Reactivity and selectivity of the N-acetyl-Glu-P-1, N-acetyl-Glu-P-2, N-acetyl-MelQx, and n-acetyl-IQx nitrenium ions: Comparison to carbocyclic N-arylnitrenium ions

Article abstract of DOI:10.1021/ja0121944

The model ultimate carcinogens 1a-d, related to the metabolites of the food-derived carcinogenic heterocyclic amines Glu-P-1, Glu-P-2, MelQx, and IQx, spontaneously decompose in neutral aqueous solution to generate the heterocyclic nitrenium ions, 2a-d. The less reactive esters 1a and 1b also undergo acid-catalyzed ester hydrolysis to generate the corresponding hydroxamic acids at pH ≤2, while the more reactive 2c and 2d are prone to rearrangement in nonaqueous solvents. The reactions of the nitrenium ions with AcO^-, HPO₄^2-, N₃^-, and 2′-deoxyguanosine (d-G) were characterized in aqueous solution by using a combination of competitive trapping methods and product isolation and identification. The reactions with N₃^- and d-G generally follow patterns previously established for carbocyclic nitrenium ions, but the reactions with AcO^- and HPO₄^2- are unusual. Similar reactions have previously only been reported for heterocyclic 1-alkyl-2-imidazolium ions. The N₃^-/solvent selectivities of these ions (5.1 × 10⁶ M^-1 for 2a, 2.3 × 10⁶ M^-1 for 2b, 1.2 × 10⁵ M^-1 for 2c, and 5.2 × 10⁴ M^-1 for 2d) are comparable to those of highly selective carbocyclic nitrenium ions. If k_az for these ions is diffusion limited at ca. 5 × 10⁹ M^-1 s^-1 the aqueous solution lifetimes of these ions range from 10 μs (2d) to 1 ms (2a). These ions are also highly selective for trapping by d-G, but comparisons to other nitrenium ions show that they are 10- to 50-fold less selective for trapping by d-G than they would be if both the N₃^- and d-G reactions were diffusion limited. This is not a consequence of their heterocyclic structures. Several carbocyclic ions show similar behavior. The relatively inefficient trapping of 2c and 2d by d-G may account for the observation of the unusual minor N-2 d-G adduct that is isolated for both of these nitrenium ions, but has not previously been observed for the reactions of other nitrenium ions with monomeric d-G.

Full text of DOI:10.1021/ja0121944

Published on Web 06/12/2002
Reactivity and Selectivity of the N-Acetyl-Glu-P-1,
N-Acetyl-Glu-P-2, N-Acetyl-MeIQx, and N-Acetyl-IQx Nitrenium
Ions: Comparison to Carbocyclic N-Arylnitrenium Ions
Michael Novak,* Krisztina Toth, Sridharan Rajagopal, Michael Brooks,
Lora L. Hott, and Matthew Moslener
Contribution from the Department of Chemistry and Biochemistry, Miami UniVersity,
Oxford, Ohio 45056
Received September 18, 2001. Revised Manuscript Received April 29, 2002
Abstract: The model ultimate carcinogens 1a-d, related to the metabolites of the food-derived carcinogenic
heterocyclic amines Glu-P-1, Glu-P-2, MeIQx, and IQx, spontaneously decompose in neutral aqueous
solution to generate the heterocyclic nitrenium ions, 2a-d. The less reactive esters 1a and 1b also undergo
acid-catalyzed ester hydrolysis to generate the corresponding hydroxamic acids at pH <2, while the more
reactive 2c and 2d are prone to rearrangement in nonaqueous solvents. The reactions of the nitrenium
ions with AcO-, HPO42-, N3-, and 2′-deoxyguanosine (d-G) were characterized in aqueous solution by
using a combination of competitive trapping methods and product isolation and identification. The reactions
-
with N3 and d-G generally follow patterns previously established for carbocyclic nitrenium ions, but the
reactions with AcO- and HPO4 are unusual. Similar reactions have previously only been reported for
2-
heterocyclic 1-alkyl-2-imidazolium ions. The N3-/solvent selectivities of these ions (5.1 × 106 M-1 for 2a,
2.3 × 106 M-1 for 2b, 1.2 × 105 M-1 for 2c, and 5.2 × 104 M-1 for 2d) are comparable to those of highly
selective carbocyclic nitrenium ions. If kaz for these ions is diffusion limited at ca. 5 × 109 M-1 s-1 the
aqueous solution lifetimes of these ions range from 10 µs (2d) to 1 ms (2a). These ions are also highly
selective for trapping by d-G, but comparisons to other nitrenium ions show that they are 10- to 50-fold
-
less selective for trapping by d-G than they would be if both the N3 and d-G reactions were diffusion
limited. This is not a consequence of their heterocyclic structures. Several carbocyclic ions show similar
behavior. The relatively inefficient trapping of 2c and 2d by d-G may account for the observation of the
unusual minor N-2 d-G adduct that is isolated for both of these nitrenium ions, but has not previously been
observed for the reactions of other nitrenium ions with monomeric d-G.
-
Introduction
ks, the ratio of the second-order rate constant for N3 trapping
of the nitrenium ion and the pseudo-first-order rate constant
for trapping the ion by the aqueous solvent.3,4
Mutagenic and carcinogenic heterocyclic amines (HCAs) are
found in parts per billion concentrations in broiled and fried
meats and other protein-containing foods, as well as in tobacco
smoke.1 A list of selected carcinogenic HCAs is presented in
Scheme 1 along with the common metabolic pathway required
for activation into their ultimate carcinogenic forms, the sulfuric
or acetic acid esters of the corresponding hydroxylamines.1,2
The HCAs are classified structurally into two groups: the IQ
type HCAs that are characterized by an imidazoquinoline,
imidazoquinoxaline, or related structure, and the non-IQ type
HCAs that are typically characterized by a 2-aminopyridine
structure.1 We have previously shown that ester derivatives of
two of the more weakly mutagenic non-IQ HCAs, Phe-P-1 and
ArC, decompose in aqueous solution to generate nitrenium ions
with low to moderate N3-/solvent selectivity measured as kaz/
Glu-P-1, Glu-P-2, IQx, and MeIQx are synthetically easily
accessible examples of the more highly mutagenic and carci-
(2) Sugimura, T. Mutat. Res. 1985, 150, 33-41. Felton, J. S.; Knize, M. G.;
Shen, N. H.; Anderson, B. D.; Bjeldames, L. F.; Hatch, F. T. EnViron.
Health Perspect. 1986, 67, 17-24. Hashimoto, Y.; Shudo, K.; Okamoto,
T. Acc. Chem. Res. 1984, 17, 403-408. Shinohara, A.; Saito, K.; Yamazoe,
Y.; Kamataki, T.; Kato, R. Cancer Res. 1986, 46, 4362-4367. Aoyama,
T.; Gonzalez, F. J.; Gelboin, H. V. Mol. Carcinogen. 1989, 2, 192-198.
Snyderwine, E. G.; Wirth, P. J.; Roller, P. P.; Adamson, R. H.; Sato, S.;
Thorgiersson, S. S. Carcinogenesis 1988, 9, 411-418. Meerman, J. H. N.;
Ringer, D. P.; Coughtrie, M. W. H.; Bamforth, K. J.; Gilissen, R. A. H. J.
Chem.-Biol. Interact. 1994, 92, 321-328. Raza, H.; King, R. S.; Squires,
R. B.; Guengerich, F. P.; Miller, D. W.; Freeman, J. P.; Lang, N. P.;
Kadlubar, F. F. Drug Metab. Dispos. 1996, 24, 385-400. Watanabe, M.;
Ishidate, M.; Nobmi, T. Mutat. Res. 1990, 234, 337-348. Wakabayashi,
K.; Nagao, M.; Esumi, H.; Sugimura, T. Cancer Res. 1992, 52, 2 (Suppl.),
2092s-2098s. Turesky, R. J.; Bracco-Hammer, I.; Markovic, J.; Richli,
U.; Kappeler, A.-M.; Welti, D. H. Chem. Res. Toxicol. 1990, 3, 524-535.
Turesky, R. J.; Markovic, J.; Bracco-Hammer, I.; Fay, L. B. Carcinogenesis
1991, 12, 1847-1855. Rich, K. J.; Murray, B. P.; Lewis, I.; Rendell, N.
B.; Davies, D. S.; Gooderham, N. J.; Boobis, A. R. Carcinogenesis 1992,
13, 2221-2226. Snyderwine, E. G.; Schut, H. A. J.; Adamson, R. H.;
Thorgeirsson, U. P.; Thorgeirsson, S. S. Cancer Res. 1992, 52 (Suppl.)
2099s-2102s. Schut, H. A. J.; Snyderwine, E. G. Carcinogenesis 1999,
20, 353-368.
* Corresponding author. E-mail: novakm@muohio.edu. Phone: 513-
529-2813. Fax: 513-529-5715.
(1) Eisenbrand, G.; Tang, W. Toxicology 1993, 84, 1-82. Hatch, F. T.; Knize,
M. G.; Felton, J. S. EnViron. Mol. Mutagen. 1991, 17, 4-19. Layton, D.
W.; Bogen, K. T.; Knize, M. G.; Hatch, F. T.; Johnson, V. M.; Felton, J.
S. Carcinogenesis 1995, 16, 39-52. Sugimura, T. EnViron. Health Perspect.
1986, 67, 5-10.
(3) Novak, M.; Xu, L.; Wolf, R. A. J. Am. Chem. Soc. 1998, 120, 1643-
1644.
9
7972
J. AM. CHEM. SOC. 2002, 124, 7972-7981
10.1021/ja0121944 CCC: $22.00 © 2002 American Chemical Society

Reactivity and Selectivity of Nitrenium Ions
A R T I C L E S
Scheme 1
Scheme 2
have not been able to isolate 3a-d presumably because of their
high reactivity with H2O. Our previous studies have shown that
the N-acetyl group of the hydroxamic acid esters can stabilize
the esters toward hydrolysis by N-O bond heterolysis by a
factor of 5 × 103-105, but the N-acetyl group has little effect
on the reactivity and selectivity of the resulting nitrenium ion.11
Typically, kaz/ks for the N-acetyl ion is only 2- to 10-fold smaller
than kaz/ks of the N-H ion, and reaction products are always
analogous.11 The data presented in this paper, along with those
previously gathered for Phe-P-1 and ArC, allow us to make
comparisons of the reactions of heterocyclic and carbocyclic
nitrenium ions, particularly with respect to selectivity for
reaction with N3- and 2′-deoxyguanosine (d-G). The results also
allow us to extend our studies into the range of the highly
selective heterocyclic nitrenium ions derived from the strongly
mutagenic HCAs.
nogenic members of both structural groups. Glu-P-1 and
Glu-P-2 are pyrolysis products of glutamic acid.1 They have
been detected in the pyrolysis mixtures of casein,5 in the oil of
charred egg yolk, a commercial food product in Japan,6 and in
Worcestershire sauce,7 among other sources.1 IQx and MeIQx
are products of the reaction of amino acids, reducing sugars,
and creatine at temperatures above 100 °C.1 They are found in
broiled and fried meats and fish, and derived products such as
beef extract and bouillon cubes.1
Results and Discussion
These amines cause cancers in the livers, small and large
intestines, forestomachs, and other organs of laboratory rats and
mice, and are presumed human carcinogens.1 Glu-P-1 binds to
DNA in the presence of rat liver microsomes to form the familiar
C-8 guanine adduct common to a wide variety of carbocyclic
and heterocyclic arylamines.1,8 MeIQx generates both the C-8
adduct and the less commonly observed N-2 adduct.9
The syntheses of 1a-d are described in the Supporting
Information. The Glu-P derivatives 1a and 1b are stable in the
absence of H2O, but the IQx derivatives 1c and 1d undergo
rearrangement into 4c and 4d in nonaqueous solvents, including
In this paper we present results of our study of the hydrolysis
reactions of 1a-d and of the subsequent chemistry of the
nitrenium ions 2a-d (Scheme 2). The esters 1a-d are acetic
acid esters of the N-acetylhydroxamic acid derivatives of the
four amines. It would be preferable to have data for the acetic
acid esters of the corresponding hydroxylamines, 3a-d, since
these are the putative carcinogens. Both 3a and 3b have been
synthesized in situ for DNA modification studies, but were not
isolated.10 Syntheses of 3c and 3d have not been reported. We
DMF used for stock solutions. Half-lives for rearrangement in
DMF at -40 °C were ca. 12-24 h, so fresh stock solutions
were prepared daily. Both compounds also underwent slow
rearrangement when stored neat at -40 °C so they were
synthesized in small batches and used within several days of
synthesis.
(4) Novak, M.; Kazerani, S. J. Am. Chem. Soc. 2000, 122, 3606-3616.
(5) Yamaguchi, K.; Zenda, H.; Shudo, K.; Kosuge, T.; Okamoto, T.; Sugimura,
T. Gann 1979, 70, 849-850.
(6) Kato, T.; Kikugawa, K.; Asanoma, M.; Sakabe, Y. Mutat. Res. 1990, 240,
259-266.
(7) Manabe, S.; Kanai, Y.; Yanagisawa, H.; Tohyama, K.; Ishikawa, S.;
Kitagawa, Y.; Wada, O. Cancer Res. 1987, 47, 6150-6155.
(8) Hashimoto, Y.; Shudo, K. EnViron. Health Perspect. 1985, 62, 209-214,
215-218.
(10) Hashimoto, Y.; Shudo, K.; Okamoto, T. J. Am. Chem. Soc. 1982, 104,
7636-7640.
(11) (a) Novak, M.; Kahley, M. J.; Lin, J.; Kennedy, S. A.; Swanegan, L. A. J.
Am. Chem. Soc. 1994, 116, 11626-11627. (b) Novak, M.; Lin, J. J. Org.
Chem. 1999, 64, 6032-6040.
(9) Turesky, R. J.; Rossi, S. C.; Welti, D. H.; Lay, J. O., Jr.; Kadlubar, F. F.
Chem. Res. Toxicol. 1992, 5, 479-490.
9
J. AM. CHEM. SOC. VOL. 124, NO. 27, 2002 7973

A R T I C L E S
Novak et al.
Scheme 3
Table 1. Kinetic Parameters for 1a-da
pK
a
compd
UV titration
kinetic fit
104ko (s-1
)
105kH (M-1 s-1
)
Figure 1. Log kobs vs pH for 1a (2), 1b (4), 1c (b), and 1d (O). The lines
are the least-squares fit for each data set to eq 1.
1a
1b
1c
1d
4.21 ( 0.05
4.17 ( 0.08
1.02 ( 0.19
4.04 ( 0.03
3.82 ( 0.06
1.21 ( 0.07
0.49 ( 0.03
5.1 ( 0.2
3.2 ( 0.2
8.1 ( 0.5
12 ( 1
199 ( 4
129 ( 2
The kinetics of the decomposition of 1a-d were monitored
at 20 °C by UV or HPLC methods in the pH range from 0.5 to
7.5 in 5 vol % CH3CN/H2O at ionic strength 0.5 (NaClO4) in
Na2HPO4/NaH2PO4, NaOAc/HOAc, and NaHCO2/HCO2H buff-
ers at pH >2.5 and in HClO4 at pH e2.5. Repetitive wavelength
scans for 1a and 1b showed that isosbestic points held for at
least 5 half-lives at pH >1.5, but did not hold at pH e1.5, while
for 1c and 1d isosbestic points held throughout the pH range
studied (Supporting Information). It was possible over the entire
pH range to monitor the reactions at wavelengths that allowed
a good fit of absorbance vs time data to the first-order rate
equation for all four compounds. Rate constants were found to
be independent of buffer concentration (0.005-0.08 M) or buffer
identity, but were dependent on pH (Figure 1 and Supporting
Information). Rate constants calculated by initial rates methods
from HPLC data confirmed that the rate constants observed at
pH e1.5 for 1a or 1b corresponded to those for their dis-
appearance.
a Conditions: 5 vol % CH3CN-H2O, T ) 20 °C, µ ) 0.5 (NaClO4).
Parameters were obtained by a fit to eq 1 or to a standard titration curve.
hydroxylamines.3,4 The term is consistent with the rate-limiting
spontaneous decomposition of the neutral form of the ester to
generate a cation that is trapped by solvent or other species in
a subsequent fast step (Scheme 3).3 The value of ko for 1a is
1.6-fold greater than that for 1b, while ko for 1c is 1.5-fold
greater than that for 1d. This is consistent with the expected
stabilizing effect of the Me substituents on the transition state
for formation of the cations.3,4
The first term of the rate law (kH(10-pH)2/(Ka + 10-pH)) is
consistent with acid-catalyzed decomposition of the conjugate
acids of 1a and 1b (Scheme 3). The site of protonation has not
been experimentally determined, but it is expected that N5 would
be the most basic site for 1a and 1b based on our observation
that the bulky substituents on N2 would disrupt solvation of
protonated N1.4 A reasonable explanation of this term would
be acid-catalyzed hydrolysis of the ester moiety of 1H+ to
generate the hydroxamic acid as its conjugate acid 5H+. The
hydroxamic acid would undergo hydrolysis under acidic condi-
tions to generate the corresponding hydroxylamine.12,13 Indeed,
6a and 6b are the major observed products of the acid-catalyzed
decomposition of 1a and 1b, respectively. This is consistent
with our observation that authentic 6a does not undergo the
Bamberger rearrangement under these conditions.
The kinetics were analyzed in terms of the mechanism of
Scheme 3 for 1a and 1b and a truncated version of that scheme
for 1c and 1d. The observed rate constants for 1a and 1b were
fit to eq 1 to obtain the parameters provided in Table 1. For 1c
and 1d only the second term of eq 1 was necessary to adequately
kobs ) (kH(10-pH)2 + Kako)/(Ka + 10-pH
)
(1)
fit the data. The apparent pKa values calculated from the kinetic
fits corresponded closely (Table 1) to those measured by fitting
initial absorbance for 1a, 1b, and 1c vs pH to a standard titration
equation. For 1c this required careful extrapolation back to
t ) 0 because of its rapid decomposition. Spectrophotometric
titration was not attempted for 1d.
The intermediacy of 5a under acidic conditions was confirmed
by a combination of kinetic and HPLC studies. The reaction is
(12) Novak, M.; Bohham, G. A.; Mohler, L. K.; Peet, K. M. J. Org. Chem.
1988, 53, 3903-3908.
(13) Ghosh, K. K.; Ghosh, S. J. Org. Chem. 1994, 59, 1369-1374. Ghosh, K.
K. Indian J. Chem. 1997, 36B, 1089-1102.
The second term of the rate law (Kako/(Ka + 10-pH)) was
observed previously for other ester derivatives of heterocyclic
9
7974 J. AM. CHEM. SOC. VOL. 124, NO. 27, 2002

Reactivity and Selectivity of Nitrenium Ions
A R T I C L E S
Table 2. Rate Constants for the Decomposition of 1a and 5a
Measured in 1.64 M HClO4 at 20 °C
Scheme 4
+
+
compd
method
105kH/[H ] (s-1
)
104kH′/[H ] (s-1
)
1a
UV (348 nm)
HPLC (256 nm)
UV (348 nm)
HPLC (256 nm)
HPLC (256 nm)
4.6 ( 0.8
6.7 ( 0.7
1.6 ( 0.3
5a
1.13 ( 0.03
1.19 ( 0.09
1.4 ( 0.3
5aa
6.5 ( 0.3
a These data refer to 5a detected as an intermediate by HPLC during the
decomposition of 1a.
inconveniently slow in the pH range, but in 1.64 M HClO4
(Ho ) -0.67)14 the hydrolysis of 1a proceeds with a half-life
of ca. 3.5 h. Under these conditions the decomposition of 1a
and authentic 5a were monitored by UV and HPLC methods.
At 348 nm the UV data for 1a were fit to an equation for two
consecutive first-order processes. The rate constants collected
in Table 2 confirm that one of these processes (kH) is associated
with the decomposition of 1a, while the other (kH′) is equivalent
to the rate constant measured for the decomposition of 5a under
the same conditions. Indeed, 5a could be detected by HPLC
during the decomposition of 1a under these conditions. Rate
constants for its appearance and disappearance, determined by
a fit of HPLC peak area vs time data to the consecutive first-
order rate equation, match those obtained for authentic 1a and
5a (Table 2).
The amines Glu-P-1 and Glu-P-2 were also generated as
minor reaction products of 1a and 1b under acidic conditions
(ca. 25%). The source of these products is not known, but
control experiments show that the amines were not derived from
5a,b or 6a,b under these conditions.
Although 1c and 1d are protonated under acidic pH condi-
tions, they do not exhibit acid-catalyzed hydrolysis at low pH
(Figure 1). This appears to be due to their much greater intrinsic
reactivity via the nitrenium ion path (ko for 1c and 1d is ca.
40-fold greater than ko for 1a and 1b, respectively) and the
smaller pKa values of their conjugate acids (1cH+ and 1dH+
are ca. 103-fold more acidic than 1aH+ and 1bH+, respectively).
These factors cause 1c and 1d to decompose ca. 104-fold more
rapidly than would 1a and 1b at pH <1 via the nitrenium ion
pathway. Since the acid-catalyzed hydrolysis pathway for 1c
and 1d is apparently not accelerated relative to that for 1a and
1b, it is not observed in the pH range of this study.
Although rate constants for the decomposition of 1a-d were
independent of buffer concentrations in phosphate and acetate
buffers, both of these buffers were incorporated into reaction
products (Scheme 4). These products were formed with the same
rate constants with which 1a-d decomposed (Figure 2). The
equilibrium between the acetate products derived from 1a and
1b, 9 and 10, was established rapidly on the time scale of the
hydrolysis reaction, but the equilibrium was slow enough that
9 and 10 appeared as separate products during HPLC, although
the two peaks tail toward each other. Attempts to isolate pure
9 or 10 by chromatographic methods always led to a mixture
of the two isomers. The approximate equilibrium constant, [10]/
[9], was ca. 2 for both 10a and 9a, and 10b and 9b. The acetate
products 4c and 4d derived from 1c and 1d are the same
materials formed by rearrangement of 1c and 1d. Fresh stock
solutions of 1c or 1d contain very little 4c or 4d, and the yield
of 4 can be shown to increase with increasing [AcO-], becoming
the only product at high buffer concentrations. These products
are not observed in phosphate buffers, confirming their inter-
molecular origin.
The phosphate adducts 8a and 8b were derived from 1a and
1b, respectively, while 12c and 12d were obtained from 1c and
1d, respectively (Scheme 4). Phosphate and acetate adducts were
identified primarily by 1H and 13C NMR spectroscopy, including
COSY, HSQC, HMBC, and DEPT experiments. The cis-
stereochemistry of 8, 9, and 10 was established by modeling
with MOPAC (AM1) coupled with observed H,H coupling
constants for H7, H8 and H8, H9. One stereoisomer was detected
for 12. The cis-structure (Scheme 4) is assumed. Structural
analyses, including stereochemical analyses, are provided in the
Supporting Information.
Although only the cis-isomers of 8, 9, and 10 were detected,
small amounts of the trans-isomers (e5% of the yields of the
cis-isomers) could have escaped detection. Nonetheless, the
(14) Markham, A. E. J. Am. Chem. Soc. 1941, 63, 874-875. Yates, K.; Wai,
H. J. Am. Chem. Soc. 1964, 86, 5408-5413.
9
J. AM. CHEM. SOC. VOL. 124, NO. 27, 2002 7975

A R T I C L E S
Novak et al.
Figure 3. Product yields vs [phosphate]T for the decomposition of 1b in
pH 6.1 phosphate buffer (fb ) 0.25): 7b (b) and 8b (9).
Figure 2. Time course for the disappearance of 1b (2) and the appearance
of 7b (b) and 8b (9) at pH 4.92 in a 0.01 M NaH2PO4/Na2HPO4 buffer.
All data were fit to a first-order rate equation to obtain an average rate
constant, kobs, of (3.0 ( 0.3) × 10-4 s-1. The calculated kobs for 1b at this
Scheme 5
pH, based on eq 1 and the data in Table 1, is (3.0 ( 0.2) × 10-4 s-1
.
reaction is remarkably diastereoselective for the cis-isomers. The
AM1 calculations suggest that the origin of this selectivity is
not product stability. The cis-isomers do tend to be more stable
than the trans-isomers according to the calculations, but not by
sufficient amounts to influence transition state stability to the
extent required to account for the results. For example, the heat
of formation of the most stable conformer of cis-9b is calculated
to be only 0.84 kcal/mol more negative than the heat of
formation of the most stable conformer of trans-9b at the AM1
level.
In the absence of buffers the 8,9-diol, 7, is the major product
formed from 1a and 1b. Only one isomer was detected, but
overlapping 1H NMR peaks make assignment of coupling
constants problematic. The diol 7 is also subject to decomposi-
tion after removal from aqueous solution so detailed structural
assignments are difficult to make. The relative stereochemistry
at C8 and C9 is left unassigned for these reasons. Under all
conditions a small amount of the reduction product 13 (ca. 5%)
can be detected in reaction mixtures of 1a and 1b. The
mechanism of the formation of this reduction product is
unknown. The diol 7, as well as 8, 9, and 10, could serve as
hydride donors to the cation 2 to generate 13. Greater yields of
13 were observed under conditions designed for product
isolation (ca. 1 × 10-6 to 5 × 10-5 M in 1 for kinetics and
HPLC studies, ca. 5 × 10-4 to 1 × 10-3 M for product
isolation). Under these conditions higher concentrations of the
potential hydride donors are available so increased yields of 13
does not interfere with product studies performed promptly
within 10 half-lives of the decomposition of 1c (ca. 350 s) or
1d (ca. 540 s).
Significant reaction of carbocyclic nitrenium ions with acetate
or phosphate buffers in buffers of moderate concentration (0.01
M to 0.02 M) has not been observed, but McClelland and co-
workers have previously reported that 1-alkyl-2-imidazolyl-
nitrenium ions, 14, react with phosphate to generate products
15 that are analogous to the phosphate product 12 reported in
this study (Scheme 5).15 In the case of 14 both cis- and trans-
products were detected.15 The structure of 15 and the discovery
that the rate of disappearance of 14 generated by laser flash
photolysis is dependent on phosphate concentration indicate that
15 is directly generated by attack of phosphate on 14.15
Product studies show that the yields of the phosphate-
containing product 8 or 12 are dependent on buffer concentra-
tion, and are produced at the expense of the solvent-derived
products 7 or 11 as buffer concentrations increase (Figure 3).
The instability of 1c and 1d complicated product studies, but
the problem was overcome by initiating all studies in a given
series simultaneously from the same freshly prepared stock
solution. The trapping ratio kphosT/ks was calculated from the
product yield data as a function of total phosphate as previously
described for other nucleophiles in the cases of the nitrenium
ions derived from Phe-P-1 and ArC.3,4 These ratios were found
to be linearly dependent on the fraction of HPO42- in the buffer
-
(Figure 4). The components of the trapping by H2PO4 and
a
HPO42- were determined by fitting kphosT/ks to eq 2, where kphos
/
ks is the trapping ratio for H2PO4 and H2O, kphosb/ks is the
-
2-
corresponding ratio for HPO42-, and fb is the fraction of HPO4
in the buffer. The trapping ratios are reported in Table 3. The
would be expected. The solvent-derived products isolated from
the reactions of 1c and 1d are 11c and 11d. The same products
are derived from the spontaneous hydrolysis of 4c and 4d that
are minor impurities in 1c and 1d, but this reaction is slow
enough (kobs for 4c at pH 7 is (1.0 ( 0.1) × 10-6 s-1) that it
kphosT/ks ) kphosa/ks + (kphosb - kphosa)fb/ks
(2)
2-
results show that HPO4 is at least a 10-fold more efficient
-
trap for 2 than its conjugate acid. The ratio kAcO /ks was
9
7976 J. AM. CHEM. SOC. VOL. 124, NO. 27, 2002

Reactivity and Selectivity of Nitrenium Ions
A R T I C L E S
Table 3. Rate Constant Ratios for Trapping 2a-d by Phosphate, N3-, and d-Ga
ratio
2a
2b
2c
2d
-1
-
kAcO /ks (M
)
280 ( 80
kphosa/ks (M-1
)
)
40 ( 1
58 ( 6
7 ( 4
20 ( 10
kphosb/ks (M-1
550 ( 5
520 ( 10
520 ( 20
440 ( 50
b
kaz/kphos
(9.3 ( 0.2) × 103
(5.1 ( 0.1) × 106
91 ( 4
(4.4 ( 0.2) × 103
(2.3 ( 0.1) × 106
60 ( 1
230 ( 30
120 ( 10
kaz/ks (M-1
)
(1.2 ( 0.2) × 105
1.6 ( 0.5
(5.2 ( 0.5) × 104
2.1 ( 0.4
b
kd-G/kphos
k
d-G/ks (M-1
)
(5.0 ( 0.2) × 104
(3.1 ( 0.1) × 104
(8.4 (2.6) × 102
(9.1 ( 2.1) × 102
a Rate constant ratios are defined in the text or in Schemes 4 and 6.
Scheme 6
Figure 4. kphosT/ks vs fb for 1b in phosphate buffer.
calculated for 2c by monitoring the yield of 4c as a function of
[AcO-] in pH 4.8 acetate buffer. That value is also included in
Table 3.
The cations 2a-d are trapped by N3- and 2′-deoxyguanosine,
d-G. Products are summarized in Scheme 6. These products
are formed in phosphate concentrations sufficiently high to
swamp out the solvent-derived products (0.02-0.20 M). The
yields of N3- trapping products exceed 95% in 0.02 M pH 6.9
Na2HPO4/NaH2PO4 buffer at [N3-], well below 0.01 M for all
four cations. The d-G trapping is less efficient, but still exceeds
70% of reaction products at 0.02 M d-G under these buffer
conditions for all four ions. This trapping occurs without rate
acceleration (Table 4). The rate constant ratios kaz/ks and kd-G
/
ks can be obtained indirectly by monitoring product yields as a
function of nucleophile concentration in buffers of constant
phosphate concentration sufficiently high that the yield of 7 or
11 is insignificant (Figure 5). The ratio of second-order rate
constants kaz/kphosb or kd-G/kphosb can be determined from these
data since [HPO42-] is constant and the reactions are run under
b
Alternatively, kaz/kphos can also be obtained by monitoring
-
product yields as a function of [HPO42-] at constant [N3
]
buffer conditions in which the contribution of H2PO4 is
-
negligible. The nucleophile/solvent selectivities can then be
obtained by multiplication. The rate constant ratios are sum-
marized in Table 3. The values of kaz/ks are somewhat larger
for the Me-substituted cations 2a or 2c than for 2b or 2d,
respectively. The higher N3-/solvent selectivity of 2a and 2c is
expected if kaz is diffusion limited because those ions should
react with solvent more slowly than their unsubstituted ana-
logues.
sufficiently high that the solvent-derived product is not observed.
This was done for 1b to obtain kaz/kphos of (5.3 ( 1.1) × 103.
b
The resulting value of kaz/ks ((2.7 ( 0.4) × 106 M-1) is in good
agreement with the value reported in Table 3.
The C-8 adducts 16 are generated by the reaction of all four
cations with d-G. These adducts are analogous to those identified
from the reaction of other heterocyclic and carbocyclic nitrenium
ions with d-G.3,4,16 They are the most commonly identified DNA
(15) McClelland, R. A.; Panicucci, R.; Rauth, A. M. J. Am. Chem. Soc. 1985,
107, 1762-1763. McClelland, R. A.; Panicucci, R.; Rauth, A. M. J. Am.
Chem. Soc. 1987, 109, 4308-4314. Bolton, J. L.; McClelland, R. A. J.
Am. Chem. Soc. 1989, 111, 8172-8181. Gadosy, T. A.; McClelland, R.
A. J. Am. Chem. Soc. 1999, 121, 1459-1465.
(16) (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. (c) McClelland, R. A.; Ahmad, A.; Dicks, A. P.; Licence, V.
E. J. Am. Chem. Soc. 1999, 121, 3303-3310.
9
J. AM. CHEM. SOC. VOL. 124, NO. 27, 2002 7977

A R T I C L E S
Novak et al.
Table 4. Rate Constants for the Decomposition of 1a-d in the
respectively. This is another characteristic reaction of nitrenium
ions.18
The N3 adducts 19, 20, and 21 are also characteristic
-
Presence and Absence of N3 and d-G
104kobs (s-1
)
-
conditions
1a
1b
1c
1d
substitution products similar to those previously reported for
other heterocyclic and carbocyclic nitrenium ions.3,4,11,19 Both
2a and 2b generate the 3-N3 product 19, but only 2b gives a
detectable yield of the 9-N3 product 20b in the N3- concentration
no added nucleophilea 5.35 ( 0.01 3.28 ( 0.15 205 ( 4 121 ( 4
-
b
0.01 M N3
5.99 ( 0.03 3.75 ( 0.32 195 ( 6 142 ( 2
5.14 ( 0.04 3.21 ( 0.02
0.01 M d-Gc
0.02 M d-Gd
180 ( 9 107 ( 4
range used in our product studies (5 × 10-6 to 2 × 10-3 M).
-
a Conditions: 5 vol % CH3CN-H2O, T ) 20 °C, µ ) 0.5 (NaClO4) in
0.02 M 2/1 Na2HPO4/NaH2PO4 buffer, pH 6.9. b Calculated % trapping by
N3- exceeds 98% for 2a-d. c Calculated % trapping by d-G exceeds 97%
for both 2a and 2b. d Calculated % trapping by d-G exceeds 70% for both
2c and 2d.
The dependence of the yields of 19b and 20b on [N3
]
(Supporting Information) suggests that the initial addition
intermediates produced by attack of N3- on C9 or C3 of 2b can
yield either final product. A significant portion of the yield of
20b is apparently generated at the expense of 19b at higher
[N3-], and the yield of 19b decreases from a high of around
82% at 6 × 10-5 M N3- to about 68% at [N3-] > 1 × 10-3 M.
The yield of 19b remains constant at that level up to at least 2
× 10-2 M N3-. A quantitative fit of the product yield data,
provided in the Supporting Information, suggests that ca. 67%
of the initial attack on 2b occurs on C9. The C3 product, 19b,
predominates at all [N3-] because the addition intermediate
-
derived from attack of N3 on C9 preferentially generates 19b
even at very low [N3-]. No product equivalent to 20b was
detected during the N3- trapping studies of 2a, but at very high
-
N3 concentrations ([N3-] g 0.01 M) small amounts of the
diazide product 22a were detected. This shows that attack at
C9 also occurs for 2a, although the extent of such attack cannot
be quantitatively assessed.
-
Figure 5. Product yields for 8a (9) and 19a (1) as a function of [N3
]
during the decomposition of 1a. Buffer conditions are the following: 0.16
M phosphate buffer, fb ) 0.67, pH 6.9.
adducts from in vitro and in vivo experiments involving the
amines or their metabolites.1,8 The adducts 16a,b are stable at
neutral pH, but are subject to hydrolysis of the C-N glycosidic
bond under acidic pH conditions to generate 17a,b. The
hydrolytic stability of 16c,d was not examined. In addition, 2c,d
also generate the unusual N-2 adducts 18c,d. Only C-8 adducts
have previously been reported from the reactions of most
heterocyclic and carbocyclic esters with monomeric d-G,3,4,16
but MeIQx and IQ derivatives have been shown to generate
both C-8 and N-2 adducts from reaction with monomeric d-G.9
Typically N-2 adducts are reported for most ester derivatives
of N-arylhydroxylamines only when they react with native
DNA.17 The C-8/N-2 adduct ratio is 3.5/1.0 for 2c. The C-8/
N-2 ratio observed for the reaction of NOHMeIQx with d-G
in the presence of Ac2O in 0.1 M pH 7.0 phosphate buffer was
11/1.9 The reported C-8/N-2 ratio in calf thymus DNA treated
similarly is 5.2/1, indicating that formation of the N-2 adduct
is more favorable in DNA than in d-G even for the IQ type
HCAs.9
-
N3 attack on most carbocyclic nitrenium ions such as
N-acetyl-4-biphenylylnitrenium ion, 23a, and N-acetyl-2-fluo-
renylnitrenium ion, 24a, occurs predominately at a position ortho
to the nitrenium ion center if the para position is blocked.19,20
-
The apparent predominant attack of N3 on 2b at C9 shows
that the regiochemistry of trapping reactions on heterocyclic
nitrenium ions can be profoundly affected by the structure of
the heterocyclic ring. The ions 2a and 2b in several of their
reactions behave more like highly conjugated iminium ions than
as typical nitrenium ions. This tendency to react with N3 at
conjugated sites other than the ortho carbon is also shared with
-
-
the N-acetyl-4-stilbenylnitrenium ion, 25, that reacts with N3
predominately at the â-vinyl carbon (85%) with only a minor
amount of reaction occurring at the ortho carbon (13%).21 This
ion also reacts with H2O to generate 1,2-diol products across
the R- and â-vinyl carbons much like those observed for 2a
The reactions of 2a-d with phosphate and acetate to produce
the addition products 8, 9, 10, and 12 are unusual, and suggest
that 2a-d behave differently from most carbocyclic nitrenium
ions, but 2a-d still possess sufficient nitrenium ion character
to react with d-G at the exocyclic N. Additionally, both 2a and
2b are efficiently reduced by I- to generate 13a and 13b,
and 2b, but it did not react significantly with AcO- or HPO4
.
2- 21
The N3- trapping data provide a lower limit for the lifetime
of 2a-d in aqueous solution since kaz cannot exceed the
diffusion limit of ca. 5 × 109 M-1 s-1 established for a wide
(18) Pelecannou, M.; Novak, M. J. Am. Chem. Soc. 1985, 107, 4499-4503.
(19) Novak, M.; Kahley, M. J.; Eiger, E.; Helmick, J. S.; Peters H. E. J. Am.
Chem. Soc. 1993, 115, 9453-9460.
(20) (a) Davidse, P. A.; Kahley, M. J.; McClelland, R. A.; Novak, M. J. Am.
Chem. Soc. 1994, 116, 4513-4514. (b) Novak, M.; Kahley, M. J.; Lin, J.;
Kennedy, S. A.; James, T. G. J. Org. Chem. 1995, 60, 8294-8304.
(21) Novak, M.; Kayser, K. J.; Brooks, M. E. J. Org. Chem. 1998, 63 5489-
5496.
(17) Kriek, E. Cancer Res. 1972, 32, 2042-2048. Westra, J. G.; Kriek, E.;
Hittenhausen, H. Chem.-Biol. Interact. 1976, 15, 149-164. Kriek, E.;
Hengeveld, G. M. Chem.-Biol. Interact. 1978, 21, 179-201. Meerman, J.
H. N.; Beland, F. A.; Mulder, G. J. Carcinogenesis 1981, 2, 413-416.
Beland, F. A.; Dooley, K. L.; Jackson, C. D. Cancer Res. 1982, 42, 1348-
1354.
9
7978 J. AM. CHEM. SOC. VOL. 124, NO. 27, 2002

Reactivity and Selectivity of Nitrenium Ions
A R T I C L E S
Table 5. Comparison of Log(kaz/ks) and Log(kd-G/ks) for
Carbocyclic and Heterocyclic Nitrenium Ionsa
ion
log(kaz/ks)
log(kd-G/ks)
2ab
6.71
6.36
5.08
4.72
2.97
3.45
2.89
3.60
3.50
3.79
3.85
4.53
6.59
4.76
5.07
2.73
3.68
2.48
4.65
4.54
4.70
4.49
2.92
2.96
2.51
3.04
2.15
2.94
2.88
3.11
3.18
3.75
4.36
3.88
4.46
e0.65
e1.26
1.92
3.94
3.91
2bb
2cb
2db
23ac
23bd
23ce
23de
23ee
23fe
23ge
23he
23ie
24af
24bg
26ah
26bi
27j
28ak
28bk
a Conditions: 5 vol % CH3CN-H2O, µ ) 0.5, T ) 20 °C, unless
otherwise specified. Multiple measurements by different methods were
averaged. b Source: this work. c Source: refs 16b, 19, 20a, and 24. d Source:
refs 16b, 19, 22a, and 24. e Source: refs 23 and 24. Conditions: 20 vol
% CN3CN-H2O, µ ) 0. f Source: refs 11a, 16b, 20a, and 24. g Source:
refs 22a and 24. h Source: ref 11a and unpublished results. i Source: ref
22c. Condition: H2O. j Source: ref 3. k Source: ref 4.
Figure 6. Log (kd-G/ks) vs log(kaz/ks) for heterocyclic (b) and carbocyclic
(2) nitrenium ions that exhibit reactivity with C-8 of d-G. The correlation
line is log(kd-G/ks) ) log(kaz/ks) - 0.67. Ions that fall off the line are
identified.
Scheme 7
range of other nitrenium ions under these conditions.20,22 The
kaz/ks ratios for 2a and 2b are at the upper limit of those observed
for carbocyclic nitrenium ions that have kaz at or near the
diffusion limit,23 so it is possible that kaz for 2a and 2b is
significantly below 5 × 109 M-1 s-1. Nevertheless, the estimated
lifetimes, 1/ks, for 2a and 2b of 1.0 and 0.5 ms, respectively,
show that these are quite stabilized cations capable of high
selectivity in their reactions with nucleophiles because of their
long aqueous solution lifetimes. The estimated lifetimes of 2c
and 2d of 24 and 10 µs, respectively, are in the same range as
those for 24a and 24b, both of which react with N3- at or near
the diffusion-controlled limit.20a,22a These are also quite stabi-
lized nitrenium ions, although less stable than 2a,b, and the
highly stabilized 23i.23,24
Table 5 provides a comparison of log(kaz/ks) and log-
(kd-G/ks) for 2a-d with a series of carbocyclic and heterocyclic
ions 23-24 and 26-28 identified in Scheme 7. The data were
obtained either by competition studies similar to those described
here or by direct measurements of rate constants on nitrenium
ions generated by laser flash photolysis.3,4,11,16,19,20,22-24 A plot
of log(kd-G/ks) vs log(kaz/ks) (Figure 6) shows that 2a-d and
several other ions exhibit 1 to 2 orders of magnitude lower
selectivity toward d-G than predicted by the correlation line
for the majority of the ions. The data points for those ions fall
on a line of unit slope that would be expected if both kaz and
kd-G are diffusion limited. In fact, for the ions that fall on the
line, measured values of kaz do cluster around 5 × 109 M-1 s-1
considered to be the diffusion-controlled limits for these
reactions. Measured values of kaz for two of the ions that fall
below the line (23i and 26b) are within a factor of 2 of the
apparent diffusion limits, so the points that fall significantly
below the line apparently do so because kd-G is 1 to 2 orders
of magnitude below the diffusion limit for those ions. The onset
of activation-limited reactions with d-G is not related to the
magnitude of ks for these ions, since the ions that fall
significantly below the line have ks ranging from ca. 103 to 107
s-1. McClelland has suggested that the activation-limited
reaction of certain nitrenium ions with d-G is related to the
ability of substituents in the ions to localize the positive charge
at µ ) 0.5 and 1 × 1010 M-1 s-1 at µ ) 0.0, while kd-G is ca.
-1
(1-2) × 109 M-1
s
at µ ) 0.0 or 0.5.16,20,22-24 These are
(22) (a) McClelland, R. A.; Davidse, P. A.; Hadzialic, G. J. Am. Chem. Soc.
1995, 117, 4173-4174. (b) Bose, R.; Ahmad, A. R.; Dicks, A. P.; Novak,
M.; Kayser, K. J.; McClelland, R. A. J. Chem. Soc., Perkin Trans. 2 1999,
1591-1599. (c) Ramlall, P.; McClelland, R. A. J. Chem. Soc., Perkin Trans.
2 1999, 225-232.
(23) Ren, D.; McClelland, R. A. Can J. Chem. 1998, 76, 78-84.
(24) McClelland, R. A.; Gadosy, T. A.; Ren, D. Can. J. Chem. 1998, 76, 1327-
1337.
9
J. AM. CHEM. SOC. VOL. 124, NO. 27, 2002 7979

A R T I C L E S
Novak et al.
Scheme 8
in the Supporting Information, by reacting the hydroxylamines with 2
equiv of AcCl in THF. The hydroxamic acid 5a was obtained by basic
hydrolysis of 1a in THF/H2O as described in the Supporting Informa-
tion.
Kinetic and Product Studies. Purification of solvents and general
procedures for performing kinetic studies, product studies by HPLC
methods, and data handling have been described.19,28 All reactions were
performed in 5 vol % CH3CN-H2O at an ionic strength of 0.5
maintained with NaClO4 except for those reactions that occurred in
1.64 M HClO4. Buffers used to maintain pH were Na2HPO4/NaH2-
PO4, NaOAc/HOAc, or HCOONa/HCOOH. At pH e2.5 HClO4
on a heteroatom other than the nitrenium N.22c It was argued
that this would reduce the electrophilic reactivity at the nitrenium
N necessary for formation of the C-8 adduct. All seven of the
ions that fall below the correlation line do have a dominant
resonance structure in which the charge is localized onto a
heteroatom (O or N) with all heavy atoms maintaining an octet
of electrons, but so do two of the ions that fall on the line
(28a,b).
If kaz is diffusion limited for 2a-d, kd-G for all four cations
is in the range of (4-9) × 107 M-1 s-1. The relatively slow
trapping of these ions by d-G may play a role in the detection
of the N-2 adducts, 18c,d. If the reaction of 2c with d-G to
form the C-8 adduct did proceed at the diffusion-controlled limit,
while the rate constant for formation of the N-2 adduct remained
unchanged, the relative yield of 18c would be ca. 1% rather
than the observed 22%.
solutions were used. Initial concentrations of 1a-d of ca. 5 × 10-5
M
for kinetics and product studies were obtained by 15 µL injection of a
0.01 M stock solution in DMF into 3.0 mL of the solution that had
been incubated at 20 °C for 15 min. Repetitive wavelength scans were
collected between 200 and 400 nm for at least 5 half-lives of the
reaction. Absorbance vs time data for 1a were collected at 256 (pH
0.8-4.4) and 263 nm (pH 4.8-7.4). For 1b data were collected at 252
(pH 0.9-3.7) and 262 nm (pH 3.9-7.4). Kinetic data for 1c and 1d
were collected at 266 nm at all pH values. Data were fit to the first-
order rate equation to obtain kobs.
28 Rate constants for the disappearance
of 1a were also determined in acidic solutions by the initial rates
method, using HPLC with UV detection at 256 nm. Spectrophotometric
titrations were carried out at 257 nm for 1a, 255 nm for 1b, and 266
-
nm for 1c. Kinetics of decomposition of 1a-d in the presence of N3
and d-G were performed in 0.02 M 2/1 Na2HPO4/NaH2PO4 buffers at
pH 6.90. These reactions were followed at the following wavelengths:
1a, 323 (N3-) and 314 nm (d-G); 1b, 272 (N3-) and 316 nm (d-G);
and 1c and 1d, 266 (N3-) and 335 nm (d-G).
The results obtained for the heterocyclic nitrenium ions
2a-d, 27, 28a, and 28b show that these ions retain many of
the characteristic reactions of carbocyclic nitrenium ions:
Reaction products of 1a and 1b in phosphate and acetate buffers
and in HClO4 solutions were monitored by HPLC on a C8 reverse-
phase analytical column with UV detection at 320 nm, using 50/50
MeOH/H2O eluent buffered with 0.05 M 1:1 NaOAc/HOAc at a flow
rate of 1 mL/min. Product yields were determined by triplicate 20 µL
injections after 10 half-lives of the reaction. Identical conditions were
used for 1c and 1d except that products were monitored at 300 nm
and the eluting buffer was 40/60 MeOH/H2O. For the determination
of kphosT/ks selectivity ratios, 0.08 M phosphate buffers were made with
varying buffer ratios. These buffers were diluted to 0.04, 0.02, 0.01,
and 0.005 M while maintaining the ionic strength. After 10 half-lives
of the reaction, product yields were determined by HPLC methods.
Details of the isolation and characterization of the buffer and solvent-
derived products 4c,d, 7a,b, 8a,b, 9a,b, 10a,b, 11c,d, and 12c,d are
described in the Supporting Information.
Product Studies in the Presence of N3- and d-G. All buffers were
Na2HPO4/NaH2PO4 (B/A ) 2/1; pH 6.9), µ ) 0.5, 5 vol % CH3CN/
H2O. All reactions were carried out at 20 °C. The buffer solutions had
been incubated at 20 °C for 15 min prior to the addition of 1a-d. The
product yields were determined by HPLC (UV detection at 256, 300,
and 320 nm, using 50/50 or 40/60 MeOH/H2O eluent buffered as
described above).
-
selective reaction with N3 to generate substitution products,
reaction with d-G to form C-8 adducts, and efficient reduction
by I-. The more stable ions 2a-d exhibit unusual reactions
with HPO4 or AcO- to generate addition products in which
2-
the aromatic character of one or more rings is lost.
In carbocyclic ions typically only H2O reacts to form stable
addition products in which aromaticity is lost. This occurs
presumably because of the poor leaving group ability of OH-
in products such as 29.20b Attack of a second equivalent of H2O
(Scheme 8) is kinetically disfavored at moderate pH (except
for 25 and related stilbenyl ions) because such reactions appear
to require specific acid catalysis and do not lead to products
stabilized by recovery of aromatic character.20b Apparently the
aromatic character of the heterocyclic rings of 2a-d is
considerably weaker allowing better leaving groups such as
-
HPO4 and AcO- to form addition products (Scheme 4) that
do not revert to the nitrenium ion. Attack of H2O on the inital
addition product to form products such as 7, 8, 9, 10, and 12 is
more favorable in these cases because there is less driving force
for rearomatization.
The product yields for 1a in N3- solutions were determined in 0.16
M total buffer with N3- concentrations ranging from 3 × 10-5 to 5 ×
10-4 M. The initial concentration of 1a was ca. 5 × 10-6 M, which
was obtained by 15 µL injections of a 0.001 M stock solution in DMF
into 3.0 mL of the buffer solution. The product yields were determined
by HPLC as described above.
Experimental Section
Synthesis. The syntheses of Glu-P-1, Glu-P-2, MeIQx, IQx, and
the nitro derivatives NO2Glu-P-1, NO2Glu-P-2, NO2MeIQx, and
NO2IQx have been described in the literature.25,26 The hydroxylamines
NOHGlu-P-1, NOHGlu-P-2, NOHMeIQx, and NOHIQx were syn-
thesized by the method of Kazerani and Novak by N2H4 reduction of
the nitro compounds.27 The esters 1a-d were prepared, as described
-
The N3 selectivity ratio for 1b was determined by two methods:
(1) varying phosphate concentrations while maintaining constant azide
concentration and (2) varying azide concentrations while maintaining
constant phosphate concentration. (1) Buffers were prepared with 0.02-
0.16 M total phosphate, and constant 5 × 10-6 M N3-. The initial
(25) Takeda, K.; Shudo, K.; Okamoto, T.; Kosuge, T. Chem. Pharm. Bull. 1978,
26, 2924-2925. Hashimoto, Y.; Shudo, K.; Okamoto, T. J. Am. Chem.
Soc. 1982, 104, 7636-7640. Bristow, N. W.; Charlton, P. T.; Peak, D. A.;
Short, W. F. J. Chem. Soc. 1954, 616-629. Astik, R. R.; Thaker, K. A. J.
Ind. Chem. Soc. 1981, 58, 1013-1014. Saint-Ruf, G.; Loukakou, B.;
N’Zouzi, C. J. Heterocycl. Chem. 1981, 18, 1565-1570.
(27) Kazerani, S.; Novak, M. J. Org. Chem. 1998, 63, 895-897.
(28) Novak, M.; Pelecanou, M.; Roy, A. K.; Andronico, F. M.; Plourde, F. M.;
Olefirowicz, T. M.; Curtin, T. J. J. Am. Chem. Soc. 1984, 106, 5623-
5631.
(26) Grivas, S.; Olsson, K. Acta Chem. Scand. 1985, B39, 31-34. Grivas, S. J.
Chem. Res. 1988, 84.
9
7980 J. AM. CHEM. SOC. VOL. 124, NO. 27, 2002

Reactivity and Selectivity of Nitrenium Ions
A R T I C L E S
concentration of 1b was ca. 1.0 × 10-6 M. (2) Buffers with constant
0.02 M buffer and 5 × 10-6-2 × 10-3 M NaN3 were prepared. The
initial concentration of 1b was ca. 1 × 10-6 M. This concentration
was obtained by 15 µL injection of a 2 × 10-4 M stock solution in
DMF into 3.0 mL of the buffer solution. The product yields were
determined by HPLC.
trans-9b with a 9-phosphoryl group and minimizing the energy
of the molecule. Heats of formation vs dihedral angle calcula-
tions were generated by energy minimizations of cis-9b and
trans-9b with fixed N10-C9-O9-C9OAc dihedrals initially at
10° intervals.
Procedures for 1c and 1d were similar to those followed for 1a except
that initial concentrations of the esters were kept at ca. 1 × 10-5 M.
Acknowledgment. This work was supported by a grant from
the American Cancer Society (RPG-96-078-03-CNE). NMR
spectra were obtained on equipment made available through an
NSF grant (CHE-9012532) and upgraded through an Ohio Board
of Regents Investment Fund grant. MALDI-TOF and LC-MS
mass spectra were obtained on equipment funded by the NSF
(CHE-9413529) and the Ohio Board of Regents Investment
Fund, respectively. High-resolution mass spectra were obtained
at the Ohio State University Chemical Instrumentation Center.
The concentration range of N3 in these studies was 5 × 10-5-1 ×
-
10-3 M. Decomposition reactions of 1a-d in d-G containing solutions
were carried out in phosphate buffers with 0.02 M total buffer
concentration. Reaction products were monitored by HPLC with UV
detection at 300 or 320 nm as described above.
-
Details of the large-scale isolation and characterization of the N3
and d-G adducts 16a-d, 18c,d, 19a,b, 20b, 21c,d, and 22a are
described in the Supporting Information.
Supporting Information Available: Synthesis of 1a-d and
5a, table of rate constants obtained for the decomposition of
1a-d, isolation and characterization of the buffer and solvent
derived products 4c,d, 7a,b, 8a,b, 9a,b, 10a,b, 11c,d, and 12c,d,
and the N3- and d-G adducts 16a-d, 18c,d, 19a,b, 20b, 21c,d,
Calculations
The AM1 semiempirical calculations were carried out on a
Dell Optiplex GX200 desktop computer with an Intel Pentium
III processor and 256 MB RAM, using CambridgeSoft MOPAC
Pro 5.0 software. The lowest energy conformation of 9b in both
the cis- and trans-configurations was found by energy minimiza-
tion starting from several different conformations of the 2-ac-
etamido and the 9-O-acetyl groups. The lowest energy confor-
mation of cis- and trans-8b was found by substituting the 9-O-
acetyl groups in the lowest energy conformations of cis-9b and
-
and 22a, kinetic analysis of the N3 dependence of the yields
of 19b and 20b, and repetitive wavelength scans for 1a-d
(PDF). This material is available free of charge via the Internet
at http://pubs.acs.org.
JA0121944
9
J. AM. CHEM. SOC. VOL. 124, NO. 27, 2002 7981