Direct EPR Spectroscopic Evidence for an Allylic Radicla Generated from (E)-2'-Fluoromethylene-2'-deoxycytidine 5'-Diphosphate by E. coli Ribonucleotide Reductase

Full text of DOI:10.1021/ja9740273

4252
J. Am. Chem. Soc. 1998, 120, 4252-4253
Direct EPR Spectroscopic Evidence for an Allylic
Radical Generated from
(E)-2′-Fluoromethylene-2′-deoxycytidine
5′-Diphosphate by E. coli Ribonucleotide Reductase
Wilfred A. van der Donk,† Gary J. Gerfen,¥ and
JoAnne Stubbe*
Departments of Chemistry and Biology
Massachusetts Institute of Technology
Cambridge, Massachusetts 02139
Figure 1. (A) 9.39 GHz EPR spectrum of a sample containing RNR
(50 µM), [6′-13C]-(E)-FMCDP (0.5 mM), and ATP (1.6 mM) frozen 30
s after addition of the inhibitor (solid line) overlayed with the spectrum
of Y122• (- - -). The intensity of the Y• signal was normalized to reflect
the radical concentration at t ) 30 s as quantitated by the absorbance at
410 nm.2c (B) New radical signal at t ) 30 s after subtraction of the
Y122• signal. Instrument settings: power 10 µW, modulation amplitude
0.39 mT, T ) 109 K. The overlaid (‚‚‚) spectrum derives from [6′-12C]-
(E)-FMCDP, showing that the peaks marked with an asterisk most likely
arise from residual unlabeled material. (C) Family of 10 best-fit
simulations of the [6′-13C] spectrum using the parameter set given in Table
1.
ReceiVed NoVember 26, 1997
(E)-2′-Fluoromethylene-2′-deoxycytidine, (E)-FMC, has potent
antiproliferative activity against a wide range of tumor cell lines.1
Its cytotoxicity is thought to arise from a synergistic effect
between its triphosphate, a DNA chain terminator,1 and its
diphosphate (FMCDP), a stoichiometric inactivator of ribonucle-
otide reductase (RNR).2 RNRs catalyze the conversion of
nucleotides to deoxynucleotides, the rate-determining step in DNA
biosynthesis. The class I RNRs are composed of two subunits:
R1 and R2. Initiation of the reduction process is thought to
proceed via a thiyl radical on R1 abstracting the 3′-hydrogen atom
from the nucleotide. While much indirect evidence supports this
proposal, direct evidence for a 3′-nucleotide radical intermediate
has been elusive. Our recent studies indicate that FMCDP is a
mechanism-based inhibitor of RNR. The inactivation is ac-
companied by loss of fluoride ion, stoichiometric alkylation of
the R1 subunit, loss of the essential tyrosyl radical (Y•) on the
R2 subunit, and formation of a new nucleotide-based radical.2b,c
To establish the structure of this radical, [6′-13C]-(E)-FMCDP was
synthesized and incubated with RNR. EPR studies of the resulting
radical establish that it is allylic, requiring that RNR catalyzes
3′-hydrogen atom abstraction.
On the basis of biochemical and EPR data, two mechanisms
for nucleotide radical generation from FMCDP have been
proposed.2c The initial step in both is 3′-hydrogen atom abstrac-
tion to generate a fluorinated allyl radical (Scheme 1). This
radical can be reduced by hydrogen atom transfer from one of
the three cysteine residues in the active site. Reduction from the
R-face or the â-face of the nucleotide (pathways A and B,
respectively), followed by loss of fluoride ion, results in the
generation of 1 and a cysteinyl radical. In pathway A, Glu441
adds to the exocyclic methylene of 1 to generate 2 which is
converted to allyl radical 3 via hydrogen atom abstraction by a
thiyl radical on the R-face of the nucleotide. In pathway B, the
thiyl radical of Cys439 adds directly to 1, resulting in formation
Table 1. Hyperfine Values for the 13C-Labeled Radical As
Determined by the Simulations in Figure 1Ca
HR
Hâ
13C
A1
A2
A3
Aiso
2.1(0.2)
0.7(0.3)
1.4(0.3)
1.4(0.3)
1.3(0.3)
1.4(0.3)
1.4(0.3)
1.4(0.3)
1.1(0.4)
0.2(0.3)
5.5(0.2)
2.3(0.2)
a The numbers in parentheses are estimated uncertainties. g1
2.0030, g2 ) 2.0042, and g3 ) 2.0018. For principal axis orientations,
see ref 9.
)
of the ketyl radical 4. A distinction between these two mecha-
nistic possibilities could not be made unambiguously; however,
the allyl radical pathway A was favored on the basis of simulations
of 9 and 140 GHz EPR data from [6′-2H]- and [6′-1H]-(E)-
FMCDP.2c
To establish the structure of the radial, [6′-13C]-(E)-FMCDP
was synthesized on the basis of the route developed by McCarthy
and co-workers3 starting with the [13C]-methylation of thiophenol
(Scheme 2).2b,4-7 [6′-13C]-(E)-FMCDP was obtained, and the
NMR analysis indicated 96-97% 13C-incorporation into 6′-C
(Supporting Information).
[6′-13C]-(E)-FMCDP was incubated with E. coli RNR, and after
30 s the sample was frozen in liquid N2, and the EPR spectrum
was recorded at 9.39 GHz. The resulting spectrum is shown in
Figure 1A overlayed with the spectrum of Y•. Subtraction of
0.82 equiv of Y• signal gave rise to the signal shown in Figure
1B. Spin quantitation revealed 0.14 equiv of new radical/equiv
of RNR.2c Since the 13C-labeled nucleotide contains 3-4% of
12C-labeled material, a minor component of the signal is derived
from unlabeled radical (Figure 1B).
* Corresponding author. Phone: (617) 253-1814. Fax: (617) 258-7247.
E-mail: stubbe@mit.edu.
† Present address: Department of Chemistry, University of Illinois at
Urbana-Champaign, Urbana, IL 61801.
‡ Present address: Department of Physiology and Biophysics, Albert
Einstein College of Medicine of Yeshiva University, 1300 Morris Park Ave.,
Bronx, NY 10461.
(1) McCarthy, J. R.; Sunkara, P. S. In Design, Synthesis and Antitumor
ActiVity of an Inhibitor of Ribonucleotide Reductase; McCarthy, J. R., Sunkara,
P. S., Eds.; CRC Press: Boca Raton, FL, 1995; pp 3-32. (b) McCarthy, J.;
Sunkara, P. S.; Matthews, D. P.; Bitonti, A. J.; Jarvi, E. T.; Sabol, J. S.;
Resvick, R. J.; Huber, E. W.; van der Donk, W. A.; Yu, G.; Stubbe, J. ACS
Symp. Ser. 1996, 639, 246-264 (c) Bitonti, A. J.; Dumont, J. A.; Bush, T.
L.; Cashman, E. A.; Cross-Doersen, D. E.; Wright, P. S.; Matthews, D. P.;
McCarthy, J. R.; Kaplan, D. A. Cancer Res. 1994, 54, 1485-1490. (d) Bitonti,
A. J.; Bush, T. I.; Lewis, M. T.; Sunkara, P. S. Anticancer Res. 1995, 15,
1179-1182.
(2) (a) Sunkara, P. S.; Lippert, B. J.; Snyder, R. D.; Jarvi, E. T.; Farr, R.
A. Proc. Am. Assoc. Cancer Res. 1988, 29, 324. (b) van der Donk, W. A.;
Yu, G.; Silva, D. J.; Stubbe, J.; McCarthy, J. R.; Jarvi, E. T.; Matthews, D.
P.; Resvick, R. J.; Wagner, E. Biochemistry 1996, 35, 8381-8391. (c) Gerfen,
G. J.; van der Donk, W. A.; Yu, G.; McCarthy, J. R.; Matthews, D. P.; Jarvi,
E. T.; Farrar, C.; Griffin, R. G.; Stubbe, J. J. Am. Chem. Soc., in press.
To establish the structure of this radical, simulations were
generated using a simulated annealing protocol8 (see the Sup-
porting Information). Proton hyperfine parameters and g-values
(3) McCarthy, J. R.; Matthews, D. P.; Sabol, J. S.; McConnell, J. R.;
Donaldson, R. E.; Duquid, R. US Patent 5,589,587, 1996.
(4) Ono, N.; Miyake, H.; Saito, T.; Kaji, A. Synthesis 1980, 952-953.
(5) Johnson, C. R.; Keiser, J. E. Org. Synth. 1967, 46, 791-793.
(6) (a) McCarthy, J. R.; Matthews, D. P.; Paolini, J. P. Org. Synth. 1994,
72, 209-215. (b)Robins, M. J.; Wnuk, S. F. J. Org. Chem. 1993, 58, 3800-
3801.
(7) (a) Yoshikawa, M.; Kato, T.; Takenishi, T. Tetrahedron Lett. 1967,
50, 5065-5068. (b) Hoard, D. E.; Ott, D. G. J. Am. Chem. Soc. 1965, 87,
1785-1788.
S0002-7863(97)04027-4 CCC: $15.00 © 1998 American Chemical Society
Published on Web 04/22/1998

Communications to the Editor
J. Am. Chem. Soc., Vol. 120, No. 17, 1998 4253
Scheme 1
Scheme 2a
diagnostic for allyl radicals.12 The isotropic hyperfine coupling
value expected for a ketyl radical, 4, has been reported as ∼1.0
mT,13 which is well outside the uncertainty of our measured value.
In addition giso is 2.003, also outside the range for reported ketyl
radicals. Thus, the 13C hyperfines and the excellent simulations
unambiguously establish that the radical signal derived from
FMCDP is allylic in nature and lacks a fluorine. This structure
can only be generated by removal of the hydrogen atom from
C3′ of FMCDP. Thus, direct evidence for a nucleotide radical
and the importance to 3′ chemistry in its formation is provided.
Acknowledgment. This work was supported by a grant from the
National Institutes of Health to J.S. (GM-29595) and by a fellowship
from the Jane Coffin Childs Fund to W.A.v.d.D. (Project 61-960). We
thank Drs. J. R. McCarthy and R. E. Richardson for advice on the
preparation of [6′-13C]-(E)-FMCDP.
a Conditions: (a) 13CH3I, DBU; (b) NaIO4; (c) DAST, SbCl3; (d)
Oxone; (e) CIP(O)(OEt)2, 2 equiv of LiHMDS; (f) 6, KOtBu; (g) Bu3SnH,
AIBN; (h) CsF, MeOH; (i) POCl3; (j) H2O; (k) CDI, Bu3NH‚H2PO4.
Supporting Information Available: 1H and 31P NMR spectra for
(E)-FMCMP, a table of changes in the previously reported NMR data
for compounds shown in Scheme 2 that are due to the introduction of a
13C-label, and full details of the simulated annealing protocol (3 pages,
print/PDF). See any current masthead page for ordering information and
Web access instructions.
were constrained to those determined in our previous X- and
D-band EPR studies,2c while 13C principal values were allowed
to vary between 0 and 2.5 mT (A1 and A2) and 4.0-6.5 mT (A3).9
The 10 best fit simulations are shown in Figure 1C.9
JA9740273
The measured isotropic 13C hyperfine coupling of 2.3 mT
(Table 1) yields a spin density on C6′ of 0.64,10,11 which is very
(10) The estimate of spin density was made using the equation11
3
3
(8) (a) Press, W. H.; Flannery, B. P.; Teukolsky, S. A.; Vetterling, W. T.
Numerical Recipies; Cambridge University Press: Cambridge, 1986; pp 326-
334. (b) Hustedt, E. J.; Smirnov, A. I.; Laub, C. F.; Cobb, C. E.; Beth, A. H.
Biophys. J. 1997, 74, 1861-1877.
Aiso ) (QCs
+
QCCXj)Fi +
QCXjCFj
∑
∑
j)1
j)1
in which Fi is the spin density on the labeled carbon, Fj indexes spin density
(9) The X-band simulations were insensitive to tensorial orientations:
annealing runs which allowed these orientations to vary did not produce better
fits (fit quality determined as described in the Supporting Information). Thus,
with the exception of A1 and A2 of HR, all hyperfine principal axis orientations
were taken as collinear with the correspondingly indexed g-principal axis
orientations. A1 and A2 of HR were rotated by 30° from the corresponding
g-principal axes to achieve better fits of 139.5 GHz EPR spectra reported
previously.2c Each 13C hyperfine value and associated uncertainty listed in
Table 1 are the mean and 2 standard devations obtained from 60 annealing
runs, with 2500 simulations calculated per run. The larger uncertainties in
the 13C A1 and A2 values reflect the broader range of convergence for these
values in the simulated annealing runs. This broader range is expected for
components whose splittings are not resolved in the EPR spectrum.
on adjacent atoms Xj, and the Q values are calculated spin polarization
constants (QsC ) -1.27 mT, QCCC′ ) 1.44 mT, QCCH ) 1.95 mT, QCCÃ ) 1.77
mT, QCC′C ) -1.39 mT)11,14 In addition, the unknown atom X in 3 has been
assumed to have a Q value equal to that of oxygen, the spin density on
hydrogen and on the unknown atom has been considered to be negligible,
and the spin density on the 2′-carbon has been assumed to be -0.19, consistent
with a typical allyl radical.12
(11) Karplus, M.; Fraenkel, G. K. J. Chem. Phys. 1962, 35, 1312-1323.
(12) Heller, C.; Cole, T. J. Chem. Phys. 1962, 37, 243-250.
(13) Camaioni, D. M.; Walter, H. F.; Jordan, J. E.; Pratt, D. W. J. Am.
Chem. Soc. 1973, 95, 7978-7992.
(14) Das, M.; Fraenkel, G. J. Chem. Phys. 1965, 42, 1350.