Carcinogenesis vol.20 no.6 pp.1085–1089, 1999
Use of UvrABC nuclease to quantify benzo[a]pyrene diol epoxide–
DNA adduct formation at methylated versus unmethylated CpG
sites in the p53 gene
Moon-shong Tang2, Jessica B.Zheng,
Mikhail F.Denissenko1, Gerd P.Pfeifer1 and Yi Zheng
Department of Carcinogenesis, University of Texas M.D. Anderson Cancer
Center, Science Park, Smithville, TX 78957, USA and 1Department of
Biology, Beckman Research Institute of the City of Hope, Duarte,
CA 91010, USA
2To
whom correspondence should be addressed
Email: [email protected]
We have used the UvrABC nuclease incision method in combination with ligation-mediated polymerase chain reaction
(LMPCR) techniques to map and quantify (K)anti-7β,
8α-dihydroxy-9α,10α-epoxy-7,8,9,10-tetrahydrobenzo[a]pyrene (BPDE) adduct formation in the p53 gene of human
cells. We found that BPDE adduct formation, as revealed
by UvrABC incision, preferentially occurred at methylated
CpG sites that correspond to the mutational hotspots
observed in human lung cancers. Our hypothesis is that
it is this methylated CpG sequence-dependent preferential
adduct formation, rather than selective growth advantage,
that is the major determinant of the p53 mutation pattern in
human cancers. Given the far reaching ramifications of such
conclusions for cancer etiology, a legitimate question is raised
regarding the reliability of using the UvrABC incision
method for quantifying and determining the sequencedependency of adduct formation. Is the higher frequency of
UvrABC cutting at methylated versus unmethylated CpG
sites due to the preference of the nuclease for cutting at those
sites or due to the preferential formation of BPDE adducts
at those sites? In order to distinguish between these two
possibilities, we have analyzed the kinetics of UvrABC
incision at BPDE adducts formed at either methylated CpG
sites versus other sequences, or unmethylated CpG sites
versus other sequences in exon 5 of the p53 gene. We have
found that the UvrABC cutting kinetics are identical for
both cases. On the basis of these results we conclude that
under proper cutting conditions, UvrABC nuclease reacts
with and incises with equal efficiency, BPDE adducts formed
at methylated or unmethylated CpG sites as well as other
sequences, and that the extent of UvrABC incision accurately
reflects the extent of BPDE–DNA adduct formation. These
conclusions were further supported by results obtained using
a DNA synthesis blockage assay.
Introduction
The mutational spectrum and signature mutations in the tumor
suppressor p53 gene in human cancers have revealed a wealth
of information regarding the etiology of cancer (1–3). This
information is important for cancer prevention and potentially
useful for cancer treatment. Interestingly, although .200 sites
Abbreviations: BPDE, (6)anti-7β,8α-dihydroxy-9α,10α-epoxy-7,8,9,10-tetrahydrobenzo[a]pyrene; DTT, dithiothreitol; LMPCR, ligation-mediated polymerase chain reaction.
of mutation have been identified in the p53 gene, 30% of total
mutations are located at codons 156, 157, 175, 245, 248, 249,
273 and 282, all of which (except codon 249) contain a
guanine in a CpG sequence, and it appears that different
cancers often share certain common mutational hotspots. There
are certain mutational hotspots which appear to be tissue
specific; for example, hotspot mutations at 175, 248 and 273
are observed in ovarian, brain, breast and stomach cancers and
leukemia/lymphoma. On the other hand, prostate cancers show
only one hotspot mutation at codon 273, and hepatoma cases
from the East Asia and sub-Sahara Africa areas have a single
hotspot mutation at codon 249 (1–3). Cigarette smokingrelated lung cancers, while having mutational hotspots at
codons 248, 273 and 282 that are commonly observed in other
cancers, also have one hotspot mutation at codon 157 that is
not a mutational hotspot in other cancers (1,2,4–6). The crystal
structure of the p53 protein has revealed that the amino acids
encoded by codons 175, 248, 273 and 282 are important for
DNA contact and/or pivotal for the integrity of the p53 protein
structure (6,7). However, it is not clear whether mutations at
these particular mutational hotspot codons contribute more
effectively to either loss of tumor suppressor function or gain
of oncogenic function than mutations in other codon sequences.
While it is possible that the specific effects of different
mutations on p53 gene function, either on its ability to suppress
cancer formation or in activating cancer formation, may be
tissue or cell dependent, it is also possible that the p53
mutational hotspots in different cancers may result from
different carcinogen exposures in different tissues and/or that
p53 genes from different tissues have different susceptibilities
for carcinogen binding (8). In order to test the latter possibility, we recently mapped the adduct distribution along the
p53 gene in normal human bronchial epithelial cells treated
with the activated cigarette smoke component, (6)anti-7β,8αdihydroxy-9α,10α-epoxy-7,8,9,10-tetrahydrobenzo[a]pyrene
(BPDE), by using the UvrABC nuclease incision method in
combination with ligation-mediated polymerase chain reaction
(LMPCR) techniques. We found that UvrABC nuclease
incisions preferentially occurred at sites corresponding to the
p53 mutational hotspots observed in human lung cancers (4).
Based on these results, and since we had shown previously
that under our conditions UvrABC nuclease incises BPDE–
DNA adducts specifically and quantitatively, we concluded
that BPDE adducts preferentially form at p53 mutational
hotspot sites in lung cancers (4,9). Our subsequent studies
have confirmed that: (i) all these preferred sites for BPDE
binding are guanine residues within CpG dinucleotides (10,11);
(ii) methylation of cytosine at such sites is the determining
factor for preferential binding (10,11); and (iii) repair of BPDE
adducts formed at these sites is significantly slower than at
other sequences (11). These results not only provide a molecular
link between a known carcinogen and lung cancer but also
lead us to hypothesize that it is this methylated CpG sequencedependent preferential adduct formation, rather than selective
1085
M.-s.Tang et al.
Fig. 1. UvrABC incision time course on methylated and unmethylated DNA
modified with BPDE. Methylated and unmethylated 59-32P-labeled 259 bp
fragments of exon 5 of the p53 gene were modified with the same
concentration of BPDE (5310–5 mg/ml) as described previously (8) and
then reacted with UvrABC for different lengths of time. The reactions were
stopped by phenol and ether extractions followed by ethanol precipitation.
The resultant DNAs were separated by denaturing 8% PAGE. Lanes 1 and
5–11, methylated DNA; lanes 2–4 and 12–18, unmethylated DNA. Meth,
CpG methylation; 1, with CpG methylation; –, without CpG methylation.
Codons containing a guanine in a CpG site are indicated with an asterisk at
the right of the panel. C* marks the methylated cytosine at CpG sites
indicated by a missing band in the C-specific reaction. G, GA and TC
represent Maxam–Gilbert sequencing reactions. UvrABC nuclease makes
dual incisions of seven bases 59 to, and four bases 39 to, a BPDE–guanine
adduct. Therefore, the position of the UvrABC incision band corresponding
to the adducted guanine is seven bases lower than the position of that
guanine in the Maxam–Gilbert guanine ladder. The 259 bp fragments
were obtained by amplification (20 cycles) of exon 5 region of the p53 gene
from p53 gene containing plasmid (pAT153P53π obtained from L.Crawford
and S.P.Tuck, Imperial Cancer Research Fund Laboratories, London, UK).
The plasmids were hybridized with two oligonucleotide primers 59-TGCCCTGCTTTCAACTCTGTCTCC-39 and 59-CCAGCCCTGTCGTCTCTCCAGCC-39 with the former labeled at the 59 end with [γ-32P]ATP according
to the method described previously (8). The 32P single-end labeled DNA
amplified fragments (259 bp) were purified through 8% PAGE. The 32P
single-end labeled DNA fragments were subjected to SssI methylase
treatment with S-adenosylmethionine to methylate all cytosines at CpG sites
according to vendor’s instructions (8). The UvrABC nuclease reactions were
carried out in a total volume of 25 µl containing 50 mM Tris–HCl pH 7.5,
10 mM MgCl2, 0.1 mM EDTA, 1 mM ATP, 100 mM KCl, 1 mM
dithiothreitol (DTT), 15 nM UvrA, 15 nM UvrB, 15 nM UvrC and substrate
DNA (2 nM). UvrA, UvrB and UvrC proteins were isolated from the
Escherichia coli K12 strain CH296 (recA, endA/F’lacIQ) carrying plasmids
pUNC45 (uvrA), pUNC211 (uvrB), pDR3274 (uvrC) (18). The proteins
were purified as described previously (19).
1086
Fig. 2. Kinetics of UvrABC incision on methylated and unmethylated DNA
modified with BPDE. Eighteen well-separated bands in Figure 1 were
quantified using Bio-Image Open Windows (Version 3 System) with a
Howtek Scanmaster 31 and whole band analysis software. The band
intensities were adjusted by the amount of sample applied onto the gel, and
then normalized to the band intensities at 60 min incubation. (A)
Quantification from methylated DNA; (B) quantification from unmethylated
DNA. The sequences of the codons are described in the right of the panel.
The asterisk represents the guanine residue modified with BPDE. The lines
in the panel represent the mean value of the different sequences (right of
the panel); for clarity, only four sequence points are presented in the panel
and the rest of the sequence points are within the length of the bar.
growth advantages, that is the basis of the p53 mutation
patterns seen in human cancers (8). These results also suggest
that an epigenetic factor, cytosine methylation at CpG sites, may
play a significant role in determining cancer susceptibility (8).
Obviously, the methodology used to obtain these results
merits careful evaluation, particularly with regard to the
question of whether UvrABC incision represents the sequencedependent adduct formation or whether it merely reflects the
sequence-dependent UvrABC incision preference. Does BPDE
preferentially bind guanines at methylated CpG sites or does
UvrABC preferentially incise BPDE–guanine adducts formed
at methylated CpG sites?
In order to resolve this issue, we have used analyses of the
UvrABC incision kinetics of adducts formed at different
sequences to assess whether there are significant sequencedependent and/or methylation-dependent adduct cutting
preferences by this enzyme system. We also further investigated the sequence dependence versus methylation dependence
of adduct formation by determining the relative effectiveness of blockage of Sequenase-mediated DNA synthesis on
BPDE-modified CpG-methylated versus unmethylated DNA
templates.
Although UvrABC nuclease is able to incise BPDE adducts
UvrABC incision of BPDE adducts at CpG sites in p53
Fig. 3. The effects of cytosine methylation at CpG sites on BPDE–DNA adduct formation. Twenty-four well-separated bands in lanes 10 and 17 of Figure 1
were quantified and normalized to the amount of 32P applied to the gel. The y-axis represents the relative intensity of UvrABC cutting of methylated DNA
(top) and unmethylated DNA (bottom). The codons with a guanine at CpG sites are indicated.
formed in plasmid DNA quantitatively (9), it had not previously
been determined whether this enzyme system would show
differences in the rate constant of cutting of adducts formed
at different sequences, or at methylated versus unmethylated
CpG sites. Although DNA fragments with a site-specific adduct
might represent ideal substrates for determining the role of
DNA sequence or cytosine methylation at CpG sites on an
adduct’s susceptibility for UvrABC cutting, the number of
sequences that would need to be tested makes it unfeasible to
take this approach. Furthermore, both the relatively short
lengths of oligonucleotides containing site-directed adducts
[which, for practical reasons, are generally limited to ~60 bp
(12,13)] and the purity of the constructs may adversely affect
the UvrABC cutting efficiency which consequently obscures
the determination of the DNA sequence effect per se on adduct
formation and adduct–enzyme affinity.
For these reasons, we used 259 bp DNA fragments PCR
amplified from exon 5 of the p53 gene. These DNA fragments
were 59-32P-labeled by using one-end labeled primer and
methylated at CpG sites by CpG methylase (8,10,11). Methylated and unmethylated DNA fragments were subsequently
modified with the same concentration of BPDE under conditions which produced an average of less than one adduct per
DNA fragment. The DNA fragments were then treated with
UvrABC under conditions in which the enzyme:DNA molar
ratio was ù6.
It is worth noting that there are no DNA polymerases and
ligases present under our reaction conditions and, therefore,
the results of UvrABC incision should be irreversible. We
previously have demonstrated that C5 cytosine methylation
results in much stronger UvrABC incision bands at CpG sites
in BPDE-modified DNA; the extent of enhancement appears
to be sequence dependent, and varies from 2- to 10-fold (8).
If this enhancement of UvrABC cutting at methylated CpG
sites is due to the enzymes having higher cutting efficiency
towards BPDE adducts formed at methylated CpG sites versus
unmethylated CpG or other sequences, then the kinetics of
UvrABC incision should be different for methylated CpG sites
and unmethylated CpG sites or other sequences. To test this
possibility, the time course of UvrABC cutting for each of
these different substrates was determined. A typical result is
shown in Figure 1, demonstrating that UvrABC incision at all
BPDE-binding sites, in both methylated and unmethylated
DNA, is a function of incubation time and appears to plateau
after 30 min of incubation.
Eighteen well-resolved bands representing UvrABC cutting
within the middle region of the DNA fragment, including eight
CpG sites and 10 non-CpG sites, were quantified. Since it has
been found that efficient UvrABC cutting requires a minimal
DNA length, no attempt was made to quantify DNA adducts
formed near the ends of the DNA fragments (14). Figure 2
shows the UvrABC cutting time courses for each of the 18
sites. These results demonstrate that 90% of UvrABC cutting
occurs during the first 30 min, and that the cutting kinetics
for both methylated CpG sites versus other sequences, and
unmethylated CpG sites versus other sequences are nearly
identical, although the final extent of cutting in methylated
CpG sites is higher than at their unmethylated counterparts
(Figure 3).
The relationship between UvrABC cutting and BPDE–DNA
adduct formation can be summarized as Pn 5 Kn·E·Sn, where
Sn represents the DNA adducts formed at any particular
sequence, Pn represents the resulting UvrABC cutting at this
sequence, and E represents the molar concentration of UvrABC.
There are two possible explanations for the different Pn that
we have observed at different sequences, and at methylated
versus unmethylated CpG sites; either Kn or Sn must be
sequence dependent and C5 cytosine methylation dependent.
If Kn is sequence/methylation dependent, one would expect
that after long incubation times, Pn would eventually be equal
for methylated, unmethylated CpG sites and other sequences,
since in our reaction conditions E . S. However, results in
Figure 1 show that: (i) after 30 min of incubation, UvrABC
cutting (P) for all sequences reached plateaus and (ii) even
after further incubation (90 min) the different extents of
UvrABC cutting for different sequences in both methylated
and unmethylated DNA remain the same as those observed at
30 min of incubation. Furthermore, the rate constants of
UvrABC cutting for methylated and unmethylated CpG sites
and other sequences are identical (Figure 2). These results rule
out the possibility that Kn is sequence and C5 methylation
dependent. Therefore, the different values of Pn for methylated
versus unmethylated CpG and for different sequences must
reflect differences in adduct formation; Sn must be C5
methylation and sequence dependent.
On the basis of these results we conclude that (i) the degree
1087
M.-s.Tang et al.
BPDE. These BPDE modified plasmid DNAs were used as
templates for determining Sequenase-mediated DNA synthesis
at an exon 8 region using 59-32P-labeled primers. The results
in Figure 4 show that CpG-methylated DNA templates resulted
in significantly more DNA synthesis blockages at CpG sites
(codons 273, 282, 283, 290 and 298) than observed with
unmethylated DNA templates. These results are consistent
with our previous finding that CpG methylation in exon 8
DNA fragments increases BPDE binding at CpG sites in
codons 273, 282, 283 and 290 (8).
It is possible that UvrABC cutting may show some small
degree of sequence-dependent preference. However, under the
conditions in which the enzyme–DNA adduct molar ratio is
excessive and in which the period of incubation time allows
the completion of .90% of cutting, such a small sequencedependent preference of UvrABC cutting should not affect the
validity of using enzyme cutting for adduct quantitation at the
sequence level.
Acknowledgements
We thank Dr Gerry Adair and Miss Yen-Yee Tang for critical review of this
manuscript. This research was supported by grants ES03124, ES08389
(M.-s.T.), CA65652 and CA69449 (G.P.P.) from the National Institutes
of Health.
References
Fig. 4. Determination of the position and the degree of DNA synthesis
blockage using BPDE modified DNA templates. Plasmid pAT153P53π
was methylated to completion with SssI DNA methylase or mock treated as
described (8). The plasmids were linearized with EcoRI, phenol extracted
and ethanol precipitated. Both methylated and unmethylated plasmids were
modified with BPDE at final concentrations of 0.5 or 10 mM in the dark for
90 min. Unreacted BPDE was removed by repeated diethyl ether and
isoamylalcohol extractions, and the DNA was ethanol precipitated. A
p53-specific primer (59-AAGAGGCAAGGAAAGGTGATA) was 59-32Plabeled with T4 polynucleotide kinase and [γ-32P]ATP. BPDE-modified
or unmodified plasmid (3 µg) was mixed with the 32P-labeled primer
(0.5 pmol) in 40 mM Tris–HCl pH 7.7, 50 mM NaCl in a volume of 15 µl,
heated to 95°C for 5 min, and then the primer was annealed at 48°C for
30 min. Nine microliters of 20 mM DTT, 20 mM MgCl2 and 0.25 mM
dNTPs was added together with 5 U of Sequenase 2.0 and the reaction was
incubated at 48°C for 20 min. Sequenase was heat inactivated at 67°C for
15 min, the DNA was ethanol precipitated, denatured in formamide loading
buffer and loaded onto an 8% polyacrylamide gel. The gel was dried and
exposed to X-ray film. The position of the guanine residue in the sequence
was identified by Maxam–Gilbert sequencing. The codon number is
depicted on the right and the asterisk indicates the CpG site.
of UvrABC incision (P) represents the extent of BPDE–DNA
adduct formation and (ii) enhancement of UvrABC cutting at
C5 methylated CpG sites is due to the preferential binding of
BPDE at methylated CpG sites, not due to more efficient cutting
by UvrABC at BPDE–DNA adducts formed at methylated CpG
sites.
To further strengthen these two conclusions, we have used
a DNA replication stoppage assay to demonstrate the effect
of methylation at CpG sites on BPDE–DNA binding. It has
been demonstrated that BPDE-modified guanines in the DNA
template strand block the progression of DNA chain elongation
(15–17). If BPDE binds more efficiently at guanines within
methylated CpG sites than to guanines at unmethylated CpG
sites or other sequences, then this different extent of BPDE
binding should also be reflected by the extent of DNA synthesis
blockage. Plasmid DNAs with a p53 gene insert were treated
with and without CpG methylase, and then modified with
1088
1. Greenblatt,M.S., Bennett,W.P., Hollstein,M. and Harris,C.C. (1994)
Mutations in the p53 tumor suppressor gene: clues to cancer etiology and
molecular pathogenesis. Cancer Res., 54, 4855–4878.
2. Hollstein,M., Shomer,B., Greenblatt,M., Soussi,T., Hovig,E., Montesano,R.
and Harris,C.C. (1996) Somatic point mutations in the p53 gene of human
tumors and cell lines: updated compilation. Nucleic Acids Res., 24,
141–146.
3. Ruddon,R.W. (1995) Cancer Biology. Oxford University Press, New
York, NY.
4. Denissenko,M.F., Pao,A., Tang,M.-s. and Pfeifer,G.P. (1996) Preferential
formation of benzo[a]pyrene adducts at lung cancer mutational hotspots
in p53. Science, 274, 430–432.
5. Hollstein,M., Sidransky,D., Vogelstein,B. and Harris,C.C. (1991) p53
mutations in human cancers. Science, 253, 49–53.
6. Levine,A.J. (1997) p53, the cellular gatekeeper for growth and division.
Cell, 88, 323–331.
7. Cho,Y., Gorina,S., Jeffrey,P.D. and Pavletich,N.P. (1994) Crystal structure
of p53 tumor suppressor–DNA complex: understanding tumorigenic
mutations. Science, 265, 346–355.
8. Chen,J.X., Zheng,Y., West,M. and Tang,M.-s. (1998) Carcinogens
preferentially bind at methylated CpG in the p53 mutational hot spots.
Cancer Res, 58, 2070–2075.
9. Tang,M.-s., Pierce,J.R., Doisy,R.P., Nazimiec,M.E. and MacLeod,M.C.
(1992) Differences and similarities in the repair of two benzo[a]pyrene
diol epoxide isomers induced DNA adducts by uvrA, uvrB and uvrC gene
products. Biochemistry, 31, 8429–8436.
10. Denissenko,M.F., Chen,J.X., Tang,M.-s. and Pfeifer,G.P. (1997) Cytosine
methylation determines hot spots of DNA damage in the human p53 gene.
Proc. Natl Acad. Sci. USA, 94, 3893–3898.
11. Denissenko,M.F., Pao,A., Pfeifer,G.P. and Tang,M.-s. (1998) Slow repair
of bulky DNA adducts along the nontranscribed strand of the human p53
gene may explain the strand bias of transversion mutations in cancers.
Oncogene, 16, 1241–1247.
12. Zou,Y., Liu,T., Geacintov,N.E. and Van Houten,B. (1995) Interaction of
the UvrABC nuclease system with a DNA duplex containing a single
stereoisomer of dG-(1)- or dG-(–)-anti-BPDE. Biochemistry, 34, 13582–
13593.
13. Mekhovich,O., Tang,M.-s. and Romano,L.J. (1998) Rate of incision of
N-acetyl-2-aminofluorene and N-2-aminofluorene adducts by UvrABC
nuclease is adduct- and sequence-specific: comparison of the rates of
UvrABC nuclease incision and protein–DNA complex formation.
Biochemistry, 37, 571–579.
14. Van Houten,B., Gamper,H., Sancar,A. and Hearst,J.E. (1987) DNase I
footprint of ABC excinuclease. J. Biol. Chem., 262, 13180–13187.
15. Hruszkewycz,A.M., Canella,K.A. and Dipple,A. (1991) DNA polymerase-
UvrABC incision of BPDE adducts at CpG sites in p53
mediated nucleotide incorporation adjacent to hydrocarbon-deoxyadenosine and hydrocarbon-deoxyguanosine adducts. Carcinogenesis, 12,
1659–1663.
16. Hruszkewycz,A.M., Canella,K.A., Peltonen,K., Kotrappa,L. and Dipple,A.
(1992) DNA polymerase action on benzo(a)pyrene–DNA adducts.
Carcinogenesis, 12, 2347–2352.
17. Reardon,D.B., Bigger,C.A.H., Strandberg,C., Yagi,H., Zerina,D.M. and
Dipple,A. (1989) Sequence selectivity of the reaction of optically active
hydrocarbon dihydrodiol epoxides with rat H-ras DNA. Chem. Res.
Toxicol., 2, 12–14.
18. Thomas,D.C., Levy,M. and Sancar,A. (1985) Amplification and purification
of UvrA, UvrB and UvrC proteins of Escherichia coli. J. Biol. Chem.,
260, 9875–9883.
19. Tang,M.-s., Nazimiec,M.E., Doisy,R.P., Pierce,J.R., Hurley,L.H. and
Alderete,B.E. (1991) Repair of helix-stabilizing anthramycin-N2 guanine
DNA adducts by UVRA and UVRB proteins. J. Mol. Biol., 220, 855–866.
Received September 15, 1998; revised February 18, 1999;
accepted March 1, 1999
1089