Spectroscopic characterization of the initial C8 intermediate in the reaction of the 2-fluorenylnitrenium ion with 2'-deoxyguanosine

Article abstract of DOI:10.1021/ja9836702

Irradiation of 2-azidofluorene in an aqueous solution containing 2'- deoxyguanosine (dG) gives good yields of 8-(2-fluorenylamino)dG. This is the C8 adduct implicated in the carcinogenicity of 2-aminofluorene, and formed in vivo from the reaction of DNA-guanine and an ester of N-hydroxy-2- aminofluorene. Flash photolysis of the azide reveals two intermediates on the pathway that forms this adduct, the 2-fluorenylnitrenium ion, and a subsequent longer-lived species formed in the reaction of this ion with dG. A number of pieces of evidence identify this later intermediate as the initial C8 adduct derived from addition of the nitrenium ion to the C8 carbon prior to loss of the C8 proton. Both spectroscopic and kinetic analyses show that the latter actually exists in both a cationic acid form and a base form, with a pK(a) for the acid of 3.9. The base form is a tautomer of the final product and is the species present at pH 7. There is also evidence that the reaction of the nitrenium ion and dG that forms this intermediate proceeds directly by addition at C8. In other words, the substitution of ArNH⁺ for H⁺ at the C8 position of 2'-deoxyguanosine is a straightforward electrophilic aromatic substitution.

Full text of DOI:10.1021/ja9836702

J. Am. Chem. Soc. 1999, 121, 3303-3310
3303
Spectroscopic Characterization of the Initial C8 Intermediate in the
Reaction of the 2-Fluorenylnitrenium Ion with 2′-Deoxyguanosine
Robert A. McClelland,* Abid Ahmad, Andrew P. Dicks, and Victoria E. Licence
Contribution from the Department of Chemistry, UniVersity of Toronto,
Toronto, Ontario M5A 3H6, Canada
ReceiVed October 19, 1998
Abstract: Irradiation of 2-azidofluorene in an aqueous solution containing 2′-deoxyguanosine (dG) gives good
yields of 8-(2-fluorenylamino)dG. This is the C8 adduct implicated in the carcinogenicity of 2-aminofluorene,
and formed in vivo from the reaction of DNA-guanine and an ester of N-hydroxy-2-aminofluorene. Flash
photolysis of the azide reveals two intermediates on the pathway that forms this adduct, the 2-fluorenylnitrenium
ion, and a subsequent longer-lived species formed in the reaction of this ion with dG. A number of pieces of
evidence identify this later intermediate as the initial C8 adduct derived from addition of the nitrenium ion to
the C8 carbon prior to loss of the C8 proton. Both spectroscopic and kinetic analyses show that the latter
actually exists in both a cationic acid form and a base form, with a pKa for the acid of 3.9. The base form is
a tautomer of the final product and is the species present at pH 7. There is also evidence that the reaction of
the nitrenium ion and dG that forms this intermediate proceeds directly by addition at C8. In other words, the
substitution of ArNH+ for H+ at the C8 position of 2′-deoxyguanosine is a straightforward electrophilic aromatic
substitution.
2-Aminofluorene has received extensive investigation since
it and its N-acetyl derivative were found to be potent carcinogens
over 50 years ago.1 These two compounds are representative
of a diverse class of carcinogenic aromatic and heteroaromatic
amines present in the environment in tobacco smoke, automobile
exhaust, broiled meats and fish, and as side products of industrial
processing.2 A common feature of the class is the transfer of
the arylamine group to DNA. The predominant site of attach-
ment on DNA is guanine. Within this base, the major adduct is
3 involving the C8 position attaching to the carcinogen’s
nitrogen (Scheme 1).1,3
Although these C8 adducts were first structurally character-
ized over 30 years ago, mechanistic details have been uncertain.
It is now generally accepted that bioactivation through oxidation
and (in most cases) O-esterification is required. The esters 1 so
formed have been generally assumed to react via an SN1
pathway, with an an electrophilic intermediate, the arylnitrenium
ion 2, as the species that actually reacts with guanine.4
Suggestions however have also appeared of a direct interaction
at the ester stage, in what is essentially an SN2 reaction of
guanine at the ester nitrogen.5 Only recently have experiments
from the Novak group conclusively established the nitrenium
Scheme 1
pathway. They found that hydrolysis of three examples of 1 in
the presence of 2′-deoxyguanosine (dG) gave substantial
amounts of the C8 adduct, but with no change in the rate
constant.6 This is classic evidence for the nucleophile reacting
with an intermediate and not in the rate-limiting step of the
reaction. A similar observation has recently been reported with
a DNA oligomer.7
Our group has corroborated this conclusion through laser flash
photolysis (LFP) experiments involving the same nitrenium
ions.8,9 In addition to the direct observation of these electro-
philes, the absolute rate constants kw and kdG were measured,
the former representing the decay of the cation in the solvent
alone and the latter, the reaction with added dG. One of the
* To whom correspondence should be addressed.
(4) Miller, J. A. Cancer Res. 1970, 30, 559-576. Kriek, E. Biochim.
Biophys. Acta 1974, 335, 177-203. Miller, E. C. Cancer. Res. 1978, 38,
1479-1496. Miller, E. C.; Miller, J. A. Cancer 1981, 47, 2327-2345.
Miller, J. A.; Miller, E. C. EnViron. Health Perspect. 1983, 49, 3-12.
Garner, R. C.; Martin, C. N.; Clayson, D. B. In Chemical Carcinogens,
2nd ed.; Searle, C. E., Ed.; ACS Monograph 182: American Chemical
Society: Washington, DC, 1984; Vol. 1, pp 175-276.
(1) Beland, F. A.; Kadlubar, F. F. In Handbook of Experimental
Pharmacology; Cooper, C. S., Grover, P. L., Eds.; Springer-Verlag:
Heidelberg, 1990; Vol. 94/I, pp 267-325. Kriek, E. J. Cancer Res. Clin.
Oncol. 1992, 118, 481-189. (c) Heflich, R. H.; Neft, R. E. Mutat. Res.
1994, 318, 73-174. Hoffman, G. R.; Fuchs, R. P. P. Chem. Res. Toxicol.
1997, 10, 347-359. Patel, D. J.; Mao, B.; Gu, Z.; Hingerty, B. E.; Gorin,
A.; Basu, A. K.; Broyde, S. Chem. Res. Toxicol. 1998, 11, 391-407.
(2) Weissberger, J. H. In Carcinogenic and Mutagenic Responses to
Aromatic Amines and Nitroarenes; King, C. M., Romano, L. J., Schueltze,
D., Eds.; Elsevier: New York, pp 3-19. Sugimura, T. Science 1992, 258,
603-607. Vineis, P. EnViron. Health Perspect. 1994, 102, 7-10. Layton,
D. W.; Bogen, K. T.; Knize, M. G.; Hatch, F. T.; Johnson, V. M.; Felton,
J. S. Carcinogenesis 1995, 16, 35-52.
(5) Novak, M.; Martin, K. A.; Heinrich, J. L. J. Org. Chem. 1989, 54,
5430-5431. Ulbrich, R.; Famulok, M.; Bosold, F.; Boche, G. Tetrahedron
Lett. 1990, 31, 1689-1692. Helmick, J. S.; Martin, K. A.; Heinrich, J. L.;
Novak, M. J. Am. Chem. Soc. 1991, 113, 3459-3466.
(6) (a) Novak, M.; Kennedy, S. A. J. Am. Chem. Soc. 1995, 117, 574-
575. (b) Kennedy, S. A.; Novak, M.; Kolb, B. A. J. Am. Chem. Soc. 1997,
119, 7654-7664.
(3) Dipple, A. Carcinogenesis 1995, 16, 437-441.
(7) Novak, M.; Kennedy, S. A. J. Phys. Org. Chem. 1998, 11, 71-76.
10.1021/ja9836702 CCC: $18.00 © 1999 American Chemical Society
Published on Web 03/25/1999

3304 J. Am. Chem. Soc., Vol. 121, No. 14, 1999
McClelland et al.
Table 1. Reactivities of Arylnitrenium Ions with Water and
Scheme 2
2′-Deoxyguanosinea
b
c
cation
kw (s-1
)
kdG (M-1 s-1
)
C6H5C6H4-4-NH+ (8)
1.8 × 106
2.7 × 105
2.7 × 104
1.6 × 103
2.0 × 109
1.5 × 109
7.6 × 108
3.6 × 107
4-MeC6H4C6H4-4-NH+ (9)
fluorenyl-2-NH+ (10)
4-MeOC6H4C6H4-4-NH+ (11)
a At 20 °C, in 20% acetonitrile, pH 7. b Reference 8d for (8)(10), 8e
for (8)(11). c Reference 8f.
Results
important observations to emerge was that the selectivities kdG
:
Generation of Nitrenium Ions. The 2-fluorenylnitrenium ion
10 was the principle electrophile employed in these studies;
however, to investigate structural effects, a few experiments
were performed with the biphenylyl derivatives 8, 9, and 11.
The rate constants kw and kdG for these ions are given in Table
1. All four show a high selectivity for dG. For example, the
nucleoside will trap 53, 85, 96, and 96% of 8-11, respectively
when its concentration is 1 mM. This means that the intermedi-
ate that arises in the nitrenium-dG reaction can be generated
in excellent yield even at low dG concentrations.
kw were in excellent agreement with those found by the Novak
group in their analysis of the products of the ground-state ester
hydrolyses.8c,f Thus, there can be no doubt that the transients
observed by LFP are ground-state arylnitrenium ions, that these
same species are formed in the ground-state hydrolysis of ester
precursors, and that these electrophiles do react with guanine
derivatives to form the C8 adduct.
The problem then arises that C8 is not regarded as the normal
position of electrophilic addition in guanine.3,10 Experimental
evidence for initial bond formation at N7 was claimed by
Humphreys, Kadlubar, and Guengerich (HKG). By employing
a system that generated the 2-fluorenylnitrenium ion in the
presence of 8-methylguanine derivatives, an unstable species
and its reduction product were obtained and characterized as
N7 adducts 4 and 5 (Scheme 2).10 Kennedy, Novak, and Kolb
(KNK) then performed similar experiments with 8-MedG and
precursors to both the N-acetyl and the parent 4-biphenylylni-
trenium ions. They also observed an unstable species and a
reduction product, but in this case assigned C8 structures 6 and
7.6b,11 These seemingly contradictory results are difficult to
reconcile, particularly considering that both groups included a
nitrenium ion ArNH+ reacting with 8-MedG. This uncertainty
thus leaves the question as to the initial site of attachment still
open.
In our LFP experiments with dG present, we observed a
growth of absorbance at 300-370 nm occurring with the same
rate as the decay of the nitrenium ion. Since there was no further
spectral change, we had assigned this growth to the formation
of the final product of the reaction. A lamp flash photolysis
apparatus however has revealed that such changes were occur-
ring, only at much longer times. This species therefore is an
intermediate, moreover, one that that forms in the reaction of
the nitrenium ion and dG. This paper addresses the structure of
this intermediate, and the implication for the mechanism for
the formation of the final C8 adduct.
All four cations were generated photochemically from azide
precursors.8b,d,e The intermediate that is initially formed is a very
short-lived (∼100 ps)8d singlet nitrene, and the nitrenium ion
arises by protonation by the solvent (eq 1). Pathways involving
ArN3
9
hυ8 1[ArN] H2O8 ArNH+
(1)
ring expansion to a didehydroazepine or formation of triplet
nitrene12 also compete, so that the yield of nitrenium-derived
products is not 100%.8d,e A detailed discussion of the photo-
generation of arylnitrenium ions via this approach is given in
previous publications.8b-f An important feature is that for the
4-biphenylyl- and 2-fluorenylnitrenium ions, there is a high
chemical and quantum yield even at neutral pH. This arises since
the singlet nitrene is quenched very efficiently by solvent.
C8 Adduct from Azidofluorene. Irradiation of 2-azidofluo-
rene in an aqueous solution containing dG gives rise to
substantial amounts of the C8 adduct derived from the 2-fluo-
renylnitrenium ion. The adduct was isolated and purified from
a scaled-up reaction and then employed as a standard in
quantitative HPLC analysis of experiments at varying dG
concentration (Figure 1).
The data were fit to eq 2, which pertains to a competition
(kdG:kw)[dG]
%C8 adduct ) (%C8max
)
(2)
(8) (a) Davidse, P. A.; Kahley, M. J.; McClelland, R. A.; Novak, M. J.
Am. Chem. Soc. 1994, 116, 4513-4514. (b) McClelland, R. A.; Davidse,
P. A.; Hadzialic, G. J. Am. Chem. Soc. 1995, 117, 4173-4174. (c)
McClelland, R. A.; Kahley, M. J.; Davidse, P. A. J. Phys. Org. Chem. 1996,
9, 355-360. (d) McClelland, R. A.; Kahley, M. J.; Davidse, P. A.; Hadzialic,
G. J. Am. Chem. Soc. 1996, 118, 4794-4803. (e) Ren, D.; McClelland, R.
A. Can. J. Chem. 1998, 76, 78-84. (f) McClelland, R. A.; Gadosy, T. A.;
Ren, D. Can. J. Chem. 1998, 76, 1327-1337.
(9) (a) Arylnitrenium ions have also been studied by LFP by Falvey and
co-workers. (b) Anderson, G. B.; Falvey, D. E. J. Am. Chem. Soc. 1993,
115, 9870-9871. (c) Robbins, R. J.; Yang, L. L.-N.; Anderson, G. B.;
Falvey, D. E. J. Am. Chem. Soc. 1995, 117, 6544-6552. (d) Srivasta, S.;
Falvey, D. E. J. Am. Chem. Soc. 1995, 117, 10186-10193. (e) Robbins, R.
J.; Laman, D. M.; Falvey, D. E. J. Am. Chem. Soc. 1996, 118, 8127-8135.
(f) Moran, R. J.; Falvey, D. E. J. Am. Chem. Soc. 1996, 118, 8965-8966.
(g) Srivasta, S.; Toscano, J. P.; Moran, R. J.; Falvey, D. E. J. Am. Chem.
Soc. 1997, 119, 11552-11553.
(
)
1 + (kdG:kw)[dG]
between dG and water for the nitrenium ion (Scheme 1), but
where the maximum yield of adduct is less than 100%. This
provided kdG:kw ) (2.3 ( 0.4) × 104 M-1, in good agreement
with the ratio from LFP (2.8 × 104, Table 1). The maximum
yield of C8 adduct is 81%. The 19% not accounted for probably
arises from the other reactions of the singlet nitrene, as just
discussed.
Laser Flash Photolysis. Growth of Intermediate. As shown
in Figure 2, the 2-fluorenylnitrenium ion is strongly absorbing
with a λmax at ∼450 nm. A smaller peak is seen at 335 nm, but
this is due to some other species formed from the singlet
nitrene.8d,e In the solvent alone, the OD decreases from 315 to
600 nm, with the same rate constant throughout. There is even
(10) Humphreys, W. G.; Kadlubar, F. F.; Guengerich, F. P. Proc. Natl.
Acad. Sci. U.S.A. 1992, 89, 8278-8282 and references therein.
(11) KNK still suggested N7 as the initial site of attack, on the basis of
a correlation of reactivity with N7 basicity.6b This latter argument breaks
down when imidazoles are included in the correlation.8f
(12) Schuster, G. B.; Platz, M. S. AdV. Photochem. 1992, 17, 69-143.

C8 Adduct in Reaction of dG and Nitrenium Ion
J. Am. Chem. Soc., Vol. 121, No. 14, 1999 3305
Figure 1. Yield of C8-adduct upon 300-nm irradiation (Rayonet) of
2-azidofluorene (20 µM) in 20% acetonitrile (0.002 M Na2HPO4:0.002
M NaH2PO4) containing varying amounts of 2′-deoxyguanosine. The
yield is based on the amount of azide reacted for irradiation times
sufficient to cause 30-40% conversion. The curve is drawn on the
basis of eq 2, using the parameters given in the text.
Figure 3. Kinetic traces at 450 and 316 nm for the solution of Figure
2 containing 0.0012 M dG. The insert shows the dependence of the
first-order rate constants measured at these two wavelengths on the
concentration of dG: (0) 450 nm, (b) 316 nm.
Figure 4. Spectra obtained on flash lamp excitation of the solution of
Figure 2 (with 4 µM 2-azidofluorene). (9) ∆OD obtained after the
lamp flash; (0) ∆OD after completion of the first-order kinetic process.
The inserts show these kinetic traces at 335 nm (rise) and 365 nm (fall).
The curves labeled FlN3 and C8 are the spectra (true ODs) in 20%
acetonitrile of equal concentrations (17 µM based on extinction
coefficients) of 2-azidofluorene and the final C8 adduct. The curve
Diff is a ∆OD, the OD of the C8 adduct minus the OD of the azide.
The curves for FlN3, C8, and Diff have been placed on the figure by
multiplying every absorbance by the same correction factor, calculated
so that the ∆OD for Diff at 330 nm is identical to the ∆OD at this
wavelength for the open squared point.
Figure 2. Spectra obtained on 308-nm laser irradiation of 2-azido-
fluorene (20 µM) in 20% acetonitrile (0.002 M Na2HPO4:0.002 M
NaH2PO4). (9) data after the laser pulse, (0) data at the completion of
the first-order process with 0.0012 M 2′-deoxyguanosine present, and
(O) data at the completion of the first-order process when there is no
2′-deoxyguanosine present.
a small decay at 315-350 nm, indicating that even here there
is a some nitrenium absorbance. At the completion of the decay,
there is little OD left, other than the band for the other species
at 335 nm.
The decay is accelerated with dG present, by a factor of 30
at the concentration of Figure 2. Now, however, there is a
significant growth of absorbance below ∼380 nm, with some
residual absorbance even out to 420 nm. This growth is observed
whenever there is sufficient dG to compete with the solvent. It
always occurs with the same first-order rate constant, within
experimental error, as that for the decay of the nitrenium ion
(Figure 3).
Growths producing similar final spectra are seen with 8, 9,
and 11 in the presence of dG, again with rate constants that are
identical to those measured for the decay at the λmax of the
nitrenium ion.
Lamp Flash Photolysis. Decay of Intermediate. The
experiments that demonstrate that the species whose growth is
observed by LFP is also an intermediate are shown in Figure
4. The curve labeled Diff in this figure is the spectrum of a
sample of the authentic C8 adduct 3 (labeled C8 in the Figure),
corrected by subtracting the spectrum for the same concentration
of 2-azidofluorene (FlN3). Since the flash photolysis experi-
ments measure the OD after irradiation minus the OD before,
Diff mimics what would be observed if the azide were replaced
by the ultimate product, the C8 adduct 3.13 Comparing this with
the spectrum of the LFP product shows that while both have
an apparent maximum near 330 nm, the C8 adduct does not
absorb beyond 350 nm, where the LFP product has a pronounced
shoulder. This led us to suspect that the latter was not the final
product, and we sought conditions where we could observe a
further reaction. This proved possible with lamp flash photolysis.
The further changes occur on the millisecond time scale, and
constitute a decrease in OD above 345 nm, with a quite
substantial increase below. The process is first-order, with the
(13) This applies to wavelengths above 310 nm only. Below this
wavelength, the dG present in the solution also absorbs.

3306 J. Am. Chem. Soc., Vol. 121, No. 14, 1999
McClelland et al.
Figure 5. Log k(obs)-pH profile for the reaction of the intermediate
from the 2-fluorenylnitrenium ion and 2′-deoxyguanosine. Condi-
tions: 20% acetonitrile, 20 °C, ionic strength 0.1 M (NaClO4). The
data below pH 3.5 were obtained in dilute HClO4 solutions, from pH
3.5-5.3 in acetate buffers, from pH 5.5-6.9 in cacodylate buffers,
and from pH 7-9 in Tris buffers. In the case of the buffer solutions,
the rate constants were obtained by extrapolating to zero buffer
concentration. The curve is drawn according to eq 3 (see Discussion)
using parameters given in the text.
Figure 6. Buffer data for the reaction of Figure 5. k(cat) is the slope
of the plot of k(obs) versus total buffer concentration at a given buffer
ratio, and R(A-) is the fraction of buffer in the base form. (9)
cacodylate, (0) acetate, (4) corrected acetate data where k(cat) has
been divided by R(B) (see discussion).
same rate constant in the increasing and decreasing regions. It
can be noted that the “initial” spectrum in Figure 4 is that of
the LFP product, since the fast process occurs within the ∼100
µs lamp flash. Although an exact match is not expected because
of the other photochemical products, the final spectrum is close
to that of the C8 adduct. Moreover, there is no further change,
as seen by recording spectra of the irradiated solutions on a
diode array spectrometer.
Kinetics of Decay of Intermediate. Kinetic experiments were
performed by following the millisecond time scale decay at 365
nm at 20 °C in 20% acetonitrile with the ionic strength at 0.1
M maintained with NaClO4. Initial experiments demonstrated
that rate constants measured in the same buffer solution were
independent of both the initial azide concentration (from 2 to
20 µM) and the initial dG concentration (from 100 µM to 2
mM). The later experiments then involved 4 µM azide and 500
µM dG. No OD change was observed in the absence of dG or
the absence of the azide.
Figure 7. Optical densities at completion of fast kinetic process on
308-nm laser irradiation of 2-azidofluorene (20 µM) in 20% acetonitrile
containing 1 mM 2′-deoxyguanosine. The main figure compares spectra
at pH 7 (9) and pH 3 (0). The insert shows the pH dependence of the
OD at 315 nm.
(a) Rate-pH Profile. The intermediate from the 2-fluor-
enylnitrenium ion was the one investigated in detail, and its
rate-pH profile is shown in Figure 5.
significant isotope effect is observed in all three: k8H:k8D
5.6 (A), 7.4 (B), and 6.1 (C).
)
C8-Deuterium Isotope Effect on the Nitrenium Ion Decay.
The isotope effect on kdG, the fast reaction of the nitrenium ion
with dG, was kdG-8H:kdG-8D ) 0.98 ( 0.02 (9), 0.955 ( 0.02
(10), and 0.88 ( 0.025 (11).
Spectroscopic Acidity Constant. As shown in Figure 7, there
is a slight difference in the intermediate’s spectra at pH 7 and
3. By working at 315 nm where the difference is quite
significant, a titration curve was obtained, and from this, an
acidity constant of (1.3 ( 0.2) × 10-4 M (pKa ) 3.9).
(b) Buffer Catalysis. The reaction is catalyzed by buffers,
in some cases quite significantly. The data were analyzed by
plotting k(obs) at a given buffer ratio versus total buffer
concentration. These plots were linear, and the slopes, k(cat),
were in turn plotted against R(A-), the fraction of the buffer in
the base form (see Figure 6 for two examples). For cacodylate,
phosphate, and Tris buffers, these plots follow the expected
linear relation. For acetate however, the plot is curved.
(c) Aryl Substituent Effect. The decrease in OD at 365 nm
(and increase at lower wavelengths) is also seen with the
intermediates from 8, 9, and 11. Rate constants for all four
intermediates were compared in three solutions: (A) 0.001 M
HClO4, (B) a pH 7.6 phosphate buffer containing 0.007 M Na2-
HPO4:0.003 M NaH2PO4, and (C) a pH 4.6 acetate buffer
containing 0.005 M HOAc:0.005 M NaOAc. In each solution
the rate constants for the four were within (5%: 1.45 × 103
s-1 (A), 2.7 × 102 s-1 (B), 4.9 × 102 s-1 (C).
Discussion
Kinetic Analysis. Since this has some impact on the question
of structure, we begin by presenting an analysis of the kinetics
of the reaction of the intermediate. There are two key observa-
tions here. The first is the isotope effect which shows that the
C8 proton is being removed in the rate-limiting step of the
reaction. The second is that the intermediate exists in acid:base
forms with a pKa of 3.9. Combining these with the observed
kinetic patterns gives rise to a mechanistic scheme where both
the acid and base forms are reacting, and moreover they each
(d) C8-Deuterium Isotope Effect. Rate constants for the
intermediates from dG and 8-deuterio-dG with the 2-fluor-
enylnitrenium ion were obtained in the same three solutions. A

C8 Adduct in Reaction of dG and Nitrenium Ion
J. Am. Chem. Soc., Vol. 121, No. 14, 1999 3307
Scheme 3
Scheme 4
Table 2. Rate Constants for the Reaction of the Intermediate from
the 2-Fluorenylnitrenium Ion and 2′-Deoxyguanosinea
catalyst
acetate
cacodylate
phosphate dianion
tris
kA(M-1 s-1
)
k+A(M-1 s-1
)
3 × 103
3.4 × 104
5.5 × 104
1.8 × 104
2.1 × 107
1.4 × 101
2.5 × 105
2.5 × 106
1.5 × 107
b
hydroxide
water
the pH profile are also included. The Bronsted plots using the
phosphate, cacodylate, acetate, and water rate constants (with
the water constant converted to second-order), are two parallel
lines with slopes of 0.6 separated by 2 log units. It is noteworthy
that the rate constants for the acid and base form differ by only
a factor of about 100-fold.
1.7 × 103
a 20% acetonitrile, 20 °C, ionic strength 0.1 M (NaClO4). b Acid
intercept not statistically significant. Units are s-1
.
do so with water, hydroxide, and the buffer base removing the
C8 proton (Scheme 3).
As mentioned at the start of this section the primary kH:kD is
an key aspect of this model. This effect was observed in three
solutions, and we can now provide a breakdown of the species
contributing in each. In the HClO4 solution (A), the rate constant
being measured represents 100% of the reaction of water with
the acid form, in the phosphate buffer (B), 99% of the reaction
of phosphate dianion with the base form, and in the acetate
buffer (C) about 50% each of water and acetate reacting with
the acid form, with only a small percentage (2%) of acetate
reacting with the base form. The observations of the primary
kH:kD in (A) and (C) are important since there is a kinetically
equivalent alternative to the processes responsible for the rate
constants k+w and k+A. This involves processes where H3O+ or
the buffer acid reacts directly with the base form. This possibility
however seems highly unlikely in terms of the kinetic isotope
effect. Particularly considering the nature of the final product,
this surely must be indicating a reaction where a base is
removing the C8 proton in the rate-limiting step.
Structure of the Intermediate. There are a number of
experimental observations that provide clues as to the structure
of the intermediate. (i) It has an absorption spectrum that extends
out to 400 nm, especially in its base form. (ii) Its reaction must
involve a mechanism where the C8 proton of guanine is being
transferred in the rate-limiting step. (iii) There is no effect of
the aryl substituent on the rate constant for its decay. (iv) It
exists in acid:base forms with a pKa of 3.9. (v) Both the acid
and the base forms are reactive.
The structure that best fits these criteria is the one where the
nitrenium ion has added directly to the guanine C8 carbon, so
that the acid form is the cation 12 and the base 13 is obtained
by loss of the N1 proton (Scheme 4). These structures are
consistent with the isotope effect (point ii), since both the acid
and the base form can proceed to product by simply losing the
C8 proton (point v). There should be little dependence on the
Ar substituent, which is well removed from the site of
deprotonation (point iii). Both forms are highly conjugated,
consistent with the absorption spectra (point i). In terms of the
pKa value (point iv), 12 is required to be a little over 5 log
units more acidic than parent dG. There is no direct comparison
to show whether this is reasonable. N7 alkylation is known to
increase acidity, as shown, for example, by 7-methyl-2′-
deoxyguanosine 14 (Scheme 5), which has a pKa of 6.4.15 This
is less than the effect observed in this study, but 12 might be
expected to be more somewhat more acidic, since its conjugate
base can be written as the charged neutralized structure 13. The
conjugate base of 14 must be written as a zwitterion (e.g., 15).
This provides the rate expression
k+w[H+] + (k+OHKw + ko Ka) + koOHKwKa/[H+]
w
k(obs) )
[H+] + Ka
(3)
The experimental date were fit to this equation with four
adjustable parameters including Ka. The values so obtained were
k+w ) 1.65 × 103 s-1, (k+OHKw + kw Ka) ) 1.53 × 10-3 M‚s-1
,
o
koOHKwKa ) 2.9 × 10-12 M2‚s-1, and Ka ) 1.25 × 10-4 M.
The kinetic Ka obviously agrees very well with the spectroscopic
one. In terms of the rate-pH profile, the acid region (pH <
5.5) is due to the acid form reacting with solvent (k+ term),
w
with the break at pH 4 as the equilibrium shifts from acid to
base. The dependence at high pH represents hydroxide ion
deprotonating the base (ko term). Kinetic ambiguity exists
OH
around pH 7, where there is a short region where the rate is
approximately pH independent (k+OHKw + kwoKa term), and
which could represent hydroxide reacting with the acid or water
with the base. In fact, it must be the latter. Even with k+
)
OH
1 × 1010 M-1 s-1, the contribution of k+OHKw to the observed
(k+OHKw + kwKa) is only 1%.14
For the buffer component, k(cat) is given by
k+AKHA
k(cat)
)
(1 - R(A-)) + ko R(A-)
(4)
A
(
)
Ka
R(B)
where KHA is the acidity constant of the buffer acid and R(B)
is the fraction of the intermediate in the base form. With pKa
) 3.9, R(B) is essentially unity for all pHs for the cacodylate,
phosphate, and Tris buffers, and the plots of k(cat) versus R(A-)
are linear. Only in acetate buffers does R(B) change, and thus
the k(cat) - R(A-) plot curves. As can be seen in Figure 6,
correcting by dividing by R(B) does produce a reasonably linear
plot.
According to eq 4 the plots of k(cat) or k(cat)/R(B) versus
R(A-) have an intercept at unit R(A-) of ko and an intercept
A
at zero R(A-) of k+AKHA/Ka. Table 2 provides the catalytic
coefficients so obtained, using pKa ) 3.9 to calculate k+
.
A
Values for the water and hydroxide rate constants obtained from
(14) The pKw in 20% acetonitrile is 14.8.8d
(15) Zoltewicz, J. A.; Clark, D. F.; Sharpless, T. W.; Grahe, G. J. Am.
Chem. Soc. 1970, 92, 1741-1750.
(16) Haines, J. A.; Reese, C. B.; Todd, L. J. Chem. Soc. 5281-5288.
Hecht, S. M.; Adams, B. L.; Kozarich, J. W. J. Org. Chem. 1976, 41, 2303-
2311.

3308 J. Am. Chem. Soc., Vol. 121, No. 14, 1999
McClelland et al.
Scheme 5
For the migration mechanism, the rate-limiting deprotonation
requires that the migration step be an equilibrium. The observed
rate constants in Table 2 would then be the product of an
equilibrium constant and a rate constant, i.e., k+A of Table 2 )
K+eqk+
(or kA ) Koeqkobase). Thus the actual k+
or ko
base
base base
would be larger than the values in Table 2 by a factor of 1/Keq.
To avoid processes occurring faster than diffusion, the smallest
value possible for Keq is then 10-3, since some kA are already
of the order of 107. The equilibrium condition is satisfied when
kbase[base], the rate constant for deprotonation, is significantly
less than kr, the rate constant for the reverse C8-to-N7 migration.
The largest value of kA(obs)[base] measured in this work was
1 × 104 s-1. Using a value of Keq of 10-3, the actual value of
kbase[base] is 1 × 107 s-1, and kr must then be greater than 108
s-1. Although such a rate constant is not impossible, it does
seem very large for a reaction in which the migrating group
must move from C8 by bonding with the N7 lone pair which is
actually pointing in the opposite direction. Finally, it can be
noted the larger rate constants for kbase seem inconsistent with
the Brønsted â value and the near maximal isotope effect, both
of which suggest a transition state for the proton transfer that
is close to being symmetrical. In summary, although the
migration mechanism cannot be unequivocally ruled out, it
seems unlikely, particularly when the above considerations are
coupled with the evidence from the pKa and spectra.
Scheme 6
An N7 structure must also be considered, particularly in light
of the HKG results. In this case the acid form is the cation 16
and the base a zwitterion such as 17 (Scheme 6). These two
structures are directly analogous to the N7-methylated 14 and
15, and we would argue that the latter system predicts a pKa
closer to 6. The arylamino at N7 in 16 is electron-withdrawing
relative to methyl, but it is difficult to imagine this effect
increasing the acidity by the 2.5 log units required to achieve
the pKa observed for the intermediate in this work. The
comparison with the N7-methylated system also shows that there
is a problem with the spectra, since neither 14 or 15 exhibit
much absorbance above 300 nm.16 There is obviously an
arylamino chromophore in 16, 17 as well, but the parent
arylamines also do not absorb much beyond 320-330 nm. Since
there is no conjugative interaction between the two chro-
mophores, their combination should not lead to absorbance out
to 400 nm. A caveat here is that there could be some form of
charge-transfer interaction.
Direct Conversion of Nitrenium Ion to C8 Intermediate.
There is further evidence in favor of the C8 intermediate in the
form of the small inverse isotope effect on kdG, the rate constant
for the reaction of the nitrenium ion and dG. The important
system here is the 4′-methoxy derivative. The 4′-methylbi-
phenylyl- and 2-fluorenylnitrenium ions were not expected to
show much of an effect, since these react at close to the diffusion
limit of 2-2.5 × 109 M-1 s-1 8f
. These were included in the
analysis so as to validate the experimental procedure. Indeed
their kH:kD values are close to unity, albeit slightly inverse at
0.98 and 0.955. The 4′-methoxy derivative, on the other hand,
reacts well below the diffusion limit (Table 1), and its kH:kD of
0.88 is significantly less than 1. An inverse secondary isotope
effect of this nature is the result expected for a reaction where
the isotopic substitution takes place on a carbon undergoing a
change of hybridization from sp2 to sp3. The implication is that
the nitrenium ion must be bonding at C8 in the transition state
when it reacts with dG. By combining the information on the
intermediate as discussed in the previous section, the most
consistent interpretation by far is found to be one where the
long-lived species is the C8 intermediate, and this forms directly
in the nitrenium-dG reaction.
There are two possible mechanisms to explain the isotope
effect if the intermediate were N7, and these are illustrated for
the acid form in Scheme 6. One is the suggestion by HKG,10
rate-limiting deprotonation at C8 to give a zwitterion, followed
by rapid transfer of the arylamino group. The second is the KNK
proposal,6b which has the arylamino group migrating from N7
to C8, followed by deprotonation.
a
The zwitterion mechanism appears to be immediately dis-
missable. Guanine derivatives exchange their C8 proton by a
mechanism that involves N7 protonation, followed by rate-
limiting deprotonation at C8 to give a zwitterion.17 The kinetics
require that hydroxide ion be the base, with no indication of a
process where water removes the C8 proton.17 Moreover, an
upper limit of 10-5 s-1 can be placed on such a water
Conclusions. Irradiation of 2-azidofluorene in aqueous solu-
tions containing dG results in significant yields of the C8 adduct
3, and flash photolysis has revealed two intermediates of the
reaction. One is the nitrenium ion and the second, longer-lived
species, is the C8 adduct prior to loss of the C8 proton. This
exists in a cationic acid form 12 and a neutral conjugate base
form 13, with a pKa of 3.9. The base is a tautomer of the final
product and is the species present at pH 7. The C8 intermediate
appears to form directly in the nitrenium:dG reaction. In other
words, substitution of ArNH+ for H+ at the C8 position of a
guanine derivative is a straightforward electrophilic aromatic
substitution.
deprotonation.18 This is 108 lower than the 103 value of k+
,
w
the rate constant that would represent water deprotonating 16
to form the zwitterion 18 if Scheme 6 were the mechanism.
(17) Tomasz, M.; Olson, J.; Mercado, C. M. Biochemistry 1972, 11,
1235-1241.
One of the arguments against electrophilic addition at C8 is
that this position is less nucleophilic than some of the others in
guanine. The present results, however, demonstrate that the
intermediate cation 12 is relatively very stable, at least in a
kinetic sense. This cation is directly analogous to the cyclo-
(18) At pH 3 and 37 °C, 1-methylguanosine (1-MeG) exchanges its C8
proton with a rate constant of 1.1 × 10-6 s-1 17
.
Since the pKa of
N7-protonated 1-MeG is 2.3,17 the maximum rate constant for water
deprotonating the cation to form a zwitterion is 7 × 10-6 s-1. This,
moreover, is an upper limit, since the kinetics shows no sign that such a
process is occurring.

C8 Adduct in Reaction of dG and Nitrenium Ion
J. Am. Chem. Soc., Vol. 121, No. 14, 1999 3309
in acetonitrile just before irradiation. Spectra were constructed point-
by-point, with three measurements being made and averaged at each
wavelength.
hexadienyl cation intermediates of electrophilic substitution of
benzene derivatives. A relevant comparison here is with 19,
whose rate constant for deprotonation by water is 5.8 × 105
Conventional Flash Photolysis. Experiments were performed using
an apparatus previously described.24 Samples were 4 µM substrate and
were irradiated with a broad band flash lamp of ca. 100-µs duration.
Quantitative Analysis of the C8 Adduct. Analysis (Figure 1) was
carried out with HPLC. The solution contained 20 µM 2-azidofluorene
and was irradiated in a Rayonet reactor at 300 nm for a time sufficient
to give 30-50% conversion. The HPLC conditions were as follows:
µ-Bondepak C18 column, 2 mL/min of 85:15 water:acetonitrile isocratic
for 7 min and then a linear gradient over 7 min to 20:80 water:
acetonitrile, monitoring wavelength ) 320 nm. The C8 adduct had a
retention time of 7 min, and the 2-azidofluorene, 12.5 min. The areas
of the peaks were converted to moles using a correction factor obtained
by injecting known amounts of the authentic samples. The yield of C8
adduct was calculated as moles C8 adduct divided by moles azide
reacted, where the latter was determined from the peaks before and
after irradiation.
s-1 19
This cation would be regarded as being relatively
.
stabilized, since it has three methoxy groups correctly positioned
to directly conjugate the three Cδ+ centers. Less stabilized
examples are much shorter-lived, and in fact require less basic
solvents for their observation by LFP.19,20 Even with the
stabilizing substituents, 19 is over 2 orders of magnitude more
short-lived than 12 (see k+w in Table 2). Thus, in terms of kinetic
stability, 12 appears as a highly stabilized “cyclohexadienyl”
cation. In this respect it is perhaps not so surprising that the
nitrenium ions can react readily at C8 to form such a cation.
Interestingly, recent experiments in our laboratories have shown
that the arylnitrenium ions of this study are quenched by 1,3,5-
trimethoxybenzene, with rate constants that are similar to kdG
.
Determination of Isotope Effect on Decay of Nitrenium Ion in
the Presence of dG. A large quantity of a stock buffer containing 20%
acetonitrile and 0.001 M Na2HPO4:0.001 M Na2HPO4 was prepared.
This was added to two 100-mL volumetric flasks, one of which
contained a weighed amount of 2′-deoxyguanosine, and the other
contained an equivalent amount of 8-deuterio-2′-deoxyguanosine.
Concentrations of the dG derivatives were in the range 0.5-2 mM.
The two solutions were divided in three parts (with a small amount
being set aside for HPLC analysis), and 10 µM of 4′-methyl-4-
azidobiphenyl, 2-azidofluorene, and 4′-methoxy-4-azidobiphenyl were
added to each from a 20-40 mM stock solution of the azides in
acetonitrile. Three solutions of the three azides in the stock buffer (no
dG) were also prepared. The nine solutions were analyzed by laser
flash photolysis with 308 nm irradiation, with the first-order rate
constants for the decay of the nitrenium ions being determined at the
following λmax: 490 nm (9), 450 nm (10), and 500 nm (11). Five to
six measurements were made with each solution (with the laser cuvette
being replenished with fresh solution after each pulse). The relative
dG concentration in the two solutions was accurately determined by
HPLC. The HPLC conditions were 95:5 water:acetonitrile at a flow
rate of 2 mL/min through a C18 column with the detector set at 270
nm. Exactly 80 µL of each the two solutions was injected alternatively,
with four (sometimes five) injections of each being made. The isotope
effect was calculated with the equation
The present results have implications for the reaction with
DNA-guanine. In particular the deprotonation of the intermediate
that occurs with the monomer at pH 7 involves a proton that is
involved in hydrogen bonding in the DNA double helix. We
also argue that the cationic form 12 acquires a considerable
stabilization by delocalization involving electrons on N1 and
the C2-nitrogen.21 These centers are again involved in hydrogen
bonding. In both respects it is therefore interesting that double-
stranded DNA appears to be very much less efficient at trapping
the N-acetyl-2-fluorenylnitrenium, as compared to both single-
stranded DNA and the monomer dG.7 A source of this effect
could be the inhibition of the resonance stabilization/deproto-
nation imposed by the double helix structure.
Experimental Section
The azide precursors were available from previous studies.8d,e 2′-
Deoxyguanosine was commercially available and used as received. This
was converted to its 8-deuterio form by refluxing in D2O overnight.22
The 1H NMR of this material showed it to be 99% deuterated
(integrating the C8-H signal versus the signals for the deoxyribose).
8-(2-Fluorenylamino)-2′-deoxyguanosine was obtained by irradiating
(300 nm) 2-azidofluorene (50 mg) in 500 mL of 20% acetonitrile
containing 0.5 g of 2′-deoxyguanosine. The volume of the solution was
then reduced to 10 mL, and the C8 adduct was isolated by preparative
HPLC (0.5-mL injections onto 25 × 250 mm C18 semipreparative
column, with 20:80 acetonitrile:water eluting at 20 mL/min, and eluant
collected from 25 to 30 min) followed by short column chromatography
(silica gel, 50:50 ethyl acetate:hexanes). The sample so obtained had
NMR23 and UV10 identical to those in the literature.
Laser Flash Photolysis. Experiments involved ca. 20-ns pulses at
308 nm (60-120 mJ per pulse) from a Lumonics excimer laser. A
pulsed Xenon lamp provided monitoring light. After passing through
a monochromator, the signal from the photomultiplier tube was digitized
and sent to a computer for analysis. Solutions were 10-20 µM azide
in 20% acetonitrile, with the substrate being added from a stock solution
kdecay(dG-8H) - kdecay(no dG)
(
)
)
kdG-8H
HPLC area(dG-8H)
kdecay(dG-8D) - kdecay(no dG)
HPLC area(dG-8D)
)
(5)
kdG-8D
(
where the rate constants kdecay were the average of the measurements
made in the three solutions, and the HPLC areas were the average of
the four or five injections. It can be noted that the solutions were
prepared in such a way that the concentrations of dG-8H and dG-8D
were very similar. We felt, however, that the HPLC method was more
accurate than the weights in determining the relative concentrations. It
can also be noted that the concentrations of dG were such that the rate
constant with dG was generally greater than 5 times faster than the
rate constant without dG, especially with 10 and 11. This minimizes
the error in subtracting the latter from the former.
In general, we found that the HPLC areas were determined with a
standard deviation of (1.2% and the rate constants with a standard
deviation of 1.5%. To tighten the error limits on the measurement, the
entire procedure was performed a total of 18 times. The numbers given
in the Results are the average of these, with 1 standard deviation.
Spectroscopic Acidity Constant. Solutions were prepared that
contained 1.00 mM 2′-deoxyguanosine, 20% acetonitrile, and various
combinations of phosphate, acetate, and formate buffers and HClO4 so
as to provide solutions of pH varying from 2.5 to 8. Immediately before
(19) Steenken, S.; McClelland, R. A. J. Am. Chem. Soc. 1990, 112,
9648-9649.
(20) Mathivanan, N.; Cozens, F.; McClelland, R. A.; Steenken, S. J. Am.
Chem. Soc. 1992, 114, 2198-2203.
(21) The delocalization involving the NH2 group explains why inosine
is less effective than dG at trapping nitrenium ions at C8.6b,8f
(22) Tsang, P.; Vold, R. R.; Vold, R. L. J. Magn. Reson. 1987, 71, 276-
283.
(24) Allen, A. A.; Kresge, A. J.; Schepp, N. P.; Tidwell, T. T. Can. J.
Chem. 1987, 65, 1719.
(23) Kriek, E.; Westra, J. G. Carcinogenesis 1980, 1, 459.

3310 J. Am. Chem. Soc., Vol. 121, No. 14, 1999
McClelland et al.
LFP, 2-azidofluorene was added from a stock solution in acetonitrile
so as to give a final concentration of 20 µM. This solution was subjected
to 308 nm LFP, with three traces being recorded at both 450 and 315
nm, with fresh solution being employed in each case. The time scale
at 315 nm was set such that the first-order increase had reached its
final value by the end of the trace. The insert to Figure 7 is this OD
plotted against the measured pH of the solution. The acidity constant
was obtained by fitting the data to the equation
period of time to ensure that the solution received the same laser dose.
The requirements were nonetheless verified by the traces recorded at
450 nm. The initial absorbance at this wavelength measures the amount
of nitrenium ion that is formed in the laser pulse, and this was indeed
constant ((5%). The rate constant for the decay provides a measure
of the extent of trapping, through the following equation:
k(obs with dG) - k(obs no dG)
fraction of reaction with dG )
(7)
k(obs with dG)
OD(base)Ka + OD(acid)[H+]
OD(obs) )
(6)
where k(obs with dG) and k(obs no dG) are the rate constants for the
solution with 1 mM dG and a solution at the same pH without dG.
Both were effectively constant at (8.0 ( 0.6) × 105 and (3.0 ( 0.4) ×
104 s-1 from pH 3 to pH 8, respectively, so that throughout this pH
range ∼96% of the cation is trapped by dG. Below pH 3, the rate
constant with dG decreases, presumably because the nucleoside is being
converted into its protonated form.6 In consequence, the OD measure-
ments could not be extended into more acidic solutions.
Ka + [H+]
where OD(base) and OD(acid) are the optical densities of the base form
and acid form, respectively. The former was set as the average of the
three values at pH > 6.5. Since the flat region in acid could not be
reached, the latter was left as an adjustable parameter along with Ka in
the fitting process.
This experiment requires that the concentration of the intermediate
be the same at each pH, which in turn requires that the same amount
of nitrenium ion be produced and that it be trapped to the same extent
by dG. To ensure that this was the case, there was the same
concentration of the reagents in every solution, the concentration of
dG was such that a substantial fraction of the nitrenium ion reacted
with the nucleoside, and the experiment was carried out over a short
Acknowledgment. This paper is dedicated to Keith Ingold
on the occasion of his 70th birthday. The continued financial
support of the Natural Sciences and Engineering Research
Council of Canada is gratefully acknowledged.
JA9836702