Tuning of the thermochemical and kinetic properties of ascorbate by its local environment: Solution chemistry and biochemical implications

Article abstract of DOI:10.1021/ja102337n

Ascorbate (vitamin C) is a ubiquitous biological cofactor. While its aqueous solution chemistry has long been studied, many in vivo reactions of ascorbate occur in enzyme active sites or at membrane interfaces, which have varying local environments. This report shows that the rate and driving force of oxidations of two ascorbate derivatives by the TEMPO radical (2,2′,6,6′-tetramethylpiperidin-1-oxyl) in acetonitrile are very sensitive to the presence of various additives. These reactions proceed by the transfer of a proton and an electron (a hydrogen atom), as is typical of biological ascorbate reactions. The measured rate and equilibrium constants vary substantially with added water or other polar solutes in acetonitrile solutions, indicating large shifts in the reducing power of ascorbate. The correlation of rate and equilibrium constants indicates that this effect has a thermochemical origin rather than being a purely kinetic effect. This contrasts with previous examples of solvent effects on hydrogen atom transfer reactions. Potential biological implications of this apparently unique effect are discussed.

Full text of DOI:10.1021/ja102337n

Published on Web 05/17/2010
Tuning of the Thermochemical and Kinetic Properties of
Ascorbate by Its Local Environment: Solution Chemistry and
Biochemical Implications
Jeffrey J. Warren* and James M. Mayer*
Department of Chemistry, UniVersity of Washington,
Box 351700, Seattle, Washington 98195-1700
Received March 19, 2010; E-mail: jjw82@u.washington.edu; mayer@chem.washington.edu
Abstract: Ascorbate (vitamin C) is a ubiquitous biological cofactor. While its aqueous solution chemistry
has long been studied, many in vivo reactions of ascorbate occur in enzyme active sites or at membrane
interfaces, which have varying local environments. This report shows that the rate and driving force of
oxidations of two ascorbate derivatives by the TEMPO radical (2,2′,6,6′-tetramethylpiperidin-1-oxyl) in
acetonitrile are very sensitive to the presence of various additives. These reactions proceed by the transfer
of a proton and an electron (a hydrogen atom), as is typical of biological ascorbate reactions. The measured
rate and equilibrium constants vary substantially with added water or other polar solutes in acetonitrile
solutions, indicating large shifts in the reducing power of ascorbate. The correlation of rate and equilibrium
constants indicates that this effect has a thermochemical origin rather than being a purely kinetic effect.
This contrasts with previous examples of solvent effects on hydrogen atom transfer reactions. Potential
biological implications of this apparently unique effect are discussed.
Introduction
Scheme 1. Redox Chemistry of Ascorbate (with the Carbon
-
Numbering Shown for AscH )
Ascorbic acid (vitamin C) is a key biological cofactor,
necessary for human health, that has a wide range of biochemi-
1
cal roles. It is important as an antioxidant, for instance reducing
2
R-tocopheroxyl radical or oxidized glutathione, and as a
3
cofactor for metalloenzymes such as ascorbate oxidase and
4
cytochrome b561. At physiological pH, it exists predominantly
-
as the ascorbate monoanion, AscH (Scheme 1). Oxidation of
•-
ascorbate forms the ascorbyl radical Asc , which has one proton
and one electron less than AscH , and further oxidation gives
7
,8
-
systems. In such cases, the thermochemistry is in some ways
dehydroascorbate, which can hydrate to the corresponding
hemiketal in aqueous solution.
Ascorbate has traditionally been considered a one-electron
reductant, but it almost always reacts with loss of both an
better described by the O-H bond dissociation free energy
5
-1
(BDFE), 73.6 ( 1.0 kcal mol , for aqueous ascorbate (see
9
below), rather than the redox potential. The standard 0.282 mV
potential for ascorbate at pH 7 is mechanistically less informa-
-
+
electron and a proton (Scheme 1) because the neutral ascorbyl
tive because it is for the more complex 2e /1H transfer to
6
•
radical is a high-energy, highly acidic species [pK
a
(AscH ) )
produce dehydroascorbate.
6
-
-
0.45]. In many cases, ascorbate reacts to transfer the e and
H in a single kinetic step, in effect as a hydrogen atom. Njus
We recently described oxidations of ascorbate or 5,6-
+
-
10
isopropylidene ascorbate (iAscH , Scheme 1) by oxyl radicals
1
1
and co-workers have shown that ascorbate reacts as a hydrogen
atom donor to oxyl radicals, cytochrome b561, and other
or porphyrin-iron-imidazole complexes and showed that
+
-
these reactions proceed via a concerted transfer of H + e
•
(
H ) rather than by a stepwise mechanism with initial proton or
1
0
•-
•-
(
(
1) Vitamin C in Health and Disease; Packer, L., Fuchs, J., Eds.; Marcel
Dekker: New York, 1997; Vol. 5.
2) Reed, D. J. Interaction of vitamin E, ascorbic acid and glutathione in
protection against oxidative damage. In Vitamin E in Health and
Disease; Packer, L.; Fuchs, J., Eds.; Marcel Dekker: New York, 1993;
pp 269-281.
electron transfer. The ascorbyl radical (Asc or iAsc ) is
remarkably persistent in “dry” MeCN solvent, decaying over
1
0
hours at ambient temperatures. This contrasts with the rapid
disproportionation of ascorbyl radicals in water (milliseconds
(
3) Kroneck, P. M. H.; Armstrong, F. A.; Merkle, H.; Marchesini, A.
Ascorbate oxidase: molecular properties and catalytic activity. In
Ascorbic Acid: Chemistry, Metabolism and Uses; Seib, P. A., Tolbert,
B. M., Eds.; Advances in Chemistry Series 200; American Chemical
Society: Washington, D.C., 1982; pp 223-248.
(7) Njus, D.; Kelley, P. M. Biochim. Biophys. Acta 1993, 1144, 235–
248.
(8) Njus, D.; Kelley, P. M. FEBS Lett. 1991, 283, 147–151.
(9) The BDFE is a better predictor of reactivity than the more commonly
used bond dissociation enthalpy. See: Mader, E. A.; Davidson, E. R.;
Mayer, J. M. J. Am. Chem. Soc. 2007, 129, 5153–5166.
(10) Warren, J. J.; Mayer, J. M. J. Am. Chem. Soc. 2008, 130, 7546–7547.
(11) Warren, J. J.; Mayer, J. M. J. Am. Chem. Soc. 2008, 130, 2774–2776.
(
4) Njus, D.; Wigle, M.; Kelley, P. M.; Kipp, B. H.; Schlegel, H. B.
Biochemistry 2001, 40, 11905–11911.
(
(
5) Kerber, R. C. J. Chem. Educ. 2008, 85, 1237–1242.
6) Creutz, C. Inorg. Chem. 1981, 20, 4449–4452.
7
784 9 J. AM. CHEM. SOC. 2010, 132, 7784–7793
10.1021/ja102337n  2010 American Chemical Society

Tuning Thermochemical and Kinetic Properties of Ascorbate
A R T I C L E S
at 10- M concentrations). The surprisingly long lifetime has
allowed us to directly measure equilibrium constants, as
described below, and thereby determine the O-H BDFE of
4
12
-
-1
AscH in MeCN as 67.7 ( 0.8 kcal mol . Remarkably, this is
1
5
.9 ( 1.3 kcal mol- weaker than the aqueous value.
Such a large change in X-H homolytic bond strength with
solvent is highly unusual. The studies described here show that
this change is not due the bulk properties of the medium such
as dielectric (ε). Instead, the dramatic shift in ascorbate
homolytic bond strength is predominantly due to the local
environment about the ascorbate, especially the presence of
hydrogen bond donors. Ascorbate reactions often occur within
enzyme active sites or at membrane interfaces that are different
from bulk water, so it seems likely that Nature has taken
advantage of this unusual property of ascorbate; the last section
of this article explores three possible examples.
Figure 1. Kinetic data (O) for reaction of 0.28 mM AscH- with 8.7 mM
TEMPO in DMSO and a first-order global fit (s) from SPECFIT. Inset:
Stopped-flow UV/vis spectra over 1.1 s for this reaction; the broad
absorption centered at 460 nm is due to TEMPO.
Results
slopes give bimolecular rate constants k1,MeCN ) 500 ( 50 M-
1
1
. Ascorbate + TEMPO Rate and Equilibrium Constants
-1
-1 -1
in Pure Solvents. To examine the intrinsic reactivity of ascorbate
derivatives, we have examined their oxidation by the stable
TEMPO radical (eq 1, TEMPO ) 2,2′,6,6′-tetramethylpiperidin-
-oxyl). These are representative and easily studied reactions
that occur rapidly by hydrogen atom transfer to form TEMPOH
and the corresponding ascorbyl radical (eq 1), as previously
s
and k1,DMSO ) 100 ( 10 M
s
(see Tables 1 and 2, below).
The rate constants (and equilibrium constants, see below) are
unaffected by the presence or absence of 0.1 M Bu NPF .
4 6
Unless specifically noted, all rate and equilibrium constants were
determined at 298 K without added electrolyte. Rate constants
n
1
-
for oxidations of isopropylidene ascorbate (iAscH ) by TEMPO
have been measured similarly, under a wide range of conditions,
1
0
described.
1
0
extending the previous report (see Tables 1 and 3, below).
In MeCN and DMSO, with <10 equiv of TEMPO, reaction
proceeds to an equilibrium mixture of the four species. The
1
position of equilibrium can be measured by using the stopped-
flow instrument to do a spectrophotometric titration. Reactions
are complete after a few seconds, and the position of the
equilibrium can be determined before there is significant decay
of the ascorbyl radical product. Equilibrium constants were
The studies presented here used the acetonitrile-soluble ascorbate
•-
-
calculated by plotting [TEMPOH][Asc ]/[AscH ] vs [TEMPO].
These plots show good linearity, validating the mass balance
assumption and the equilibrium treatment (see Supporting
+
-
salt with potassium 18-crown-6 ([K18C6 ]AscH ) and the
DBU-H+ salt of the isopropylidene derivative iAscH (DBU
-
)
1,2-diazabicyclo[5.4.0]undec-7-ene), avoiding the low-melting
+
Information). The derived equilibrium constant for [K18C6 ]-
n
+
and more difficult to handle Bu
4
N
compounds used in the
-
AscH + TEMPO, K1,MeCN, is 0.14 ( 0.02 (∆G°1,MeCN ) 1.3
+
-
previous study. [K18C6 ]AscH was prepared from potassium
-1
(
0.1 kcal mol ), which is within error of the previously
13
+
-
ascorbate + 18-crown-6 in MeCN, and [DBU-H ]iAscH was
generated in situ from solutions of iAscH by adding 1 equiv
of DBU. Both of these salts are spectroscopically and kinetically
n
+
-
10
measured K for Bu N AscH , 0.12 ( 0.04. Equilibrium
4
2
constants for analogous reactions were measured similarly and
are given in Tables 2 and 3.
n
+
4
identical to their Bu N analogues.
Rate constants have also been measured in water (pH 7.1,
+
-
Rate constants for the oxidation of [K18C6 ]AscH by
TEMPO (eq 1) have been determined in MeCN, DMSO, and
water (and in MeCN with various additives, see below). The
reactions in MeCN and DMSO were performed under pseudo-
first-order conditions of excess TEMPO (15-180 equiv),
observing the growth of the ascorbyl radical optical band at
-
-
1
00 mM phosphate buffer) for both AscH and iAscH . In
water, the ascorbyl radical optical band is not observed because
•-
15
disproportionation of Asc is fast compared to the initial
-
reaction of AscH + TEMPO. Therefore, these reactions were
done with excess ascorbate (10-42 mM) and monitored using
the disappearance of the optical band for TEMPO (4.3 mM).
The data were fit using SPECTFIT to a model with an initial
3
(
77 nm by using UV/vis stopped-flow spectrophotometry
Figure 1). SPECFIT global analysis software was used to fit
the spectra over the entire observed spectral range (320-550
•-
bimolecular hydrogen atom transfer step followed by Asc
6
-1 -1
disproportionation (kdisp ) 1 × 10
M
s
in 0.045 M
14
nm). The data follow a first-order kinetic model over >4 half-
lives. A plot of the derived pseudo-first-order rate constants as
a function of [TEMPO] is linear (see Supporting Information),
15
-1 -1
phosphate buffer, pH 7 ), yielding k1,H O ) 2.4 ( 0.3 M
s
2
-1 -1
and k1i,H O ) 5.5 ( 0.5 M s . The former rate constant is in
excellent agreement with a recent determination at more acidic
pH for the reduction of TEMPO by ascorbate. The corre-
2
-
with slopes that are independent of starting [AscH ]. These
16
sponding equilibrium constants cannot be determined in water,
again due to the rapid ascorbyl radical disproportionation.
(
(
(
12) Bielski, B. H. J.; Allen, A. O. J. Am. Chem. Soc. 1970, 92, 3793–
794.
13) Tajmir-Riahi, H. A.; Boghai, D. M. J. Inorg. Biochem. 1992, 45, 73–
4.
14) Binstead, R. A.; Zuberb u¨ hler, A. D.; Jung, B. Specfit, Version 3.0.38
32-bit Windows); Spectrum Software Associates: Chapel Hill, NC,
006.
3
8
(15) Bielski, B. H. J.; Allen, A. O.; Schwarz, H. A. J. Am. Chem. Soc.
1981, 103, 3516–3518.
(
2
(16) Sajenko, I.; Pilepi c´ , V.; Brala, C. J.; Sajenko, I.; Uri sˇ i c´ , J. Phys. Chem.
A 2010, 114, 3423–3430.
J. AM. CHEM. SOC. 9 VOL. 132, NO. 22, 2010 7785

A R T I C L E S
Warren and Mayer
Table 1. Ascorbate Bond Dissociation Free Energies (BDFEs) and
Scheme 2. Kinetic Model Used To Fit Reaction 1 in the Presence
of Water or Other Additives
Rate and Equilibrium Constants for Their Reactions with TEMPO
a
(
Eq 1) in Pure Solvents
-1
-1
b
reagent
solvent
k (M
s
)
K
1
BDFE
AscH-
MeCN
DMSO
500 ( 50
100 ( 10
2.4 ( 0.3
1720 ( 150
250 ( 20
0.14 ( 0.02
0.9 ( 0.1
67.7 ( 0.8
67.6 ( 1.0
73.6 ( 1.0
66.5 ( 0.8
67.5 ( 1.0
-
AscH-
-3
c
AscH
H
2
O
∼1.2 × 10
iAscH-
MeCN
DMSO
1.2 ( 0.2
0.85 ( 0.10
-
iAscH-
Table 2. Rate and Equilibrium Constants for Ascorbate (AscH ) +
a
TEMPO (Eq 1)
a
At 298 K. BDFE values in kcal mol-1. Determined for equilibrium
b
solvent
ꢀ
water
k
1
K
1
c
measurements with TEMPO (eq 1) except as noted. Calculated from
BDFEwater(AscH ) and BDFEwater(TEMPOH) (see below).
-
5
3
b
-
MeCN
3 × 10
2.2 × 10
3.6 × 10
7.2 × 10
7.2 × 10
1.2 × 10
1
500 ( 50
410 ( 50
370 ( 40
280 ( 70
300 ( 60
260 ( 50
2.4 ( 0.3
100 ( 10
0.14 ( 0.02
0.10 ( 0.01
0.080 ( 0.008
0.048 ( 0.005
0.052 ( 0.005
0.035 ( 0.004
ND
-
MeCN/H
MeCN/H
MeCN/H
MeCN/H
MeCN/H
2
O
2
O
2
O
2
O
2
O
2
. Ascorbate Bond Dissociation Free Energies in Pure
-3
-
-
-
3
3
2
Solvents. The free energy change of eq 1 in pure solvents (∆G°
1
)
is the difference in O-H BDFEs of the ascorbate and TEMPOH
-
∆G°1/RT
in that solvent. Thus, the measured values for K
1
() e
)
2
H O
DMSO
-
1
and the well-known TEMPO-H BDFEs (66.5 ( 0.5 kcal mol
-4
b
2 × 10
0.9 ( 0.1
1
7
-1
18
in MeCN, 67.5 ( 1.0 kcal mol in DMSO ) provide the
ascorbate O-H BDFEs in those solvents (Table 1). The bond
strength of ascorbate in water is derived using a well-established
a
At 298 K. Rate constants in M-1 s-1
Karl Fischer titration.
b
.
ꢀwater in “dry” solvent from
18,19
6
procedure
1.37pK
1/2 vs NHE ). The iAscH BDFE in MeCN derived from
from the known pK
a
and E1/2 values: BDFE(X-H)
-
-
1
Rate constants for ascorbate (AscH ) oxidation by TEMPO
were measured under pseudo-first-order conditions, as described
above, in solutions with the mole fraction of water (ꢀwater) from
)
E
a
+ 23.06E1/2 + C [C
G
G 2
(H O) ) 57.6 kcal mol with
1
8b
-
1
0
this equation is within error and less precise than that derived
-
5
2
0
∼
3 × 10 (“dry” acetonitrile, see Methods) to 0.012. These
1
from K , so the latter is given in Table 1.
values correspond to concentrations of water up to 200 mM, or
0.1-720 equiv of water per ascorbate. Qualitatively, the reaction
becomes substantially slower in the presence of even quite small
amounts of water. Second-order rate constants for reactions with
For both ascorbate and isopropylidene ascorbate, the O-H
-
BDFEs are the same in MeCN and DMSO, with that for AscH
about 1 kcal mol-1 stronger than iAscH . The O-H BDFE of
ascorbate is about 6 kcal mol larger in water than in the polar
organic solvents, as noted above. These solution BDFEs are
related to the gas-phase BDFE of ascorbate by the free energies
-
-1
ꢀ
water < 0.003 were determined by simple pseudo-first-order
kinetic analysis, as above. However, in reactions with higher
water concentrations, the decay of the ascorbyl radical is more
appreciable and had to be treated explicitly when fitting the data.
Above ꢀwater ) 0.012, the decay is too rapid to extract
meaningful kinetic results. Fortunately, the isopropylidene
derivative iAsc• decays more slowly so that experiments up
to ꢀwater ) 0.028 could be analyzed. The kinetic data were
globally fit with a model of a second-order approach-to-
equilibrium step followed by the second-order decay of the
-
•
•
of solution of AscH , Asc , and H , and about half of the
difference between the organic solvents and water can be
•
21
-1
ascribed to differences in ∆G°(H )solv
.
Still, the 6 kcal mol
strengthening of the ascorbate O-H bond between MeCN/
-
DMSO vs H
2
O is significantly larger than would be expected.
For substrates with one hydroxylic moiety, the change in BDFE
-1
from MeCN to water is typically +2-4 kcal mol and from
DMSO to water is about +1-3 kcal mol-1 (see Supporting
2
2
ascorbyl radical (Scheme 2). In all cases, the spectrum of
ascorbyl radical was fixed in SPECFIT to the known spectrum
Information).
3
2
. Rate and Equilibrium Constants in Mixed MeCN/H O
1
0
in pure acetonitrile, which is reasonable since the λmax and ε
of the radicals are similar in pure MeCN (377 nm/3900 ( 200
Solvent. To probe the substantial difference in BDFEs for
ascorbate in acetonitrile versus water, reaction 1 has been studied
in acetonitrile/water mixtures. The effects of other “additives”
are explored in the next section.
-
1
-1 10
-1
-1 23
M
cm ) and H
Equilibrium constants for these reactions with added water
are also determined using Scheme 2, with Keq given by the ratio
of the rate constants k /k-1. This is best done for reactions where
a limited amount of TEMPO is used. For reactions with ꢀwater
0.003, analysis with the model in Scheme 2 gives the same
values of k and K as obtained from the pseudo-first-order plots
and plots of [TEMPOH][Asc ]/[AscH ] vs [TEMPO] in MeCN
Table 1). The derived rate and equilibrium constants for
2
O (360 nm/3500 ( 200 M cm ).
(
(
17) Mader, E. A.; Manner, V. W.; Markle, T. F.; Wu, A.; Franz, J. A.;
1
Mayer, J. M. J. Am. Chem. Soc. 2009, 131, 4335–4345.
18a
18) The BDFE in DMSO is derived from the known
E
1/2 and pK
a
values
) 71.1
(a) Bordwell,
<
using BDFE(X-H) ) 1.37pK
a
+ 23.06E1/2 + C , with C
G G
-1
+/0
9,18b
kcal mol (ferrocene electrochemical reference).
1
1
•-
-
F. G.; Liu, W.-Z. J. Am. Chem. Soc. 1996, 118, 10819–10823. (b)
Mader, E. A. Hydrogen atom transfer reactions of iron and cobalt
alpha-diimines: A study of intrinsic and thermodynamic effects. Ph.D.
Thesis, University of Washington, 2007.
(
reactions with added water are given in Tables 2 and 3 and
shown graphically in Figure 2. To pick one example, the rate
constant for the iAscH reaction is a remarkable 3 times slower
in the presence of just ꢀwater ) 0.014, and the equilibrium
constant is 8 times smaller.
(
19) Tilset, M. The thermodynamics of systems involving electron-transfer
paths. In Electron Transfer in Chemistry; Balzani, V., Ed.; Wiley-
VHC: New York, 2001; Vol. 2, pp 677-713.
-
(
20) Bond dissociation free energies derived from thermochemical cycles
-
1
typically have errors of (1-2 kcal mol•
.
(
21) A more detailed discussion of ∆Gsolv°(H ) in different solvents can be
•
found in ref 18b. Briefly, ∆Gsolv°(H ) is well approximated by
-
1
(22) The second step, Asc•- disproportionation, was fit to the kinetic model
as a bimolecular process but could be mechanistically more compli-
∆
G
solv°(H ), which is (in kcal mol ) 5.12 in MeCN, 5.61 in DMSO,
2
18b
•
and 8.98 in water. Thus, the solvation of H contributes +3.9 kcal
-
1
mol to the difference in BDFEs between MeCN and H
2
O (3.4 kcal
1
cated. The derived values for k and k-1 do not appear to be very
-
1
mol for DMSO vs H
2
O). This is only one aspect of the solvation,
sensitive to the value of the disproportionation rate constant.
(23) The λ/ε given for aqueous ascorbate is the average of the values
reported in ref 12 and the following: Schuler, R. H. Radiat. Res. 1977,
69, 417–433.
•
however, and ∆Gsolv°(XH) is not necessarily close to ∆Gsolv°(X ) for
hydroxylic substrates ROH such as ascorbate, which can form strong
hydrogen bonds with H-bond-accepting solvents.
7
786 J. AM. CHEM. SOC. 9 VOL. 132, NO. 22, 2010

Tuning Thermochemical and Kinetic Properties of Ascorbate
A R T I C L E S
Table 3. Rate and Equilibrium Constants for Isopropylidene
this question, hydrogen atom transfer reactions of TEMPO,
TEMPOH, and ascorbate with other reagents have been studied
in MeCN as a function of added water.
-
a
Ascorbate (iAscH ) + TEMPO (Eq 1)
solvent
ꢀ
water
k
1i
K
1i
-
-
-
-
-
-
-
-
5
3
3
3
3
2
2
2
b
c
c
-
MeCN
3 × 10
1.5 × 10
3.6 × 10
7.2 × 10
7.2 × 10
1.2 × 10
1.4 × 10
2.8 × 10
1
1720 ( 150
1290 ( 100
1020 ( 100
780 ( 80
745 ( 80
525 ( 70
560 ( 70
240 ( 50
5.5 ( 0.4
250 ( 20
1.2 ( 0.2
5.1. Reactions with Phenoxyl Radicals. iAscH and TEM-
MeCN/H
MeCN/H
MeCN/H
MeCN/H
MeCN/H
MeCN/H
MeCN/H
2
O
2
O
2
O
2
O
2
O
2
O
2
O
0.65 ( 0.07
0.43 ( 0.06
0.38 ( 0.04
0.25 ( 0.03
0.13 ( 0.02
0.15 ( 0.02
0.065 ( 0.006
ND
•
POH each rapidly and completely transfer H to the 2,6-di-tert-
t
•
butyl-4-methoxyphenoxyl radical ( Bu
2
(MeO)ArO ). In pure
5
MeCN, both reactions have Keq ) 3 × 10 based on the relevant
25
BDFEs. The rate constants in MeCN with various amounts
of added water are shown in Figure 4 and given in Supporting
t
•
2
Information Table S1. For TEMPOH + Bu (MeO)ArO , the
2
H O
DMSO
-
4
b
rate constant at ꢀwater ) 0.028 is slower than that in pure MeCN
by only 13%. Similarly, the rate constant for TEMPOH +
2 × 10
0.85 ( 0.10
a
At 298 K. Rate constants in M s-1
-1
b
t
•
7
17
.
ꢀwater in “dry” solvent from
Bu
PhO (Keq ) 5 × 10 ) is reduced by only 15% at ꢀwater
)
3
c
Karl Fischer titration. Reference 10.
-
0
.028. In contrast, k1i for TEMPO + iAscH decreases by a
factor of 7 at this water concentration (Table 3). The mild
dependence of the TEMPOH rate constants on small amounts
of water is the expected behavior for a hydrogen atom transfer
reaction (see below). These reactions do not show the unusual
solvent effect.
4
. Effects of Other Additives on Rate and Equilibrium
Constants. Additives other than water also have a large effect
on the rate and equilibrium constants for reaction 1 in MeCN.
These experiments were done with iAscH- for experimental
convenience (see above). We have examined the effects of
In contrast, reactions of iAscH- with Bu
t
•
adding ethylene glycol (HOCH
CH(OH)CH OH], dimethoxyethane (DME, MeOCH
OMe), diethanolamine [(HOCH CH
methyl ester [MeC(O)NHCH( Bu)C(O)OMe], DMSO, choline
chloride (HOCH
ride (Me
chosen to include both hydrogen bond donors and acceptors,
molecules with biological-type functional groups, and charged
species. Acidic compounds, such as NH
because they would protonate the ascorbate substrate under these
conditions. For each additive, k1i and K1i were determined as
described above.
Most of the additives cause a decrease in rate and equilibrium
constants, consistent with the effect of added water (Table 4
and Figure 3). The exceptions are DME, diethanolamine,
2
CH
2
OH), glycerol [HOCH
2
-
2
(MeO)ArO are
2
2
CH
2
-
substantially slowed by added water, as shown in Figure 4. For
instance, this reaction is slower by 38% at ꢀwater ) 0.0038 vs
pure MeCN. This reduction is remarkably similar to the decrease
2
2 2
) NH], N-acetyl-leucine-
i
+
-
-
2
CH
2
NMe
3
Cl ), tetramethylammonium chlo-
of 35% for TEMPO + iAscH at the same ꢀwater. This water
+
-
n
+
-
4
N Cl ), and, as noted above, Bu
4
N PF
6
. These were
concentration is 7 times less than the amount that causes only
a 13-15% decrease in the TEMPOH reactions. In sum, the
t
•
2
reaction of ascorbate with Bu (MeO)ArO shows the same
+
-
4
Cl , cannot be used
unusual slowing with added water, but the reactions of
TEMPOH do not. The similarity between the water effects for
-
t
•
2
the two reactions of iAscH , with Bu (MeO)ArO and with
TEMPO, suggests that this effect is due to the ascorbate rather
than the hydrogen abstractor.
5
.2. Hydrogen Atom Transfer from a Ruthenium Com-
plex to TEMPO. Added water could also potentially slow k
1
+
-
n
+
-
Me
the reaction. The lack of any change with 100 mM Bu
is in striking contrast to the effect of only 6.5 mM choline
4
N Cl , and Bu
4
N PF
6
, which have little or no effect on
•
and k1i by forming a hydrogen bond to the R N-O moiety of
2
n
+
-
N PF
4 6
TEMPO, which is a moderately strong hydrogen bond accep-
26
tor. To test this possibility, we have explored the effect of
-4
chloride (ꢀ ) 3.4 × 10 ), which decreases k1i and K1i by factors
•
added water on the ability of TEMPO to remove H from
of 1.6 and 2.9, respectively. DMSO has a much weaker effect
II
27
(
2
acac)
2
Ru (py-imH) [acac ) 2,4-pentanedionato, py-imH )
-3
than other additives: at ꢀ ) 7.7 × 10 , DMSO lowers k1i by
a factor of 1.3, while ethylene glycol causes a 5.7-fold rate
decrease.
These shifts in k1i and K1i presumably result from the additive
binding to or in some way differentially solvating the ascorbate
and/or ascorbyl radical. To probe such a binding, optical spectra
of ascorbate in the presence of ethylene glycol or choline
chloride have been examined, but no changes were observed.
-(2′-pyridyl)imidazole]. This reaction has been previously
-1
examined in pure MeCN, where ∆G° ) -4.4 ( 0.1 kcal mol
3
-1 -1 27b
(
K
eq ) 1800 ( 200) and k ) (1.4 ( 0.1) × 10 M s .
-5 -3
Between ꢀwater ) 3 × 10 and 7.2 × 10 , the measured rate
constant is within the 7% error bars of the value in pure MeCN
-2
(
see Supporting Information). At ꢀwater ) 2.8 × 10 , there is a
decrease in k by a factor of 1.4. In contrast, this concentration
of water reduces k1i by a factor of 7.2 (Table 3).
1
H NMR spectra, however, show downfield shifts of the
t
resonance for the proton at C4 for both AscH- and iAscH-
5.3. Hydrogen Atom Transfer from Bu
2
NOH to TEMPO.
The effect of added water on the reaction of TEMPO with
(
see Scheme 1 for numbering) with either ethylene glycol or
t
Bu
is most conveniently studied in the reverse direction, Bu
TEMPOH, but since the reaction is close to isoergic, the
2
NOH has also been explored. Experimentally, this reaction
choline chloride. The addition of choline chloride has a
qualitatively larger effect on the chemical shift at lower
concentrations than ethylene glycol, consistent with the above
t
NO•
2
+
24
approach-to-equilibrium kinetics yield the forward and reverse
rate constants as well as the equilibrium constant. In pure MeCN,
results that choline chloride has the larger effect on k
. Other Reactions of TEMPO, TEMPOH, and Ascorbate.
In principle, the changes in k and K with increasing water or
1 1
and K .
5
1
1
other additives could be due primarily to changes in the
ascorbate/ascorbyl radical couple, to changes in the TEMPO/
TEMPOH couple, or to changes in both couples. To resolve
(
25) Waidmann, C. R.; Zhou, X.; Tsai, E. A.; Kaminsky, W.; Hrovat, D. A.;
Borden, W. T.; Mayer, J. M. J. Am. Chem. Soc. 2009, 131, 4729–
4
743.
(
26) Frachi, P.; Lucarini, M.; Pedrielli, P.; Pedulli, G. F. ChemPhysChem
2
002, 3, 789–793.
-
(
24) Similar NMR shifts are observed for iAscH + choline chloride. A
(27) (a) Wu, A.; Masland, J.; Swartz, R. D.; Kaminsky, W.; Mayer, J. M.
Inorg. Chem. 2007, 46, 11190–11201. (b) Wu, A.; Mayer, J. M. J. Am.
Chem. Soc. 2008, 130, 14745–14754.
-
1
plausible quantitative analysis of the AscH H NMR shifts is given
in the Supporting Information.
J. AM. CHEM. SOC. 9 VOL. 132, NO. 22, 2010 7787

A R T I C L E S
Warren and Mayer
Figure 2. Plots of rate constants [red solid symbols, left axes] and equilibrium constants [blue open symbols, right axes] as a function of mole fraction of
added distilled water in MeCN for the reactions of (a) ascorbate with TEMPO (k , K ) and (b) isopropylidene ascorbate with TEMPO (k1i, K1i).
1 1
-
Table 4. Rate and Equilibrium Constants for iAscH + TEMPO
water (∼200 mM water) versus the reaction in pure MeCN.
a
(
Eq 1) in MeCN with Additives
With the more experimentally tractable isopropylidene ascorbate
+
-
additive
ꢀ in MeCN
k
1i
K
1i
derivative, [DBU-H] iAscH , the presence of this amount of
water decreases k1i and K1i by factors of 3.3 and 9.6, respectively.
This effect is not limited to water: the addition of ca. 1 mol %
ethylene glycol, choline chloride, or N-acetyl-leucine-methyl
ester to the MeCN solvent causes k1i to fall by more than a
factor of 2 and K1i to drop by at least a factor of 8.
none
b
1720 ( 150
1170 ( 80
300 ( 40
1.2 ( 0.2
0.65 ( 0.09
0.085 ( 0.010
0.87 ( 0.10
1.2 ( 0.2
0.90 ( 0.10
0.35 ( 0.05
0.50 ( 0.05
0.40 ( 0.05
1.1 ( 0.2
-
-
-
-
-
4
3
5
3
4
ethylene glycol
ethylene glycol
glycerol
7.0 × 10
7.7 × 10
5.1 × 10
4.8 × 10
2.7 × 10
3.0 × 10
7.7 × 10
2.8 × 10
5.5 × 10
3.4 × 10
9.5 × 10
5.2 × 10
1500 ( 100
1680 ( 130
1600 ( 120
1150 ( 100
1360 ( 100
1100 ( 100
1410 ( 120
1100 ( 100
1550 ( 150
1720 ( 150
DME
diethanolamine
N-AcO-Leu-OMe
DMSO
DMSO
choline chloride
c
-3
The results reported here show that the changes observed in
rate and equilibrium constants with various additives are due
to changes in the reactivity of ascorbate, not TEMPO or
-
-
3
2
d
-5
-
-
TEMPOH. The reactions of AscH + TEMPO, iAscH
+
-
4
choline chloride
0.46 ( 0.06
0.9 ( 0.1
1.2 ( 0.2
-
t
•
+
-
-4
-3
2
TEMPO, and AscH + Bu (MeO)ArO all show this remark-
Me
Bu
4
N Cl
n
+
-
4
N PF
6
able effect. In contrast, none of the reactions of TEMPO or
TEMPOH show substantial changes in their rate or equilibrium
constants upon addition of water, except for the reactions of
TEMPO with the ascorbates. The TEMPO/TEMPOH couple
appears to behave like a typical oxyl radical, while the ascorbate/
ascorbyl radical couple is anomalous. This conclusion is
supported by the unusual solvent dependence of the BDFE of
ascorbate, as discussed above and in the Supporting Informa-
tion).
a
At 298 K. Rate constants in M s-1
-1
b
-5
.
ꢀ
H O = 3 × 10 present in
2
c
i
“
dry” MeCN. N-Acetyl-leucine-methyl ester (MeC(O)NHCH( Bu)C(O)-
d
+
-
2 2 3
OMe). HOCH CH NMe Cl .
2
8
t
based on the reported results, TEMPO + Bu
2
NOH has ∆G°
1
)
-1.3 ( 0.1 kcal mol- (Keq ) 9.0 ( 1.5) and k ) 17.3 (
-
2
3
.6. We find that addition of water to ꢀ ) 1.4 × 10 has
essentially no effect on either the forward or the reverse rate
2
. Kinetic and Thermodynamic Results Indicate That
-
2
constants, or the equilibrium constant. At ꢀwater ) 2.8 × 10
Ascorbate Is Strongly Affected by Local Solvation. The large
variations in rate and equilibrium constants for reaction 1 cannot
be explained by variation of the bulk properties of the solvent,
such as the dielectric constant ꢀ, nor by a common kind of
kinetic solvent effect observed for other HAT reactions (see
below). The presence of water in the small amounts used here
t
•
there is no change in rate constant for TEMPO + Bu
2
NO ,
although k for the reverse reaction and the equilibrium constant
both decrease by about 15%. Thus, neither the reaction of
II
t
TEMPO with (acac)
2
2
Ru (py-imH) nor TEMPO + Bu NOH
show, the substantial effect of added water that is observed for
TEMPO + ascorbates. This effect of additives is thus indicated
to be a property of ascorbate rather than TEMPO.
2
9
(ꢀwater < 0.03) has virtually no affect on ꢀ. Furthermore, the
effects of the other additives on reaction 1 are also inconsistent
with changes in ꢀ. For example, the dielectric constants of pure
MeCN (37.5) and ethylene glycol (37.7) are nearly identical,
Discussion
3
0
1
. Overview. Most biochemical reactions of ascorbate, as is
yet a small amount of ethylene glycol in the MeCN solvent (ꢀ
4,7
-4
known but not always appreciated, initially form the ascorbyl
radical and therefore involve loss of e and H or, equivalently,
H . These reactions are best described as hydrogen atom transfer
HAT) (or concerted proton-electron transfer ). The results
) 7.0 × 10 , 13.5 mM) reduces k by 32% and K by 48%.
1i
1i
-
+
2
With the ionic material choline chloride (HOCH -
•
+
-
-4
CH NMe Cl ), addition of only 13 equiv (ꢀ ) 3.4 × 10 ,
2
3
2
5
(
6.5 mM) reduces k1 and K by factors of 1.6 and 2.9,
i 1i
reported here show that small changes in the reaction medium
cause surprisingly large changes in the kinetics and thermody-
namics of the H-atom transfer reactions of ascorbates with
respectively. This is not due to an ionic strength or ion pairing
effect, since a much larger concentration of an inert salt, 100
n
+
-
mM Bu N PF , has no effect on the rate or equilibrium
4
6
t
•
-
TEMPO and with Bu
2
(MeO)ArO . The reaction of the potas-
constants. The equilibrium constant for iAscH + TEMPO
sium-18-crown-6 salt of ascorbate with TEMPO is 1.9 times
slower and 4.0 times less favorable in MeCN with ca. 1 mol %
(
(
29) Jellema, R.; Bulthuis, J.; van der Zwan, G. J. Mol. Liq. 1997, 73-74,
1
79–193.
30) Riddick, J. A.; Bunger, W. B. In Organic SolVents: Physical Properties
and Methods of Purification; Weissberger, A., Ed.; Techniques of
Chemistry Series 2; Wiley-Interscience: New York, 1970.
(
28) Wu, A.; Mader, E. A.; Datta, A.; Hrovat, D. A.; Borden, W. T.; Mayer,
J. M. J. Am. Chem. Soc. 2009, 131, 11985–11997.
7
788 J. AM. CHEM. SOC. 9 VOL. 132, NO. 22, 2010

Tuning Thermochemical and Kinetic Properties of Ascorbate
A R T I C L E S
Figure 3. Plots of (a) rate constants [k1i] and (b) equilibrium constants [K1i] for the reaction of isopropylidene ascorbate with TEMPO (eq 1) in MeCN as
a function of mole fraction of various additives. Error bars have been omitted for clarity, see Tables 24.
3
3
substrate to the solvent, ROH · · ·Solv. Their KSE model
indicates that only the non-hydrogen-bonded substrate is reactive
toward HAT, so reactions are slower in solvents that are better
hydrogen bond acceptors. They gauge the H-bond-accepting
ability of solvent using the hydrogen bond basicity parameters
H
34
(
ꢁ
2
) developed by Abraham. This model does predict the
slower rates for reaction 1 in the better H-bond-accepting DMSO
H
H
(ꢁ
2
) 0.78) vs MeCN (ꢁ
2
) 0.44) (Tables 2 and 3). However,
the large rate deceleration with the addition of small amounts
of water is not consistent with the KSE model. Since MeCN is
H
a slightly better H-bond acceptor than H
>
2
O (ꢁ
2
(MeCN) ) 0.44
H
2 2
ꢁ (H O) ) 0.38), the KSE model predicts that H-transfer
rate constants should be unaffected by small amounts of water.
These results show that the major effects of additives on the
reactions of ascorbate are not due to a hydrogen bonding
interaction that blocks the transfer of the ascorbate hydrogen.
If this were the case, ethylene glycol and DME would be
Figure 4. Relative rate constants, krel ) k(ꢀwater)/k(ꢀwater ) 0), as a function
of mole fraction water in MeCN for TEMPOH + 2,4,6-tri-tert-butylphenoxyl
-
(
O), TEMPOH + 2,4-di-tert-butyl-4-methoxyphenoxyl (b), and iAscH
+
2,4-di-tert-butyl-4-methoxyphenoxyl (9).
1
expected to show very similar effects on the rate constants (k ).
Thus, ascorbate shows a different type of solvent effect on its
HAT reactivity that is distinct from the Ingold KSE model for
HAT reactions.
decreases by 1800% as water concentration is raised to ꢀwater
)
0
.028. This decrease is vastly larger than would be predicted
-1
from ideal solution behavior, since the reaction is 2.5 kcal mol
less favorable in pure H O vs pure MeCN (see below) and 2.8%
of 2.5 kcal mol- would imply a barely noticeable change in
It is striking that small amounts of water and other additives
affect both the rates of reaction and the equilibrium constants.
In fact, the equilibrium constants change substantially more than
the rate constants. This contrasts with the KSE model, which
does not correlate with the (ground-state) reaction thermody-
namics because it involves hydrogen bonding only with the
reactants and not with the products. Brønsted plots of ln(k) vs
ln(K) for the ascorbate and isopropylidene ascorbate reactions
at different water concentrations are linear (Figure 5). Figure
2
1
K1i of 12%.
If the large variations in rate and equilibrium constants with
small amounts of additives are not due to changes in the bulk
properties of the solvent, then they must be due to differences
in the local environment in which the HAT reaction occurs. As
shown below, it is the presence of H-bond donors in the local
environment that is causing the observed effect.
5
b also includes values for all of the other additives, showing
Solvent effects on HAT reactions have been examined in
many systems, particularly the effects of solvent polarity and
that water and the other materials all behave very similarlysa
3
1
remarkable and unexpected result. The slopes of the plots are
viscosity on selectivity. For HAT reactions involving C-H
q
3
2
the Brønsted R values, or equivalently, ∆∆G /∆∆G°. For
bonds, solvent effects are usually very small (which is why
gas-phase C-H bond strengths can be used to rationalize
reactions in solution). For hydroxylic substrates (ROH), Ingold,
Litwinienko, and co-workers have found substantial kinetic
solvent effects (KSEs) due to hydrogen bonding from the
-
-
AscH , R ) 0.48 ( 0.03. For iAscH , the slope is 0.61 ( 0.04
including all additives, and a very similar 0.60 ( 0.05 for just
-
the MeCN + H
2
O reactions of iAscH . A combined plot with
all the data shows a very strong linear trend among all the data
-
-
(
iAscH and iAscH ), with a slope of 0.58 ( 0.04 and a
correlation coefficient of 0.96 (Figure 5c).
(
(
31) Tanko, J. M.; Suleman, N. K. Solvent Effects in the Reactions of
Neutral Free Radicals. In Energetics of Organic Free Radicals;
Martinho Sim o˜ es, J. A., Greenburg, A., Liebman, J. F., Eds.; SEARCH
Series 4; Blackie: Glasgow, 1996; pp 244-293.
32) Avila, D. V.; Brown, C. E.; Ingold, K. U.; Lusztyk, J. J. Am. Chem.
Soc. 1993, 115, 466–470.
(33) Litwinienko, G.; Ingold, K. U. Acc. Chem. Res. 2007, 40, 222–230,
and references therein.
(34) Abraham, M. H.; Grellier, P. L.; Prior, D. V.; Morris, J. J.; Taylor,
P. J. J. Chem. Soc., Perkin Trans. 2 1990, 521–529.
J. AM. CHEM. SOC. 9 VOL. 132, NO. 22, 2010 7789

A R T I C L E S
Warren and Mayer
-
2
Figure 5. Plots of rate constants ln(k
1
) versus equilibrium constants ln(K
1
) for the reaction of (a) AscH in MeCN with added water [2, R ) 0.99), (b)
-
2
iAscH with TEMPO in MeCN with different additives [9, reactions with added water, R ) 0.93; 0, reactions with other additives], and (c) the aggregate
2
data set (R ) 0.96).
Scheme 3. Effect of Additives A on the Reaction of AscH- with
TEMPO
Rate/driving force relationships in organic HAT reactions
3
5
have long been studied, although we are unaware of an
example such as this with small amounts of additives. Tradition-
ally these correlations have used reaction enthalpies, but we
9
have found that free energies are more appropriate. In addition,
we have found that a Marcus theory formalism connecting k
3
6,37
and K is very valuable for understanding HAT reactions.
For reactions such as these with small driving forces (∆G° ,
The effect of additives on the equilibrium constants can be
analyzed using a thermochemical cycle such as Scheme 3,
although this is probably a simplification because the binding
is probably not well described by a 1:1 model. In this model,
λ/2), Marcus theory predicts a linear Brønsted plot with a slope
3
8
of 0.5, which is in excellent agreement with the observed
results. Thus, the obserVed changes in rate constant stem
predominantly from the changes in the driVing force of the
reactions.
the difference in the measured equilibrium constants, K1A/K
1
,
is equal to the difference in binding constants for the additive
3
. Local Solvation or Binding Shifts the Thermochemistry
-
•-
A with AscH and Asc (KAsc•/KAscH). The binding constants
AscH and KAsc• in this analysis are analogous to the K values
of Ascorbate. Small amounts of water and other materials
strongly shift the equilibrium constant for the reaction of
ascorbate and isopropylidene ascorbate with TEMPO. As shown
in the previous section, this is due to the local solvation or
binding of the additive to the ascorbate and/or ascorbyl radical.
The diverse range of additives gives insight into the nature of
this binding or solvation. Ethylene glycol at 0.8% markedly
lowers both k1i and K1i (factors of 5.7 and 20, respectively),
while DME has essentially no effect at a similar mole fraction.
Ethylene glycol and DME are similar in size and are comparable
K
d
-
•-
for AscH and Asc binding to a protein (see next section).
Qualitatively, when an additive (or enzyme binding pocket)
binds ascorbate more tightly than the ascorbyl radical, the HAT
1
step will be less favorable (K1A < K ). Since these equilibrium
constants are related to the O-H BDFE of the ascorbate,
hydrogen-bond-donating environments effectively increase the
BDFE. A more detailed discussion of BDFEs is given in the
Supporting Information.
-
•-
3
9
The difference in binding to AscH and Asc is presumably
due to the difference in charge distribution in the two anions,
as indicated in the drawings in Scheme 1. Ascorbate has a single
enolate oxygen that formally carries the negative charge, while
the ascorbyl radical has a more delocalized triketone-like
hydrogen bond acceptors, but only the former can be a
hydrogen bond donor. Thus, it is the hydrogen-bond-donating
character of the material that is critical. This is again distinct
from Ingold’s KSE observations, where the hydrogen-bond-
accepting character of the solvent is the key contributor to the
rate constant for hydrogen atom transfer. The importance of
hydrogen bond donation is also indicated by the very small effect
of DMSO, a good hydrogen bond acceptor, and by the much
-
structure. This is reflected in the difference in basicity of AscH
and Asc• (pK
-
values of the conjugate acids are 4.0 and -0.45,
a
,8
6
respectively ). This issue has been addressed by Costanzo et
al. using DFT calculations and molecular dynamics simulations.
They show that all three exocyclic oxygen atoms carry
+
-
larger effect of choline chloride (HOCH
2
CH
2 3
NMe Cl ) vs
+
-
Me
4
N Cl . All of the results reported here are consistent with
-
•-
significant negative charge in both AscH and Asc , with all
the atomic charges being smaller in the radical. Molecular
dynamics simulations indicate that in aqueous solution, the
hydrogen bond donation from the additive to ascorbate playing
the key role in the observed changes in k and K.
4
0
-
enolate oxygen of AscH is the preferred hydrogen bonding
(
35) (a) Ingold, K. U. In Free Radicals; Kochi, J. K., Ed.; Wiley-
Interscience: New York, 1973; Vol. 1, pp 69-70. (b) Russell, G. A.
In Free Radicals; Kochi, J. K., Ed.; Wiley-Interscience: New York,
site. Oxidation to the radical causes the loss of 0.6 of a hydrogen
bond at this oxygen, consistent with the change in calculated
charges. Thus, these calculations appear to support the conclu-
1
973; Vol. 1, pp 283-293.
-
(
(
36) Mayer, J. M. Annu. ReV. Phys. Chem. 2004, 55, 363–390.
sion that hydrogen bond donors bind more strongly to AscH
than to Asc .
37) Roth, J. P.; Yoder, J. C.; Won, T.-J.; Mayer, J. M. Science 2001, 294,
•-
2
524–2526.
(
(
38) Marcus, R. A. J. Phys. Chem. 1968, 72, 891–899.
39) Alkyl ethers and primary alcohols have essentially identical hydrogen
(40) Costanzo, F.; Sulpizi, M.; Vandevondele, J.; Della Valle, R. G.; Sprik,
H
bond basicities (ꢁ
2
). See ref 34.
M. Mol. Phys. 2007, 105, 17–23.
7
790 J. AM. CHEM. SOC. 9 VOL. 132, NO. 22, 2010

Tuning Thermochemical and Kinetic Properties of Ascorbate
A R T I C L E S
Figure 6. (a) Proposed structural model for bovine cytochrome b561, modified from ref 44, showing the ascorbate/ascorbyl radical conversions on both sides
of the membrane. Fe ) iron-heme and H ) histidine. (b) Depiction of a local environment for the ascorbate reduction of R-tocopheroxyl in a lipid bilayer,
as adapted from refs 49 and 52. The bars at the right delineate the regions having different levels of hydration (H-bond-donating water).
Two other reports have described unusual solvent effects on
bond-accepting character that would thermodynamically favor
formation of ascorbate or ascorbyl, respectively? Below we
briefly discuss three in ViVo processes in which modulation of
the thermochemistry of ascorbate by the local H-bonding
environment could play an important role.
1
6,41
HAT reactions of ascorbate.
The reduction of PhNO by
ascorbate is 40% slower in 1:1 water/dioxane than that in pure
water, which is twice as large as the maximum KSE-predicted
4
1
shift in k from pure water to pure dioxane. In addition, the
O/D O kinetic isotope effect increases from 2.4 to 8.5 from
H
2
2
Cytochrome b561 is a transmembrane protein, found in both
water to water/dioxane. While these effects could be a result of
bulk polarity changes, it seems likely that the issues of local
solvation implicated here are also playing a role. A very recent
study of the reduction of TEMPO by ascorbate in water/dioxane
mixtures reported large kinetic isotope effects indicative of
plants and animals, that is thought to mediate ascorbate recycling
43,44
across the membrane (Figure 6a).
it can catalyze the reduction of the ascorbyl radical to ascorbate.
In mammalian chromaffin b561, the putative Asc binding site
In Vitro studies show that
•-
on the intravesicular side of the membrane is fairly hydrogen-
bond-donating, with the sequence -xYSLHSWxGx- (x is a
1
6
hydrogen tunneling. Tunneling has also been implicated in
2
7b,28
4
3
other HAT reactions of TEMPO.
It is not clear how
hydrophobic residue). On the extravesicular side of the lipid
tunneling is contributing to the local solvent effect discussed
here. However, it seems unlikely that tunneling effects play a
dominant role, given the very strong Brønsted correlations
between rate and equilibrium constants. We also note that
Amorati et al. have reported a solvent effect that is distinct from
-
bilayer, the putative AscH binding site has fewer hydrogen-
bond-donating groups (-ALLVYRVFR-), as well as a well
conserved lysine residue (Lys85), proposed to play an electro-
43
static role in binding. Based on the results presented here,
•-
the H-bonding environment of the intravesicular Asc binding
4
2
Ingold’s KSE model. They found that the kinetics and
thermodynamics of HAT from hydroquinones to 2,2′-diphe-
nylpicrylhydrazyl radical (DPPH) or peroxyl radicals are af-
fected by hydrogen bonding to the nontransferring hydrogen
of the hydroquinone. Such a “remote” hydrogen bonding effect
may be analogous to what is reported here, but their study
involved hydrogen bond acceptors, whereas the ascorbate
reactions above are modulated by hydrogen bond donors.
-
pocket thermodynamically favors formation of AscH , and the
-
extravesicular AscH binding pocket favors the reverse reaction.
-
•-
While the mechanism of AscH /Asc interconversion may be
45
more complicated than the simple HAT reactions studied here,
it is likely that the kinetic barriers reflect the energetics, so this
•-
effect should facilitate the reduction of Asc on the intrave-
-
sicular side of the membrane and the oxidation of AscH on
the extravesicular side.
4. Biological Implications. The results reported here show that
changes in the hydrogen-bond-donating environment can have
a substantial impact on the reactivity of ascorbate. In particular,
hydrogen-bond-donating environments increase the ascorbate
O-H bond strength, making ascorbate a poorer reductant. Given
the wide range of biological processes involving ascorbate, it
seems likely that Nature has taken advantage of this apparently
unique feature. For example, do ascorbate binding pockets in
certain enzymes have hydrogen-bond-donating or hydrogen-
(
(
43) Verelst, W.; Asard, H. Genom. Biol. 2003, 4, R38.1R38.12.
44) Nakanishi, N.; Takeuchi, F.; Tsubaki, M. J. Biochem. 2007, 142, 553–
5
60.
(45) It has been suggested that the ascorbate oxidation reaction at the
extravesicular active site utilizes an essential histidine residue that acts
4
,45ab
as a proton acceptor from ascorbate.
If this histidine is bound to
the iron that accepts the electron, we would still call this H-atom
10
transfer, but it is possible that it could have different properties than
the classical HAT from ascorbate to TEMPO described here. More
detailed structural data would aid in understanding the nature of the
ascorbate/ascorbyl binding pockets and the potential proton acceptors
and hydrogen bonding interactions. (a) Bashtovyy, D.; B e´ rczi, A.;
Asard, H.; P a´ li, R. Protoplasma 2003, 221, 31–40. (b) B e´ rczi, A.;
Su, D.; Lakshminarasimhan, N.; Vargas, A.; Asard, H. Arch. Biochem.
Biophys. 2005, 443, 82–92.
(
(
41) Vuina, D.; Pilepi c´ , V.; Ljubas, D.; Sankovi c´ , K.; Sajenko, I.; Uri sˇ i c´ ,
S. Tetrahedron Lett. 2007, 48, 3633–3637.
42) Amorati, R.; Franchi, P.; Pedulli, G. F. Angew. Chem., Int. Ed. 2007,
4
6, 6336–6338.
J. AM. CHEM. SOC. 9 VOL. 132, NO. 22, 2010 7791

A R T I C L E S
Warren and Mayer
-
Ascorbate peroxidase (APX) has an ascorbate binding pocket
similar for the potassium 18-crown-6 salt of ascorbate (AscH )
46
+
that is well characterized. It contains several hydrogen bonding
interactions (e.g., Lys30, Arg172, and the heme-propionate),
and is fairly exposed to solvent. This pocket thus appears to
hold ascorbate in an environment closer to that of aqueous
solution, with a high bond strength (high one-electron/one-proton
and the more soluble DBU-H -isopropylidene ascorbate
(iAscH ). Addition of water up to 1-3 mol % causes decreases
-
-
by factors of 2-7 in k
1
and almost 4-20 in K
1
. For AscH , the
at only 1.2 mol % water is a quarter of
4.5-fold decrease in K
1
the total change from pure MeCN to pure water. Other
hydrogen-bond-donating additives such as glycerol, choline
chloride, and N-acetyl-leucine-methyl ester have similar effects.
On the other hand, the hydrogen bond acceptors DME and
DMSO do not significantly affect the reaction outside of normal
kinetic solvent effects. Thus, the changes in rate and equilibrium
constants are much larger than can be explained by changes in
bulk solvent properties such as the dielectric constant or ionic
strength. In addition, since the changes in k and Keq are due to
hydrogen bond donors but not acceptors, this effect is clearly
distinct from Ingold’s kinetic solvent effect model for hydrogen
atom transfer reactions.
The changes in rate constant directly parallel the changes in
equilibrium constant, with Brønsted plots, ln(k) vs ln(K), being
linear with slopes close to the theoretical value of 0.5 (0.48 for
AscH and 0.61 for iAscH ). Therefore, the changes in rate
are predominantly a consequence of the changes in the driving
force for the reactions. This contrasts with previously described
kinetic solvent effects in oxyl radical reactions, which are purely
kinetic effects due to hydrogen bonding to the transferring
hydrogen atom. The results show that ascorbate is a stronger
reductantsbetter able to donate a proton and an electron (or
+
-
redox potential). The aqueous 1H /1e ascorbate potential is
8
II/III
0
.33 V at pH 7, somewhat above the outer-sphere Fe
47
potential of the heme in APX of +0.206 V. Since the catalytic
48
cycle of APX requires an iron(III) resting state, it is important
that the bound ascorbate not be any more reducing than aqueous
ascorbate. Thus, the enzyme has apparently evolved to have a
strongly hydrogen-bond-donating binding site for ascorbate to
avoid reduction of the resting ferric state.
The ascorbate reduction of the R-tocopheroxyl radical, to give
the ascorbyl radical and R-tocopherol (vitamin E), is perhaps
one of the most discussed and studied reactions of ascorbate.
4
9,50
This simple HAT reaction in Vitro is very well documented,
5
1
though its in ViVo importance is debated. R-Tocopherol is
membrane bound, likely with its phenolic moiety near the
49
-
-
water-lipid interface. Conversely, ascorbate is hydrophilic and
does not significantly partition into lipids, although there is
evidence that it can localize in the polar lipid headgroups at
the surface of the membrane. This membrane surface region
is polar but less hydrated and contains fewer H-bond donors
than the cellular bulk (Figure 6b). On the basis of our results,
ascorbate at the surface of the membrane is predicted to act as
a stronger H-atom donor (reductant) to tocopheroxyl radical than
ascorbate in aqueous solution. In addition, the O-H bond in
R-tocopherol should be a little stronger in the polar headgroup
region than in the bulk lipid. Thus, the “intermediate” region
at lipid-water interfaces appears to be an ideal location for
Nature to maximize the driving force for, and therefore the rate
5
2
5
3
•
4,7,8
H ) sin local environments with fewer hydrogen bond
donors. The change in reducing power of ascorbate is also
indicated by the change in bond strength (BDFE) in pure
solvents (Table 1). The unusually large solvation/binding effect
of ascorbate is suggested to derive from the substantial differ-
ences in charge distribution between ascorbate and the more
delocalized ascorbyl radical. This study also indicates that the
reactivity of ascorbate can, in many cases, be best understood
5
4
5
5
-
of, H-transfer from AscH to R-tocopheroxyl. Rapid reduction
of R-tocopheroxyl is valuable to minimize its activity as a pro-
5
6
oxidant.
using the free energy for loss of both a proton and an electron
•
(
H ).
Conclusions
The tuning of the properties of ascorbate by its local
environment is likely to be of importance in many of the
biological actions of ascorbate. Polar, protic environments favor
ascorbate, while environments without hydrogen bond donors
favor the ascorbyl radical. Three examples illustrate different
ways in which Nature may have utilized this unique property,
to facilitate ascorbate recycling by cytochrome b561, to avoid
unwanted side effects in ascorbate peroxidase, and in the
regeneration of R-tocopherol by ascorbate. As new reactivity
and structural information emerges about ascorbate-utilizing
proteins, specific hydrogen bonding interactions should be
considered with respect to the reactivity of the enzyme.
Rate and equilibrium constants for the oxidation of ascorbate
by the aminoxyl radical TEMPO (eq 1) in acetonitrile are very
sensitive to small amounts of various additives. The results are
(
(
46) Sharp, K. H.; Mewies, M.; Moody, P. C. E.; Raven, E. L. Nat. Struct.
Biol. 2003, 10, 303–307.
47) Efimov, I.; Papadopoulou, N. D.; McLean, K. J.; Badyal, S. K.;
Macdonald, I. K.; Munro, A. W.; Moody, P. C. E.; Raven, E. L.
Biochemistry 2007, 46, 8017–8023.
(
(
(
48) Raven, E. L. Nat. Prod. Rep. 2003, 20, 367–381.
49) Burton, G. W.; Ingold, K. U. Acc. Chem. Res. 1986, 19, 194–201.
50) May, J. M.; Qu, Z.; Mendiratta, S. Arch. Biochem. Biophys. 1998,
3
49, 281–289.
(
(
(
(
51) Hamilton, I. T. J.; Gilmore, W. S.; Benzie, I. F. F.; Mulholland, C. W.;
Strain, J. J. Br. J. Nutr. 2000, 84, 261–267.
Methods
52) Crans, D. C.; Baruah, B.; Gaidamauskas, E.; Lemons, B. G.; Lorenz,
B. B.; Johnson, M. D. J. Inorg. Biochem. 2008, 102, 1334–1347.
53) Nagle, J. F.; Tristram-Nagle, S. Biochim. Biophys. Acta 2000, 1469,
Physical Techniques and Instrumentation. UV/vis stopped-
flow measurements were obtained using an OLIS RSM-1000
1
59–195.
1
instrument. H NMR spectra were recorded on a Bruker Avance
54) This prediction assumes that R-tocopherol is not affected by H-bond
donors in the same unusual way that ascorbate is and that the
R-tocopherol BDFE is not unusually shifted by local solvation. This
assumption is supported by the BDFE data in the Supporting
Information and by reports of similar solvent effects on R-tocopherol
and phenol reactions in a variety of solvents: Valgimigli, L.; Banks,
J. T.; Ingold, K. U.; Lusztyk, J. J. Am. Chem. Soc. 1995, 117, 9966–
3
00 or 500 MHz spectrometer. All reactions were performed in
the absence of air using standard glovebox and high-vacuum
techniques.
Materials. Reagent-grade solvents were purchased from Fischer
Scientific, unless otherwise noted. All other chemicals were
purchased from Aldrich. Anhydrous acetonitrile was purchased from
Honeywell Burdick and Jackson, sparged with Ar, and plumbed
9
971.
(
(
55) In general, homolytic bond strengths increase from nonpolar to polar
media. This is in part due to changes in the free energy of solvation
directly into a N
2
-filled glovebox with stainless steel tubing. The
•
supplier specification for water content in the MeCN is 5 ppm (ꢀwater
of H (see Supporting Information).
-5
56) Bowry, V. W.; Stocker, R. J. Am. Chem. Soc. 1993, 115, 6029–6044.
) 1.5 × 10 M); our independent measurement by Karl Fischer
7
792 J. AM. CHEM. SOC. 9 VOL. 132, NO. 22, 2010

Tuning Thermochemical and Kinetic Properties of Ascorbate
A R T I C L E S
O. H NMR (DMSO-d ): δ
3.24 (m, 1H), 3.34 (m, 1H), 3.45 (m, 1H), 3.55 (s, 24 H), 3.85 (d,
1H).
1
titration was 10 ppm. Benzene and DMSO were dried using a
Grubbs-type catalyst in a Seca solvent system. Deuterated solvents
were obtained from Cambridge Isotope Laboratories. TEMPOH,
and washed once with ∼10 mL of Et
2
6
2
,4,6-tri-tert-butylphenoxyl, and 2,6-di-tert-butyl-4-methoxyphe-
Stopped-Flow Kinetics. In a typical procedure, solutions of
2
5
II
iAscH
MeCN were prepared and loaded into gastight syringes inside a
-filled glovebox. Mixtures containing additives, except for water,
2
+ DBU (0.68 mM each) and TEMPO (1.3-71 mM) in
2
noxyl were prepared according to the literature. (acac) Ru (py-
27a
imH) was prepared as previously described. Deionized water was
distilled before use in kinetic experiments. 18-Crown-6 was
recrystallized from acetonitrile and dried under vacuum. Choline
chloride was recrystallized from absolute ethanol and dried under
vacuum. Glyme was freshly distilled from sodium/benzophenone
ketyl. N-Acetyl-leucine-methyl ester was prepared from leucine-
N
2
were also prepared in the glovebox. The syringes were removed
from the glovebox individually for each stopped-flow run. When
necessary, freshly distilled and degassed water was quickly
-
2
transferred from a N -filled two-neck flask to both syringes (iAscH
57
and TEMPO) via a gastight microliter syringe. The stopped-flow
syringes were capped and inverted twice to mix thoroughly and
immediately placed onto the stopped-flow. A minimum of flow
kinetic runs were collected at each [TEMPO]. The data were
analyzed using SPECFIT software as described in the text.
methyl ester (Aldrich) by the method of Applewhite, recrystallized
from ethanol/pentane, and stored in a N -filled glovebox. Ethylene
glycol was distilled from MgSO , and then a small amount of Na
2
0
4
was added. The solvent was stirred overnight and finally was freshly
distilled before use. Glycerol was added to an equal volume of
n-butanol and rotated slowly in an ice bath until crystals formed.
The crystals were collected by vacuum filtration, washed with ice-
cold acetone, and dried under vacuum. Diethanolamine was distilled
and crystallized from its melt.
Acknowledgment. The U.S. National Institutes of Health
(GM50422) and the University of Washington Department of
Chemistry are gratefully acknowledged for financial support.
Preparation of Potassium 18-Crown-6 Ascorbate. Potassium
Supporting Information Available: Pseudo-first-order plots
for reaction 1 in DMSO and MeCN; equilibrium constant/mass
balance plots for reaction 1 in MeCN and DMSO; Benesi-
1
3
ascorbate was prepared according to the literature. A swivel-frit
assembly was charged with 0.5 g (2.3 mmol) of potassium ascorbate
and 0.6 g (2.3 mmol) of 18-crown-6. The system was evacuated,
-
Hildebrand plots for AscH · · ·additive formation constants;
and approximately 20 mL of CH
transferred from CaH . The resulting yellow solution was stirred
for 1 h and filtered. The solvent was removed under vacuum to
give a yellow solid, which was recrystallized from CH CN/Et
3
CN was freshly vacuum-
-
reactions of TEMPOH or iAscH with phenoxyl radicals;
2
II
reaction of TEMPO with (acac)
2
Ru (py-imH); reaction of
t
•
2
Bu NO with TEMPOH; calculations of BDFE for TEMPOH.
3
2
O
This material is available free of charge via the Internet at http://
pubs.acs.org.
(
57) Applewhite, T. H.; Waite, H.; Neimann, C. J. Am. Chem. Soc. 1958,
0, 1465–1469.
8
JA102337N
J. AM. CHEM. SOC. 9 VOL. 132, NO. 22, 2010 7793