capacity of the cells was being overwhelmed, but was not
yet fully saturated. At the highest dose of 10 µM, the adduct levels
peaked at 2 h after treatment and achieved a steady-state adduct
level that was at least 15-fold higher than those seen at the lowest
dose.
Although further experimentation is necessary to fully explain
these observations, the data are consistent with the hypothesis
that once the repair capacity of the cell was exceeded, steady-
state levels of the N-OH-AABP adduct increased as a function
of a dose. However, the data also indicate that adduct levels did
not increase beyond the 2-h time point at the 10 µM dose,
suggesting that steady-state adduct levels up to ∼500 per 109 bases
were not compatible with cell survival. The latter contention is
supported by the cell survival data (Figure 7), which revealed a
dramatic decline in cell viability at a dose of 12.5 µM.
When viewed in unison, these results presented here provide
convincing evidence for the added value of anchoring gene
expression patterns to phenotypic markers, such as DNA adduct
levels, toxicity, and mutagenicity. Although the present study was
limited to a small number of doses and exposure time, genes
clearly showed differential responses to dose, with some being
maximally changed at the lowest doses, while others showed a
continuing dose response. Although the study lacked the statistical
power to definitively distinguish among these sets of genes, the
trends did suggest that many genes related to mutagenicity and
toxicity showed significant dose responses. Although phenotypic
anchoring of expression profiles to adduct levels could not identify
profiles that are predictive of toxicity and mutagenicity, they could
provide insights into the underlying mechanisms. Studies are
underway to validate the utility of these and other aggregate gene
expression patterns as predictive biomarkers.
In the present study, we combined this new method for
detection of dG-C8-4-ABP DNA adduct levels with data on the
toxicity, mutagenicity, and gene expression profiles induced in
TK 6 human lymphoblastoid cells exposed to different doses of
N-OH-AABP. By anchoring the observed biological outcomes
to DNA adduct levels, we hoped to gain further insight into how
DNA adduct formation and repair affect cellular responses to
mutagenic insults. Few studies have directly investigated the rates
of ABP DNA adduct formation and the repair rate in human cell
lines, especially in TK6 cells. Human B-lymphoblastoid TK6 cells
have functional wild-type P53 alleles (P53+/+) and, hence, should
be capable of mounting a normal response to DNA damage.
Consistent with the latter contention, TK6 cells have the capacity
to repair bulky adducts, such as benz[a]pyrene diol epoxide
(BPDE)-DNA with an adduct half-life of 12.3 h.39 Even though
TK 6 cells do not express the CYP1A enzyme,38 BPDE DNA
adducts can be detected in TK6 cells as early as 15 min after
treatment with BPDE.39 By contrast, previous work has demon-
strated that treatment of TK 6 cells with N-OH-AABP at
concentrations as high as 10 µM yielded no observable toxicity
at 1 h of treatment. Thus, to achieve 5, 15, and 40% toxicity at 0.5,
1, or 10 µM, treatment time was extended to 27 hour.22 Compari-
son of gene expression profiles induced by BPDE and N-OH-
AABP at doses yielding comparable levels of toxicity suggested
that differences in the abilities of TK6 cells to repair the adducts
induced by these two compounds might account for the almost
1000-fold difference in their mutagenicities and toxicities of these
directly acting mutagens.22 As a first step toward testing the latter
hypothesis, the present study compared the dG-C8-ABP DNA
adduct levels in TK6 cells exposed to the doses of N-OH-AABP
yielding 5, 15, and 40% toxicity.
The data discussed in this publication have been deposited
in NCBIs Gene Expression Omnibus (GEO, http://
www.ncbi.nlm.nih.gov/geo/) and are accessible through GEO
Series accession number GSE2345.
ACKNOWLEDGMENT
The authors thank Jimmy Flarakos for his helpful comments.
This research was supported by the National Institute of Health
(Grant Nos. RO1CA112231, U19ES011387, and P30ES07033). This
is publication number 883 from the Barnett Institute.
ABBREVIATIONS
The quantification of the dG-C8-ABP adduct content in the
human TK6 lymphoblastoid cells treated with N-OH-AABP
revealed several trends. After 2 and 6 h, there is no correlation
between dose and DNA adduct levels. However, 27 h after dosing
at all three levels, a positive correlation can be seen between
N-OH-AABP metabolite concentration and DNA adduct levels.
This observation suggests that although adducts can form im-
mediately, at the lowest dose (0.5 µM) the constitutive repair
capacity of the cells was probably able to maintain the steady-
state levels of the adduct at a background level over the 27 h of
treatment. At the intermediate dose of 1.0 µM, the adduct levels
4-ABP
4-aminobiphenyl
dG-C8-ABP
dG-C8-ABP-d9
MgCl2
N-(deoxyguanosin-8-yl)-4-aminobiphenyl
N-(deoxyguanosine-8-yl)-4-aminobiphenyl-d9
magnesium chloride
N-OH-AABP N-hydroxy-4-acetylaminobiphenyl
6TG
TK
6-thioguanine
thymidine kinase
(38) . Xia, F.; Wang, X.; Wang, Y. H.; Tsang, N. M.; Yandell, D. W.; Kelsey, K.
T.; Liber, H. L. Cancer Res. 1995, 55 (1), 12-15.
(39) . Luo, W.; Gurjuar, R.; Ozbal, C.; Taghizadeh, K.; Lafleur, A.; Dasari, R.;
Received for review April 18, 2006. Accepted June 14,
2006.
Zarbl, H.; Thilly, W. Chem. Res. Toxicol. 2003, 16, 74-80
AC0607360
6432 Analytical Chemistry, Vol. 78, No. 18, September 15, 2006