Photoinduced electron transfer, decarboxylation, and radical fragmentation of cysteine derivatives: A chemically induced dynamic nuclear polarization study

Article abstract of DOI:10.1021/ja9536678

The photoreactions of cysteine derivatives I with 4-carboxybenzophenone in D₂O were investigated by measurements of chemically induced dynamic nuclear polarization (CIDNP). The quenching mechanism is electron transfer from sulfur at every pH; even if the amino group of I is deprotonated, electron transfer from nitrogen does not participate. Decarboxylation of I^?+ to give α-aminoalkyl radicals V^? occurs on the CIDNP time scale and causes strong cooperative effects. The decarboxylation rate is increased significantly by deprotonation of the amino function; this is due to product control. V^? decays by two competing pathways. Fragmentation of the C^β - S bond in V^? yields vinylamine, which is hydrolyzed to acetaldehyde at pH ? 7.25, and thiyl radicals, which then attack the sensitizer to give combination products. Oxidation of V^? by ground-state sensitizer leads to sulfur-containing aldehydes or other products, depending on pH. Relative rates of fragmentation and oxidation were determined from CIDNP signal intensities. From the temperature dependence of the polarizations, the activation energy of β-fragmentation was estimated to be 54 kJ mol^-1.

Full text of DOI:10.1021/ja9536678

2
882
J. Am. Chem. Soc. 1996, 118, 2882-2891
Photoinduced Electron Transfer, Decarboxylation, and Radical
Fragmentation of Cysteine Derivatives: A Chemically Induced
Dynamic Nuclear Polarization Study
,†
‡
‡
Martin Goez,* Jaroslaw Rozwadowski, and Bronislaw Marciniak
Contribution from the Institut f u¨ r Physikalische und Theoretische Chemie, Technische UniVersit a¨ t
Braunschweig, Hans-Sommer-Strasse 10, D-38106 Braunschweig, FRG, and Faculty of
Chemistry, Adam Mickiewicz UniVersity Poznan, Grunwaldzka 6, PL 60-780 Poznan, Poland
X
ReceiVed October 31, 1995
Abstract: The photoreactions of cysteine derivatives I with 4-carboxybenzophenone in D2O were investigated by
measurements of chemically induced dynamic nuclear polarization (CIDNP). The quenching mechanism is electron
transfer from sulfur at every pH; even if the amino group of I is deprotonated, electron transfer from nitrogen does
•
+
•
not participate. Decarboxylation of I to give R-aminoalkyl radicals V occurs on the CIDNP time scale and causes
strong cooperative effects. The decarboxylation rate is increased significantly by deprotonation of the amino function;
this is due to product control. V decays by two competing pathways. Fragmentation of the C -S bond in V yields
vinylamine, which is hydrolyzed to acetaldehyde at pH j 7.25, and thiyl radicals, which then attack the sensitizer
•
â
•
•
to give combination products. Oxidation of V by ground-state sensitizer leads to sulfur-containing aldehydes or
other products, depending on pH. Relative rates of fragmentation and oxidation were determined from CIDNP
signal intensities. From the temperature dependence of the polarizations, the activation energy of â-fragmentation
-
1
was estimated to be 54 kJ mol .
R
The photoreactions of sulfur-containing amino acids with
-carboxybenzophenone have received some attention because
tonation at C by the sensitizer anion to give an R-thioalkyl
1,2
4
radical, which amounts to a net hydrogen abstraction, also plays
some role. The mechanisms of these photoreactions are
therefore rather complex and involve several paramagnetic
intermediates.
One of the most versatile methods for the study of complex
radical reactions is the measurement of chemically induced
2
of the biological importance of these substrates and the model
2a
character of these reactions for the damage of cell components.
It has been shown that the primary photochemical process is
electron-transfer quenching of the excited triplet state of the
sensitizer by the amino acid.2 While these molecules contain
a,b
5
two possible donor sites, the thioether group and the amino
dynamic nuclear polarizations (CIDNP). CIDNP denotes the
+
occurrence of anomalous NMR line intensities (enhanced
absorption or emission) in the products of chemical reactions
carried out in a magnetic field. This effect is caused by nuclear
spin selective intersystem crossing in intermediate radical pairs.
The high diagnostic value of CIDNP experiments draws on five
sources: First, the diamagnetic reaction products are observed
by high-resolution NMR, so their identification is often straight-
forward. Further advantages in this respect are that species with
lifetimes exceeding a few tenths of a second are detectable with
pulsed spectrometers and that the signal enhancement by the
CIDNP effect mitigates the low sensitivity of NMR. Second,
the polarization patterns, i.e., relative polarization intensities and
phases of different protons, contain similar information as the
EPR spectra of the radicals involved and therefore allow
identification of the intermediates as well. Since the generation
of spin polarizations is completed after the correlated life of
the radical pairs, a few nanoseconds, CIDNP is sensitive to faster
processes than conventional EPR spectroscopy, and it possesses
an inherent clock. Third, the overall signal phases are related
to the multiplicities of the reacting species. This information
is often unaccessible by other methods. Fourth, geminate
function, the observation of dimeric radical cations 〉S∴ S〈 has
provided evidence in several cases that the electron is transferred
2a
from sulfur. One of the most important reactions of the radical
cations is elimination of CO2 to give R-aminoalkyl radicals,
which are key intermediates with respect to the secondary
1
,2a,c,3
chemistry in these systems.
R-Aminoalkyl radicals are
strongly reducing species,3
a,4
and their oxidation by surplus
4
-carboxybenzophenone, which finally leads to aldehydes,
1,2c
provides one of their major decay pathways.
Apart from
decarboxylation of the sulfur-centered radical cations, depro-
*
Author to whom correspondence should be addressed. Present
address: Fachbereich Chemie, Martin-Luther-Universit a¨ t Halle-Wittenberg,
Kurt-Mothes-Strasse 2, D-06120 Halle/Saale, FRG.
†
Technische Universit a¨ t Braunschweig.
Adam Mickiewicz University Poznan.
Abstract published in AdVance ACS Abstracts, March 1, 1996.
‡
X
(
1) Cohen, S. G.; Ojanpera, S. J. Am. Chem. Soc. 1975, 97, 5634-5635.
(2) (a) Bobrowski, K.; Marciniak, B.; Hug, G. L. J. Am. Chem. Soc.
1
992, 114, 10279-10288. (b) Marciniak, B.; Bobrowski, K.; Hug, G. L.
J. Phys. Chem. 1993, 97, 11937-11943. (c) Bobrowski, K.; Hug, G. L.;
Marciniak, B.; Kozubek, H. J. Phys. Chem. 1994, 98, 537-544.
(3) (a) Hiller, K.-O.; Masloch, B.; G o¨ bl, M.; Asmus, K.-D. J. Am. Chem.
Soc. 1981, 103, 2734-2743. (b) Asmus, K.-D.; G o¨ bl, M.; Hiller, K.-O.;
Mahling, S.; M o¨ nig, J. J. Chem. Soc., Perkin Trans. 2 1985, 641-646. (c)
M o¨ nig, J.; G o¨ bl, M.; Asmus, K.-D. J. Chem. Soc., Perkin Trans. 2 1985,
(5) (a) Lepley, A. R., Closs, G. L., Eds. Chemically Induced Magnetic
Polarization; Wiley: New York, 1973. (b) Richard, C.; Granger, P.
Chemically Induced Dynamic Nuclear and Electron Polarizations - CIDNP
and CIDEP; Springer: Berlin, 1974. (c) Kaptein, R. AdV. Free Rad. Chem.
1975, 5, 319-381. (d) Muus, L. T., Atkins, P. W., McLauchlan, K. A.,
Pedersen, J. B., Eds. Chemically Induced Magnetic Polarization; D.
Reidel: Dordrecht, 1977. (e) Salikhov, K. M.; Molin, Yu, N.; Sagdeev,
R. Z.; Buchachenko, A. L. Spin Polarization and Magnetic Effects in Radical
Reactions; Elsevier: Amsterdam, 1984.
6
47-651. (d) Steffen, L. K.; Glass, R. S.; Sabahi, M.; Wilson, G. S.;
Sch o¨ neich, Ch.; Mahling, S.; Asmus, K.-D. J. Am. Chem. Soc. 1991, 113,
2
141-2145. (e) Bobrowski, K.; Sch o¨ neich, Ch.; Holcman, J.; Asmus, K.-
D. J. Chem. Soc., Perkin Trans. 2 1991, 353-362. (f) Bobrowski, K.;
Sch o¨ neich, Ch.; Holcman, J.; Asmus, K.-D. J. Chem. Soc., Perkin Trans.
2
1991, 975-980. (g) Davies, M. J.; Gilbert, B. C.; Norman, R. O. C. J.
Chem. Soc., Perkin Trans. 2 1983, 731-738.
4) Hiller, K.-O.; Asmus, K.-D. J. Phys. Chem. 1983, 87, 3682-3688.
(
0
002-7863/96/1518-2882$12.00/0 © 1996 American Chemical Society

CIDNP Study of Cysteine DeriVatiVes
J. Am. Chem. Soc., Vol. 118, No. 12, 1996 2883
Table 1. Formulas, Abbreviations, and pKa2 of the Cysteine
Derivatives Used
R
abbreviation
pKa2a
pKa2b
CH
CH
3
-
Ia
Ib
Ic
Id
Ie
9.16 ( 0.08
9.11 ( 0.04
8.80 ( 0.10
9.39 ( 0.06
9.28 ( 0.06
8.80
8.65
c
8.89
9.08
3
-CH
2
-
(
CH
OOC-CH
OOC-CH
)
3 3
C-
-
-
2
-
-CH
2
2
-
a
O. b In H
This work, in D
2
2
O (Smith, R. M.; Martell, A. E. Critical
Stability Constants; Plenum Press: New York, 1974, 1989; Vol. 1, 6).
c
No literature data available.
processes of radical pairs and reactions of free radicals can be
separated by their time dependence. Lastly, the polarizations
can be employed as labels to trace the pathways from the
intermediates to the products because the polarizations are
created at an earlier stage than they are detected.
Despite these advantages, so far there seems to have been
but a single CIDNP investigation of sensitized photoreactions
of chemically unmodified sulfur-containing amino acids in
6
which any polarizations could be detected, and in that study
(methionine sensitized by flavin) only extremely weak CIDNP
signals of the starting material were observed and none of any
reaction product. Methionine residues in a protein were also
found to give rise to such polarizations.7
In this work, we use CIDNP experiments to study the
photoreactions of five derivatives of cysteine with 4-carboxy-
benzophenone in aqueous solution. As we will show, electron
transfer occurs from sulfur only, even in basic medium where
the amino function is deprotonated. However, deprotonation
of this group influences the overall reaction indirectly, by
increasing the decarboxylation rate through stabilization of the
resulting radical. Our CIDNP results further reveal that
â-fragmentation of the R-aminoalkyl radicals competes with
oxidation by surplus sensitizer and provides a major decay
pathway of these intermediates. The polarization intensities
allow us to determine the rates of fragmentation and oxidation.
Figure 1. Top trace, background-free pseudo-steady-state CIDNP
spectrum observed in the photoreaction of 4-carboxybenzophenone (CB)
with S-(carboxyethyl)cysteine (Ie) in D O (pH 6.39) at room temper-
2
ature. Bottom trace, NMR spectrum of the same system recorded in
the dark. Only the spectral ranges of interest are shown. Experimental
-3
-2
parameters: [CB] ) 8 × 10 M, [Ie] ) 2 × 10 M. The assignment
of the resonances of the products Ie, II, IIIe, and IVe refers to the
structural formulas given at the top. The numbering of the protons
has been chosen to emphasize their correspondence in the different
products. CB-o denotes the signal of the ortho and ortho′ protons of
the sensitizer CB.
spectrum of Ie at pH 6.39, which is given in Figure 1, provides
a typical example of the polarizations found below pH 7. Owing
to the experimental technique employed (pseudo-steady-state
Results and Discussion
Substrates and Protonation Equilibria. The amino acids
employed, their abbreviations used in this work, and the pKa2
values of their cysteine groups are given in Table 1. pKa2 values
were determined from the turning points in plots of the chemical
9
measurements), this spectrum, as well as all other spectra of
this work, shows pure polarizations without background signals.
Line intensities are undisturbed by nuclear spin relaxation in
the diamagnetic reaction products including cross-relaxation.
Apart from resonances of the starting material, the CIDNP
spectrum of Figure 1 exhibits several polarized signals that are
due to three new products, II, IIIe, and IVe. The structural
formulas of these compounds are displayed in the figure. The
connectivity of their spin systems was determined by combining
CIDNP and double-resonance experiments. The same aldehyde,
II, is formed with all five amino acids. Addition of acetaldehyde
to the samples showed the chemical shifts of this compound
and those of II to be identical; however, the splitting pattern
1
shift of H vs pH. Since D2O was used as the solvent, the pKa2
values differ slightly from the values in H2O, by 0.2-0.5 unit.
In almost all experiments of this work, pH was larger than
6
.0, so the sensitizer 4-carboxybenzophenone was present in
8
its anionic form; the cysteine fragments of the amino acids
were present in their zwitterionic forms at pH below pKa2 and
in their anionic forms at higher pH. To simplify the nomen-
clature, we will omit the charges of the carboxy and amino
functions. Thus, for instance, we denote the anion of 4-carb-
oxybenzophenone by CB and the radical dianion of this
•
-
compound by CB .
revealed that II is in fact CDH CHO. IIIe and the correspond-
2
CIDNP Spectra below pH 7. Strong nuclear spin polariza-
tions are observed in these systems upon irradiation, indicating
the occurrence of radical pairs in the photoreactions. The
appearance of the CIDNP spectra strongly depends on pH. The
ing products in the other systems are stable and were identified
to be combination products of the sensitizer with RS-, where
R is the respective substituent at sulfur in the cysteine. IVe
and the analogous compounds with the other amino acids were
very unstable and could not be detected in the reaction mixture
directly after irradiation. Our characterization of these products
as aldehydes R-S-CH2-CHO which are known to be formed
(
(
6) Stob, S.; Kaptein, R. Photochem. Photobiol. 1989, 49, 565-577.
7) Stob, S.; Scheek, R. M.; Boelens, R.; Kaptein, R. FEBS Lett. 1988,
2
39, 99-104.
8) Inbar, S.; Linschitz, H.; Cohen, S. G. J. Am. Chem. Soc. 1981, 103,
323-7328.
(
7
(9) Goez, M. Chem. Phys. Lett. 1992, 188, 451-456.

2
884 J. Am. Chem. Soc., Vol. 118, No. 12, 1996
Goez et al.
(g ) 2.0037);14 the hyperfine coupling constants of R-protons
Table 2. Chemical Shifts, Multiplet Patterns, and Polarization
Phases of the Polarized Protons in the Products III and IV
a
11b
in a sulfur-centered radical cation are positive.
The absorption
2
3
R
III-3
IV-1
IV-2
IV-3
signal observed for H and H of Ie is thus consistent with
regeneration of the reactants by spin-allowed back electron-
CH
CH
3
-
1.88 (s, E) 9.28 (m, A) 3.24 (d, E) 2.29 (s, E)
2.44 (q, E) 9.37 (m, A) 3.33 (d, E) 2.65 (q, E)
•+
•-
3
-CH
2
-
transfer of singlet pairs I CB .
(
CH
OOC-CH
OOC-CH
3
)
3
C-
b
9.35 (m, A) 3.41 (d, E) b
-
-
The polarization pattern of the reaction products II and IVe
is obviously different from that of the starting amino acid,
showing opposite phases for vicinal protons. This pattern is
compatible with CIDNP generation, at least to some degree, in
alkyl radicals, where σ-π spin polarization serves to induce a
2
-
-CH
3.04 (s, E) 9.34 (m, A) 3.30 (d, E) 3.22 (s, E)
2
2
- 2.57 (t, E) 9.33 (m, A) 3.30 (d, E) 2.78 (t, E)
a
For the general structures of III and IV and the numbering of the
b
protons, see Figure 1. Not applicable.
R
in these reactions,1,2c is based on the following evidence. From
negative spin density at H and hyperconjugation a positive spin
â 15
1,2c
density at H . From the known pathway of formation of
chemical shifts and multiplet patterns, two product moieties are
identified. One is an -S-CH2-CHO fragment that is inde-
pendent of the amino acid employed, the other is an RS-
fragment like the one contained in the combination product III.
That these two moieties belong to the same product is
established by our observation that the relative polarizations of
the protons in these two fragments are constant regardless of
the experimental parameters, whereas, for instance, the ratio of
these polarizations to the polarizations of II or III strongly
depends on the sensitizer concentration (see below).
aldehydes IV in these systems, via decarboxylation of the sulfur-
•+
centered radical cations I to give R-aminoalkyl radicals R-S-
•
CH2- C˙ HNH2 (V ) which are then oxidized by surplus CB,
•
radical pairs containing V appear as the most natural explana-
tion for the observed polarization pattern. This mechanism has
three implications. First, the other radical contained in the pairs
•-
must still be the radical anion CB of 4-carboxybenzophenone
•+
because this is left unchanged by the decarboxylation of I .
Second, the aldehyde proton of IV must have been attached to
•
the radical center in V ; hence, for this proton a is negative.
The NMR parameters of compounds III and IV are listed in
Table 2. Lowering pH led to no changes in the relative
intensities of the CIDNP signals. Experiments below pH ≈
Third, the aldehydes IV are products of free radicals having
escaped from the cage, so ꢀ ) -1. Thus it would follow from
•
•-
.5, the pKa value of the sensitizer,10 were impracticable because
eq 1 that the g value of V is larger than that of CB . This is
4
contrary to expectation based on the data for similar radicals
the protonated form of CB is hardly soluble in water.
Identification of the Paramagnetic Intermediates. For the
(
for instance, the g value of CH3 C˙ HNEt2 is noticeably smaller
16
2
3
than that of the benzophenone radical anion). An explanation
of this apparent discrepancy will be given below.
regenerated starting compound Ie, protons H and H , which
are connected to the carbon atoms adjacent to sulfur, are seen
to be polarized with equal phase and intensity whereas the more
remote protons H and H are unpolarized (Figure 1). This
polarization pattern is in accordance with the spin density
The magnitudes of the hyperfine coupling constants of R-
and â-protons in R-aminoalkyl radicals are comparable, whereas
more remote protons only possess extremely weak hyperfine
1
4
1
7
1
•
+
coupling constants. In contrast, the aldehyde proton H of
distribution in a sulfur-centered radical cation I (compare the
11
IVe is much more weakly polarized than the aliphatic protons
data for structurally similar radical cations). The basis of this
reasoning is provided by the known fact that sign and magnitude
of the polarization Pi of proton i in the products reflect sign
and magnitude of the hyperfine coupling constant ai of this
proton in the paramagnetic intermediates; often, there is even
direct proportionality between Pi and ai.5
2
3
H , and the polarizations of H are found to be as strong as
2
those of H . The polarization pattern observed therefore cannot
•
•-
be due solely to CIDNP generated in pairs V CB . Rather
there must be a significant contribution of the primary pairs
•+
•-
I CB , in which polarizations of equal sign and magnitude
1
2
According to Kaptein’s rule for CIDNP net effects,
2
3
are generated for H and H ; the emissive phase of these
polarizations is in line with aldehyde formation by an escape
reaction.
Γ ) sgna × sgn∆g × µ × ꢀ
(1)
i
i
Radical pair type CIDNP means spin sorting, so polarizations
of escape products must be accompanied by opposite polariza-
tions of cage products. It seems strange that no resonances can
the polarization phase (Γi ) +1, absorption; Γi ) -1, emission)
of proton i is determined by magnetic parameters of the
intermediate radical pairs (sgnai, sign of the hyperfine coupling
constant of proton i; sgn∆g, sign of the difference of the g values
of the two radicals of the pair, where the radical counted first
contains proton i) and by the entry and exit channels of the
pairs (µ ) +1, pair formation from triplet precursors; µ ) -1,
pair formation from singlet precursors; ꢀ ) +1, product
formation from singlet radical pairs, which in our case means
in a cage reaction; ꢀ ) -1, product formation from triplet pairs,
i.e., via radicals escaping from the cage). 4-Carboxybenzophen-
one reacts from its electronic triplet state, so µ is +1 in all our
experiments. Sulfur-centered radical ions are known to possess
•
•-
be assigned to cage products of the radical pairs V CB (III is
also an escape product; see next section). Two factors may
provide an explanation for this anomaly. First, geminate
recombination of V CB will occur at C1 of V , so one
•
•-
•
2
expects merely a small influence on the chemical shift of H
3
and practically none on that of H ; hence, the most strongly
polarized signals of the cage products may be obscured by the
educt resonances. Second, this recombination must yield a
carbanion; owing to the reactivity of this intermediate, the
polarizations might be distributed among several products and
would thus be correspondingly weak.
very high g values (g > 2.01),1
1b,13
whereas the g value of CB
cannot differ much from that of the benzophenone radical anion
•-
As Figure 1 shows, the aromatic region of the CIDNP
spectrum is dominated by a strong absorption signal, CB-o, for
(
10) Hurley, J. K.; Linschitz, H.; Treinin, A. J. Phys. Chem. 1988, 92,
151-5159.
11) (a) Eastland, G. W.; Rao, D. N. R.; Symons, M. C. R. J. Chem.
5
(
(14) Aarons, J. L.; Adam, F. C. Can. J. Chem. 1972, 50, 1390-1400.
(15) Carrington, A.; McLachlan, A. D. Introduction to Magnetic
Resonance; Harper & Row: New York, 1969; pp 80-85.
(16) Goez, M.; Sartorius, I. J. Am. Chem. Soc. 1993, 115, 11113-11123.
(17) McLauchlan, K. A.; Ritchie, A. J. J. Chem. Soc., Perkin Trans. 2
1984, 275-279.
Soc., Perkin Trans. 2 1984, 1551-1557. (b) Rao, D. N. R.; Symons, M.
C. R.; Wien, B. W. J. Chem. Soc., Perkin Trans. 2 1984, 1681-1687.
(
(
12) Kaptein, R. J. Chem. Soc., Chem. Commun. 1971, 732-733.
13) Petersen, R. L.; Nelson, D. J.; Symons, M. C. R. J. Chem. Soc.,
Perkin Trans. 2 1978, 225-231.

CIDNP Study of Cysteine DeriVatiVes
J. Am. Chem. Soc., Vol. 118, No. 12, 1996 2885
the ortho and ortho′ protons of the sensitizer; the other protons
are almost unpolarized. There are also some weaker new
signals, which are most probably due to combination products
but which we did not assign. The polarization pattern found
for regenerated CB is in accordance with the intermediacy of
hydrogen abstraction from the solvent; owing to the small radical
concentrations in our experiments, disulfide formation would
not be expected to be significant. For the same reason, and
because it is an escape product, III must stem from attack of
•
•-
R-S to ground-state sensitizer, not to CB . As shown in eq
•
-
â
•
radical anions CB because ortho and ortho′ are the only
positions in these radicals for which one expects substantial
proton hyperfine couplings. With the known sign of ao,o′
2, the olefin resulting from C -S cleavage of V is an enamine
(vinylamine), so hydrolysis of this compound is an obvious
pathway to acetaldehyde. The exclusive incorporation of a
single deuterium atom in the â-position of acetaldehyde supports
15
(
minus), and the known multiplicitites of precursors and pairs
•
undergoing back electron-transfer (µ ) +1, ꢀ ) +1), the
the intermediacy of an enamine and thus cleavage of V
o,o′
•-
20
absorptive polarization of H indicates that the g value of CB
according to eq 2 because it is well known that the first step
of enamine hydrolysis is protonation at C , whereas this position
â
is smaller than the g value of the sulfur-containing radical, which
is consistent with the above results obtained from the polariza-
tion patterns of the starting compound Ie and the cysteine-
derived product moieties II-IVe.
is not involved in the later stages of this reaction. As will be
shown below, in experiments at higher pH the enamine is also
observed directly in the CIDNP spectra.
The same behavior as for Ie, with respect to both educt and
product polarizations, was also observed for the other amino
acids.
Rates of Radical Fragmentation. Cleavage and oxidation
by ground-state CB should be competing reactions of the
radicals V
•
Mechanism of Radical Fragmentation. The formation of
acetaldehyde II on the one hand and of a combination product
IIIe of the sensitizer with the cysteine substituent R on the other,
which are both spin-polarized, implies fragmentation at sulfur
of one of the intermediates. Given the structural formulas of
II and IIIe,18 it is natural to assume that the two products result
from the same C-S cleavage step. That this is indeed the case
is borne out by the observation that for all our systems the ratio
of polarizations of protons II-2 and III-3 does not depend on
experimental conditions such as sensitizer concentration or
temperature, whereas, for instance, the ratio of polarizations of
protons II-2 and IV-2 does (see below).
As Figure 1 shows, the polarization pattern of protons II-1,
II-2, and IIIe-3 is identical with that of protons IVe-1, IVe-2,
and IVe-3 (medium absorption, strong emission, strong emis-
sion, relative intensities being about 1:-8:-10 in both cases).
Hence, the polarizations in II and IIIe must have the same origin
as those of IVe, that is, they must stem from a superposition of
(
In eq 3, the primary products resulting from these two processes
have been omitted for clarity.) It is therefore to be expected
that relative rates of these reactions can be obtained from relative
product yields. Since the aldehydes IV are rather unstable, such
an analysis of the product distribution is not feasible with the
usual methods.
However, the same information is accessible from the CIDNP
spectra because polarization intensities are proportional to the
amount of the respective product formed. The validity of this
approach is based on the condition that the constant of
proportionality is identical for the signals compared. This
requirement is met under the following circumstances.
First, the polarizations considered must stem from the same
source. This does not imply that the reaction may only proceed
via a single radical pair; with more complex reaction mecha-
nisms, as in our case, this is still fulfilled if each type of radical
pairs makes the same relative contribution to these polarizations.
The latter has already been shown in the preceding section;
moreover, it seems to be ensured for our systems because
pathways from the sulfur-centered radical cations I to one of
the products II, III, or IV that bypass the R-aminoalkyl
radicals V are hardly conceivable from a chemical point of
view. As long as this condition holds, the polarizations present
in V are simply partitioned between the products, and every-
thing that influences concentrations and polarizations at a stage
of the reaction up to and involving V , in particular such factors
as the different efficiencies of CIDNP generation in the pairs
•+
•-
CIDNP arising in the primary pairs I CB and, to a smaller
•
•-
degree, of CIDNP arising in pairs V CB ; furthermore, II and
III are escape products, as is IV. The same result was found
for the other cysteine derivatives; there, however, the amounts
•
+
•
•-
of polarizations from V CB are somewhat larger than in the
system Ie/CB.
•
The fact that the relative contribution of polarizations from
•
•
•-
pairs V CB to the total polarizations is the same for the
fragmentation products and the products of oxidation of escaping
•
•
V by ground-state sensitizer shows that scission of the C-S
•
+
bond does not occur in the sulfur-centered radical cations I
•
•+
•-
•
•-
but in the R-aminoalkyl radicals V . â-Thioalkyl radicals are
indeed known to undergo cleavage of the bond between C and
I CB and V CB , affects the polarizations of all products
in exactly the same degree. As a further consequence of the
requirement that the polarizations evaluated have the same
origin, only protons that were identical in the radicals should
be compared, as, for instance, II-2 and IV-2 or III-3 and IV-3.
Second, errors due to nuclear spin relaxation must be avoided.
With respect to the diamagnetic products, this poses no problems
because of the experimental technique used (pseudo-steady-state
â
sulfur to give an olefin and a thiyl radical.19 This reaction,
9
CIDNP measurements), but in our case the products of interest
provides a straightforward rationalization both of the products
formed in our systems and of their polarizations. Thermody-
are formed via free radicals, and nuclear spin relaxation in these
rather long-lived paramagnetic intermediates may not always
be neglected. However, only differences of the relaxation rates
count, so relaxation in V is circumvented by considering protons
that were identical in V , as already mentioned; the same holds
•
namic considerations indicate that the thiyl radicals R-S should
terminate preferentially by reaction with the aromatic rings of
•
the sensitizer, which eventually leads to III, rather than by
•
•
+
(18) As indicated in Figure 1, the structure of the combination products
for relaxation in free radicals I . Even so, on the route to III
III is not yet known completely. However, this does not invalidate any of
our mechanistic conclusions that follow.
(20) March, J. AdVanced Organic Chemistry, 2nd ed.; McGraw-Hill:
Tokyo, 1977; p 807.
(19) Huyser, E. S.; Kellogg, R. M. J. Org. Chem. 1966, 31, 3366-3369.

2
886 J. Am. Chem. Soc., Vol. 118, No. 12, 1996
Goez et al.
Table 3. Rate Constants of the Secondary Reactions of the
•
Cysteine-Derived R-Aminoalkyl Radicals V at Room Temperature
(R is the substituent at sulfur)
a
ox/kfragb
k
ox
k
k
frag
7
9
-1
(M-1)
(10 s-1)
R
(10 M s-1
)
CH
CH
3
-
3
1.27
1.08
73
84
78
68
79
1.74
1.28
-CH
2
-
(
CH
OOC-CH
OOC-CH
3
)
3
C-
-
-
c
2
-
-CH
0.78
0.96
1.14
1.21
2
2
-
a
From ref 2a; estimated error e 20%. b This work; (10%. Error
c
(
30% owing to low signal-to-noise ratio of the signals II-2 and IVd-2
caused by predominant reaction at the substituent R.
Figure 2. Left, dependence of the polarization intensities of the
â-protons in the oxidation product IVe and the fragmentation product
II (upper trace, CIDNP signal of IVe-2; lower trace, that of II-2) on
-
3
the sensitizer concentration, [CB] (given between the traces; in 10
M). Right, ratio P(IVe-2)/P(II-2) of these polarizations as a function
of [CB]. Polarizations P were determined by integration over the
-
2
multiplets. Experimental parameters: [Ie] ) 1 × 10 M, pH ) 6.8,
room temperature.
an additional paramagnetic species, the thiyl radical, is successor
•
•
to V , whereas after the stage of V formation of both II and
IV involves diamagnetic species only. Hence, comparison of
the signals II-2 and IV-2 eliminates all effects of nuclear spin
relaxation in the radicals, whereas comparison of the signals
III-3 and IV-3 would be less reliable.
Figure 3. Plot according to eq 4 for compound Ie. For further
explanation, see text.
Third, none of the products may decompose significantly
during the time needed for acquisition of its free induction decay.
Even for IV, which is unstable, this appears to be fulfilled in
our case because no manifestations (line broadening or disper-
observed when the photoreaction of Ia was sensitized with 2
-4
22
×
10 M N-(9-methylpurin-6-yl)pyridinium cation or when
21
sive signal components) of such a fast secondary reaction could
be observed.
pulse radiolysis experiments with Ia were performed in the
-5
4
presence of 2.5 × 10 M methylviologen.
To test whether eq 3 holds, e.g., whether the products II and
III on the one hand and IV on the other are formed via a simple
competition between fragmentation and oxidation of the R-ami-
Activation Energy of Radical Fragmentation. Assuming
Arrhenius behavior, plots of ln kox/kfrag vs 1/T should be linear,
with slopes that are determined by the differences of the
•
noalkyl radicals V , we varied the concentration of CB. An
activation energies Ea of fragmentation and oxidation of V•
increase of [CB] should favor oxidation, hence formation of
the sulfur-containing aldehyde IV, over fragmentation, which
is formation of monodeuterated acetaldehyde II and combination
product III. From the spectra at the left in Figure 2, which
display the resonances of IVe-2 and II-2 as functions of the
sensitizer concentration, this is seen to be qualitatively true.
According to eq 3, the ratio of polarization intensities, P(IV-
and intercepts that reflect the ratios of the preexponential factors,
A, of both processes:
kox
Aox Ea,frag - Ea,ox
1
T
ln
) ln
+
(4)
kfrag
Afrag
R
In Figure 3, such a plot is shown for the derivative Ie. From
2
)/P(II-2), of these two signals should be equal to kox[CB]/
-
1
this graph, one obtains a value of 38.5 ( 0.7 kJ mol for the
kfrag, so a plot of this polarization ratio vs [CB] should be linear,
with vanishing intercept. This plot is given at the right in Figure
-
5
difference Ea,frag - Ea,ox and of about 1 × 10 M for the ratio
•
Aox/Afrag. The activation energies for oxidation of V by CB
2. As is evident from it, the above assumption is also
are not known. However, calculation of the rate constant kdiff
of a diffusion-controlled reaction by the Smoluchowski equation:
quantitatively true; furthermore, it shows the validity of the
described evaluation procedure of relative CIDNP intensities.
The ratios kox/kfrag obtained for our cysteine derivatives are
8
3
RT
η
•
kdiff
)
γ
(5)
listed in Table 3. The rate constants kox for oxidation of V
by ground-state CB are known2 from pulse radiolysis experi-
ments in which the rates of secondary formation of CB were
measured as functions of [CB]; they have also been compiled
in the table. With these data, kfrag can be computed. As Table
a
•
-
modified by the Debye factor γ to take into account the
Coulombic interactions between ions of charge z1 and z2,
2
3
shows, the values of kfrag are very similar for the amino acids
z z e
1
2
1
employed, as are those of kox. Hence, the substituent R has no
γ )
(6)
2
4
πꢀ ꢀ dkT
•
0 r
exp[z z e /(4πꢀ ꢀ dkT)] - 1
significant influence on the fragmentation of V nor on the
1 2
0 r
oxidation of this radical by CB.
Even in the absence of an electron acceptor such as CB, the
lifetimes 1/kfrag of the radicals V are seen to be about 70 ns
yields a value of about 3 × 10 M s-1 for a reaction in water
between two ions of charge -1 with an assumed encounter
distance d of 4-5 Å. This is not much larger than the
experimental value of kox in the system Ie/CB (compare Table
9
-1
•
only. This short life can fully rationalize why no oxidation of
•
the R-aminoalkyl radical Va by ground-state sensitizer was
3
); moreover, the Smoluchowski equation usually overestimates
(
21) Ernst, R. R.; Bodenhausen, G.; Wokaun, A. Principles of Nuclear
Magnetic Resonance in One and Two Dimensions; Clarendon Press: Oxford,
987; pp 211-215.
(22) Marciniak, B.; Hug, G. L.; Rozwadowski, J.; Bobrowski, K. J. Am.
1
Chem. Soc. 1995, 117, 127-134.

CIDNP Study of Cysteine DeriVatiVes
J. Am. Chem. Soc., Vol. 118, No. 12, 1996 2887
R-aminoalkyl radicals produced in the sensitized photoreactions
of tertiary aliphatic amines, although the mechanism is quite
different in those instances, hydrogen abstraction being followed
by reaction of the R-aminoalkyl radicals with surplus sensi-
1
6,25
tizer.
Raising pH leaves unchanged the chemical shifts of
VI. This shows that the amino group is not protonated for pH
>
7.25, so the pKa value of this group must be lower than about
7
.
As can be seen in Figure 4, the polarization pattern of VI at
1
2,2′
pH ≈ 7.6 (absorption for H , emission for H ) is qualitatively
and quantitatively identical with that of II at lower pH. This
is consistent with formation of the aldehyde II by acid-catalyzed
hydrolysis of VI. Moreover, the observation of VI provides
direct evidence for the â-fragmentation of the radicals V•
according to eq 2.
At the high magnetic field employed in our experiments, the
polarizations of the protons in a particular product should be
approximately proportional to the hyperfine coupling constants
of these protons in the radicals weighted with their number.5e
The ratio of hyperfine coupling constants of the R- and
1
2
•
â-protonssi.e., those corresponding to H and H in V sin the
1
7
model compound CH3CH2 C˙ HN(n-Pr)2 is -1:1.36, so for VI,
one would expect a polarization ratio P(H ):P(H ) of about
1
2,2′
-
1:2.7. The experimental ratio is larger than this (about -1:4
in Figure 4, center and bottom traces), thus again indicating
that the polarizations of VI and its secondary product II partly
•+
•-
arise in pairs I CB as well, which yield emissive polariza-
2
,2′
1
tions for H but none for H . With Ib these polarizations are
•
•-
seen to stem predominantly from the pairs V CB ; this is also
found for the other cysteine derivatives except Ie, for which
•+
•-
polarizations from I CB have already been shown to prevail
1
2,2′
(
P(H ):P(H ) ≈ -1:8; compare Figure 1). Common to all
Figure 4. CIDNP spectra of the system 4-carboxybenzophenone/S-
systems investigated is the unexpected sign of ∆g of the pairs
(ethyl)cysteine (CB/Ib) at different pH (bottom trace, pH ) 6.29; center
•
•-
trace, pH ) 7.63; top trace, pH ) 11.43). For the assignment of the
resonances of II and VI, see the formulas in Figure 1 and this figure,
respectively. Signals IV′ and VII are explained in the text. Experi-
V CB under these experimental conditions that is obtained
from the polarization phases with eq 1. Both this result and
the behavior of Ie that is different from the other substrates
with respect to the relative amounts of polarizations from the
two radical pairs are related; these effects will be discussed
below.
-
3
-2
mental parameters: [CB] ) 2 × 10 M, [Ib] ) 2 × 10 M, room
temperature.
•
kdiff slightly. Hence, we assume that the oxidation of V by
CB is practically diffusion controlled in this case and ap-
At high pH (top trace of Figure 4), the reaction still leads to
the vinylamine VI, but under these circumstances VI displays
the opposite polarization phases as at medium pH, namely,
emission for H and absorption for H . This change of phase
with pH is paralleled by an inversion of the CIDNP signal CB-o
due to the ortho and ortho′ protons of the sensitizer. Both these
observations can be rationalized with predominant generation
proximate Ea,ox by the activation energy of the viscosity η of
-1 24
water, which is 15.6 kJ mol . Thus, we arrive at an activation
1
2
energy of 54 kJ mol- for the fragmentation of V according to
eq 2, which may be uncertain by a few kilojoules per mole
owing to the approximation involved.
1
•
Comparable values of Ea,frag - Ea,ox were also obtained for
the other cysteine derivatives.
CIDNP Experiments in Basic Solution. Interesting changes
of the product signals and of their polarization patterns occur
when pH is raised. Figure 4 shows CIDNP spectra of Ib as
examples.
•
•-
at high pH of the polarizations of VI in the pairs V CB , the
polarization phases now indicating a “normal” g-value differ-
ence. On the basis of the hyperfine coupling constants, one
2
,2′
would expect the absorption signals of H in VI to be larger
by a factor of about 3 than the experimental intensities. This
shows that at high pH there is still an emissive contribution
Above pH ≈ 7.25, the signals from acetaldehyde II vanish,
•+
•-
and a characteristic ABX pattern appears instead. From
from the pairs I CB to the polarizations of these protons.
In principle, the dependence of the polarization patterns of
VI on pH might be caused by a change of the exit channel
leading to this product. At pH J 7, no signals from the aldehyde
IV are detectable (see below), so determination of ꢀ for VI by
comparison of its polarization phases with those of IV, which
is by necessity an escape product, is not possible any longer.
However, ꢀ can also be obtained from the phases of CIDNP
1
2
chemical shifts and coupling constants (δ(H ) ) 6.10, δ(H ) )
2′
2
3
3
3
.80, δ(H ) ) 4.10; J22′ ≈ 0 Hz, J12 ) 9.6 Hz, J12′ ) 15.5
Hz), the product giving rise to these signals was identified as
vinylamine (VI), CH2dCHNH2; the spectral parameters listed
are quite similar to those1 of substituted vinylamines. There
is ample precedent for formation of the latter compounds from
6,25
4
(
23) As the pKa value of R-aminoalkyl radicals is about 3.9, the amino
•
group of V is deprotonated under our experimental conditions.
(
24) Calculated from the values of η in Weast, R. C., Ed. Handbook of
(25) (a) Roth, H. D.; Manion, M. L. J. Am. Chem. Soc. 1975, 97, 6886-
6888. (b) Roth, H. D. In ref 5d, pp 53-61. (c) Goez, M.; Frisch, I. J.
Photochem. Photobiol. A 1994, 84, 1-12.
Chemistry and Physics, Student Edition; CRC Press: Boca Raton, FL, 1988;
p F-19.

2
888 J. Am. Chem. Soc., Vol. 118, No. 12, 1996
Goez et al.
Figure 5. CIDNP multiplet effects in the system 4-carboxybenzo-
phenone/S-(methyl)cysteine (CB/Ia) at pH 10.3. Other experimental
parameters were as in Figure 4. Shown are the resonances VI-1 (left)
and VI-2′, VII-1, and VI-2 (right). The structural formula of VI and
the assignments are given in Figure 4. The spectrum was obtained by
subtracting a set of free induction decays acquired with a flip angle æ,
2
6
of 135° from a set acquired with æ ) 45°. For an explanation on
why the two inner lines of VI-1 are missing, see ref 16, note 40.
multiplet effects, even if the sign of ∆g is unknown. As |∆g|
•
+
•-
-3
for I CB is estimated to be larger than 9 × 10 , whereas
•
•-
-4
the value for V CB should be about 7 × 10 , multiplet
5
effects from the former pairs should be vanishingly small, and
no ambiguity is expected to arise from the participation of two
Figure 6. CIDNP spectra of the system 4-carboxybenzophenone/S-
(carboxymethyl)cysteine (CB/Id) at different pH (bottom trace, pH )
radical pairs in the reaction. Kaptein’s rule12 for a multiplet
6
.33; top trace, pH ) 11.11). Only the spectral regions of interest are
effect of protons i and j:
shown. The signals VIII-XI are explained in the text. For the
assignment of the other CIDNP signals, see preceding figures.
Γ ) µ × ꢀ × sgna × sgna × sgnJ × σ
(7)
ij
i
j
ij
ij
-
3
-2
Experimental parameters: [CB] ) 2 × 10 M, [Id] ) 2 × 10 M,
room temperature.
connects the multiplet phase (Γij ) +1, E/A, i.e., emission at
the low-field edge of the multiplet, absorption to high field; Γij
we tentatively assign these signals to the imine (R-S-CH2-
)
-1, A/E) with the sign of the hyperfine coupling constants
•
CHdNH) formed by oxidation of V by ground-state CB. When
of both nuclei and the sign of the nuclear spin-spin coupling
constant Jij in the product. The parameter σij takes into account
whether both nuclei were contained in the same radical of the
intermediate pairs (σij ) +1), or in different radicals (σij ) -1);
µ and ꢀ have the same meaning as in eq 1.
pH is raised further, IV′-2 decreases but does not change phase;
for pH g pKa2, IV′-3 is buried under the educt signal I-2, which
is shifted toward higher field by the deprotonation of the amino
group.
26
The position and phase of the signals III-3 stemming from a
combination product between CB and the thiyl radicals resulting
Net and multiplet effects were separated by their dependence
on the flip angle æ of the observation pulse. Two sets of spectra
were acquired with æ being 45° and 135°, respectively.
Subtracting the second set from the first yields pure multiplet
effects, which are shown in Figure 5 for the compound Ia. The
polarization pattern is seen to be an E/A multiplet. The signs
•
from the fragmentation of V are unaffected by pH. However,
the intensity of these peaks decreases with increasing pH. The
same is observed for the polarizations of the starting amino
2
acid: Their polarization pattern (enhanced absorption for H
3
•
3
and H , no polarizations for the other protons) is independent
of the hyperfine coupling constants in V are known, as is µ; J
is positive.27 In the present case, σij must be +1; furthermore,
of pH, but the signal intensities relative to those of the products
are smaller at high pH.
this variable can be eliminated by a combined evaluation of
net and multiplet effects,28 which leads to the same result,
namely, that VI is an escape product also at high pH.
The aldehydes IV behave in a manner similar to II: Above
pH ≈ 7, the signal IV-2 vanishes and is replaced by another
signal, IV′-2. The latter is a doublet (J ) 5.2 Hz) that is high-
field shifted by about 0.1 ppm; it displays the same phase
Lastly, we mention for completeness that at pH g pKa2
a
further ABX pattern appears (VII-1, δ ) 3.96 ppm, absorption;
VII-2,2′, δ ) 2.30 ppm, emission), which has not been identified
yet. Its spectral parameters are identical in all systems
investigated, so it must be attributed to a reaction of the cysteine
moiety. From the phase of the multiplet effect (A/E, compare
Figure 5), VII is seen to be a cage product. The fact that these
signals are not observed below pH ) pKa2, nor other signals
that can be related to them, leads us to infer that this product is
due to a reaction of the deprotonated amino function.
CIDNP Spectra of S-(Carboxymethyl)cysteine (Id). The
quencher Id shows a slightly more complicated behavior than
the other four amino acids because its photoreactions also take
place at the carboxymethyl substituent, not only at the cysteine
function. This manifests itself by the occurrence (compare
Figure 6, bottom trace) of four new CIDNP signals (VIII-3,
IX-3, X-3, and XI-3) that are also found in the photoreactions
of CB with compounds possessing an HOOC-CH2-S- group
but no cysteine moiety, as, for example, thiodiglycolic acid,
S(CH2COOH)2. If the CIDNP spectrum of Id is shifted by
(emission) and the same relative intensity with respect to the
educt signals as before. The signal that was due to IV-3 remains
almost at the same position. The fate of the signal IV-1 cannot
be determined owing to its extremely low intensity. The peaks
IV′-2 and IV′-3 replacing the aldehyde signals at pH J 7 exhibit
the same dependence on the concentration of CB as do IV-2
and IV-3 at low pH. While so far we have not been able to
identify the new product unambiguously, we explain these
changes in the same way as for II, by the stopping of the
reaction at the stage of an “earlier” diamagnetic product, and
(
26) Sch a¨ ublin, S.; H o¨ hener, A.; Ernst, R. R. J. Magn. Reson. 1974, 13,
1
96-216.
(
(
27) Reference 15, p 67.
28) Goez, M.; Frisch, I. J. Am. Chem. Soc. 1995, 117, 10486-10502.

CIDNP Study of Cysteine DeriVatiVes
J. Am. Chem. Soc., Vol. 118, No. 12, 1996 2889
about 0.05 ppm toward higher field such that the signal of the
methylene protons of HOOC-CH2-S- in Id coincides with
that of the corresponding protons in thiodiglycolic acid, the
signals VIII-3, IX-3, X-3, and XI-3 in both systems coincide
as well; moreover, relative polarization intensities are very
similar. All this clearly shows that VIII-XI are products of
reactions at HOOC-CH2-S-.
token, to changes of relative decarboxylation rates. An effect
of pH on decarboxylation of the S-carboxymethyl substituent
seems hardly conceivable, since this group is deprotonated
throughout the pH range investigated. Hence, it must be
concluded that above pKa2 cysteine decarboxylation is signifi-
cantly faster than below pKa2.
Decarboxylation Pathway. Two explanations are conceiv-
able for the increase of the decarboxylation rates of the cysteine
moieties above pH g pKa2.
A detailed discussion of the new products is outside the scope
of this work, which is focused on the reactions of the cysteine
29
moiety, and will be given elsewhere. Pertinent to the present
investigation is, however, what intermediate these polarizations
stem from. Time-resolved experiments show VIII, which gives
rise to the strongest CIDNP signal in Figure 6, bottom trace,
and also by far the largest of the four new signals, to be a cage
product (ꢀ ) +1). According to eq 1, the term sgn∆g × sgna
must thus be negative, which cannot be reconciled with
On the one hand, the deprotonated amino group stabilizes
+
the resulting alkyl radicals much better than does -NH3 .
3
3
AM1 calculations gave a difference in the heats of formation
-
1
of CH3 C˙ HNH2 and CH3CH2NH2 of 81 kJ mol ; for
+
+
CH3 C˙ HNH3 and CH3CH2NH3 this difference was computed
-1
to be 134 kJ mol . Hence, deprotonation of the amino function
should increase the driving force of decarboxylation by more
than 50 kJ mol and, unless the mechanism were different for
•+
•-
generation of these polarizations in radical pairs Id CB
-1
•+
because the sulfur-centered radical cations I possess very high
g values (i.e., ∆g > 0) and positive hyperfine coupling constants
the anionic and zwitterionic forms of our amino acids, should
lower the activation barrier accordingly.
3
11b
for the protons of interest, H .
However, radical cations of
On the other hand, the deprotonated amino group is also a
potential electron donor, so it would be natural to assume that
for pH g pKa2 a pathway via nitrogen-centered radical cations
•
+
structure HOOC-CH2-X -R, where X is a heteroatom such
as N, O, or S, easily lose CO2 to give R-substituted alkyl radicals
•
CH2-X-R. This decarboxylation is rapid on the CIDNP time
•+
•+
-
I′ , R-S-CH2-CH(N H2)COO , becomes accessible, i.e.,
that the first step of the reaction is electron transfer from nitrogen
instead of sulfur. As the findings of the preceding section
scale;30 hence, it occurs within nanoseconds or even faster. The
emissive polarization of VIII-3 is consistent with CIDNP
•
•-
•
+
generation in pairs CH -S-R CB : The g value of the
suggest, decarboxylation of I′ might be considerably faster
2
•
-
R-thioalkyl radical is certainly larger than that of CB (for
instance, CH2SCH3 has g ) 2.0049), the hyperfine coupling
constants of the protons at the radical center are negative, and
the product VIII contains a -CH2-S group instead of the
carboxymethyl group.29
than decarboxylation at the cysteine moiety of the sulfur-
•
31
•+
centered radical cations I , both processes, though, leading to
15
the same R-aminoalkyl radicals.
The polarization patterns again allow a clear-cut distinction
between these two alternatives, increase of the decarboxylation
rate by product control or an additional reaction channel.
At low pH (see Figure 6, bottom trace), the CIDNP spectrum
of Id is dominated by the signal VIII-3. The polarizations of
the starting amino acidswhich were the strongest signals in
the other systemssare noticeably smaller, and the polarizations
of the products of reaction at the cysteine group are very weak.
•+
Aminium cations I′ possess substantial hyperfine coupling
1
constants of the R-protons (i.e., H ), whereas the hyperfine
coupling constants of these protons in the sulfur-centered radical
•+
•+
cations I are negligible. Unless the decarboxylation of I′
were very fast on the CIDNP time scale, polarizations from
2
The polarization pattern of the educt Id (absorption for H and
3
1
H , no polarizations of H ) is the same as in the other systems,
indicating the intermediacy of sulfur-centered radical cations
•+
•-
primary radical pairs I′ CB should thus be detectable in the
regenerated starting amino acid. Experimentally, however, we
did not observe polarizations of I-1 in any case, so either
•
+
Id . Likewise, the polarization pattern of II (absorption for
1
2
H , emission for H ) is identical with that observed with the
other amino acids. These observations show that R-aminoalkyl
radicals are indeed formed at pH ≈ 7, so decarboxylation at
the cysteine moiety occurs, but decarboxylation of the car-
boxymethyl substituent prevails.
•
+
aminium cations I′ are not formed in these reactions or they
decarboxylate extremely rapidly as to prevent back electron-
•+
•-
transfer after intersystem crossing of the pairs I′ CB .
To test the latter hypothesis, we used alanine Ala, CH3-
CHNH2COOH) as a model compound, which bears an obvious
structural similarity to our cysteine derivatives but for which
electron transfer is only possible from nitrogen. Part of the
CIDNP spectrum obtained in the photoreaction of this amino
acid with CB at pH 12 is displayed in Figure 7. As is seen in
At pH above pKa2 (top trace of Figure 6), the picture is
different. Under these circumstances, the emission signal VIII-3
is still prominent32 and still the strongest of all the CIDNP
signals that are caused by decarboxylation at HOOC-
CH2-S-. However, the most intense product signal in the
CIDNP spectrum is now due to the vinylamine VI, the product
of decarboxylation of the cysteine function. As the magnetic
parameters of the intermediate R-thioalkyl and R-aminoalkyl
1
the figure, proton H of Ala is polarized in absorption, but no
2
polarizations can be detected for protons H . This is in
accordance with the spin density distribution expected for the
2
3
radicals are constant within the pH range considered, this
change of relative polarization intensities with pH must be
related to changes of relative product yields and, by the same
•
+
1
radical cation Ala . The polarization phase of H is consistent
with regeneration of the educts by in-cage back electron-transfer
(
2
5
µ ) +1, ꢀ ) +1, ∆g > 0, aH1 > 0); the fact that no CIDNP
(29) Goez, M.; Rozwadowski, J.; Marciniak, B. Manuscript in prepara-
signals are observed for pH < pKa2 further corroborates electron-
tion.
•+
transfer quenching. Thus, nitrogen-centered radicals I′ are
(30) Bowers, R. P.; McLauchlan, K. A.; Sealy, R. C. J. Chem. Soc.,
formed in the alanine experiment and give rise to polarizations
Perkin Trans. 2 1976, 915-921.
1
(
31) Gilbert, B. C. In Sulfur-Centered ReactiVe Intermediates in Chem-
of H in the substrate. No such polarizations are observed in
istry and Biology; Chatgilialoglu, C., Asmus, K.-D., Eds.; Plenum Press:
New York, 1990; pp 135-154.
(
pH > pKa2 is due to an interaction between -COO of the 4-carboxy-
benzophenone moiety and -NH3 of the cysteine function, which restricts
•+
the cysteine systems; hence, I′ is not an intermediate in these
3
4
reactions. The increased decarboxylation rate at pH above
32) The fact that VIII-3 is an AB system at pH < pKa2 and a singlet at
-
+
(33) Dewar, M. J. S.; Zoebisch, E. G.; Healy, E. F.; Stewart, J. J. P. J.
Am. Chem. Soc. 1985, 107, 3902-3909.
2
9
rotation.

2
890 J. Am. Chem. Soc., Vol. 118, No. 12, 1996
Goez et al.
Chart 1
polarizations attributable to RP1) than decarboxylation of the
cysteine moieties of the sulfur-centered radical cations I• at
pH e pKa2 (significant polarizations from RP1). This is in line
with expectation because in the former instance decarboxylation
occurs at an R-position with respect to the heteroatom, so there
is a direct interaction, whereas in the latter case decarboxylation
+
â
36
occurs at C . It is known that sulfur-centered radical cations
â
bearing carboxy groups at C are stabilized by formation of
five-membered rings possessing a two-center three-electron bond
between S and O (see Chart 1). It seems very likely that
â-decarboxylation also proceeds via this intermediate. For
completeness, we mention that for radical structures as shown
1
in Chart 1 the spin density at H should be negligible, so pairs
containing these intermediates would not give rise to polarization
patterns that differ substantially from those caused by the open-
chain cations I .
Figure 7. Top trace, part of the CIDNP spectrum of the system
-carboxybenzophenone/alanine (CB/Ala) at pH 12.0. Bottom trace,
•
+
4
the NMR spectrum of the same system in the dark. Only the resonances
Third, the polarizations of the products of RP2 are not simple
superpositions of polarizations from RP1 and RP2. Rather, they
can be described as arising in a radical pair with the combined
properties of RP1 and RP2 weighted with the respective
lifetimes. These so-called cooperative effects are well known,
and examples have even been reported where neither RP1 nor
of the educt protons (left, Ala-1, δ ) 3.17 ppm; right, Ala-2, δ ) 1.08
-3
ppm) are shown. Experimental parameters: [CB] ) 2 × 10 M, [Ala]
-
2
)
1 × 10 M, room temperature.
pKa2 must thus be due to the stabilizing influence of the
deprotonated amino group on the alkyl radical produced by
RP2 on its own causes polarizations but in combination they
•
+
decarboxylation of I .
35b,c
do.
The seemingly anomalous g-value difference of the pairs
Decarboxylation of I•+ and CIDNP. Fast chemical trans-
formations of radical pairs, RP1, into other radical pairs, RP2,
affect the CIDNP signals if these reactions occur on a time scale
•
•-
V CB (RP2) at low pH can be explained in this way. In
these reactions, ∆g is expected to be very large and positive
-
3
•+
•-
-
10
-6
35
(≈9 × 10 ) for RP1, I CB , and smaller by more than a
1
/k of some 10 s10 s. First, the polarizations in the
-4
factor of 10 and negative (≈ -7 × 10 ) for RP2. Since the
transformation RP1 f PR2 is not extremely fast on the CIDNP
time scale under these experimental conditions, the positive
contribution of RP1 to the effective value of ∆g outweighs the
negative contribution of RP2. With respect to the hyperfine
coupling constants, the effect is smaller because the coupling
constants have comparable magnitudes in both radical pairs.
At high pH, the decarboxylation is significantly faster, so the
influence of RP1 is correspondingly smaller, and the anomaly
is not observed.
products of RP1 decrease with increasing k because both the
yields of these products decrease and the shorter life of the
primary pairs reduces the probability of intersystem crossing
by Zeeman and hyperfine interactions, i.e., interferes with the
generation of nuclear spin polarizations in RP1. A manifestation
of this effect is the low relative intensity of the polarizations of
the starting amino acid that is observed in the photoreactions
of Id.
Second, the products of RP2 exhibit polarizations from both
radical pairs, their relative amounts depending on the magnetic
parameters of RP1 and RP2 and on k. This effect is found for
the decarboxylation products. From the relative contributions
of RP1, it can be inferred that the decarboxylation of the
Likewise, the behavior of Ie, for which RP1 influences the
relative polarizations of the products to a larger degree than in
the case of the other amino acids although the hyperfine coupling
constants should be almost identical, is rationalized by a lower
rate of decarboxylation. The carboxyethyl group can effect a
stabilization of the sulfur-centered radical cation in the same
way as the cysteine substituent can (Chart 1), and Ie is the only
one of our substrates that is capable of such an additional
stabilization. It should be noted that this compound provides
further evidence that decarboxylation is governed by product
•
+
carboxymethyl substituent in Id is much faster (almost no
(34) A referee raised the question whether the observation of exclusive
electron transfer from sulfur in our systems could be explained with
electrochemical or gas-phase ionization data. As electrochemical oxidations
of both amines and thioethers are irreversible, equilibrium potentials are
inaccessible, and the measured peak potentials deviate from them by an
amount depending on the rates of charge transfer and/or the rates of
subsequent chemical reactions of the radical cations. While correlations
within a class of compounds may thus still be permissible, comparison of
such data for compounds from different classes is certainly problematic.
With respect to gas-phase data, the ionization potentials (IPs) of simple
model compounds, e.g., Et2S and i-PrNH2, 8.43 eV (Lias, S. G.; Bartmess,
J. E.; Liebman, J. F.; Holmes, J. L.; Levin, R. D.; Mallard, W. G. J. Phys.
Chem. Ref. Data 1988, 17 (Suppl. 1)) and 8.63 eV (Aue, D. H.; Webb, H.
M.; Bowers, M. I. J. Am. Chem. Soc. 1976, 98, 311-317), respectively,
indeed seem to indicate a preference of electron transfer from sulfur.
However, the difference of these values is comparable to the precision of
the measurements and the changes of the IPs with the substituents at S or
N. Hence, we are reluctant to draw definitive conclusions from the IPs.
•+
control: For Ie , decarboxylation of the S-carboxyethyl group
would be conceivable as well as decarboxylation of the cysteine
moiety; however, the CIDNP spectra give no indication that
the first pathway is taken. This is consistent with the fact that
decarboxylation of the S-carboxyethyl substituent would yield
•
a primary alkyl radical, CH-CH2-S-R, which is much more
•
unstable than Ve .
In principle, the rates of the processes RP1 f RP2 could be
obtained from a quantitative evaluation of the CIDNP signal
3
5
(
35) (a) Kaptein, R. J. Am. Chem. Soc. 1972, 94, 6262-6269. (b) den
intensities. Unfortunately, this is not very accurate at present
Hollander, J. A. Chem. Phys. 1975, 10, 167-184. (c) Kaptein, R. In ref
5
1
d, pp 257-266. (d) den Hollander, J. A.; Kaptein, R. Chem. Phys. Lett.
(36) Marciniak, B.; Bobrowski, K.; Hug, G. L.; Rozwadowski, J. J. Phys.
Chem. 1994, 98, 4854-4860.
976, 41, 257-263.

CIDNP Study of Cysteine DeriVatiVes
J. Am. Chem. Soc., Vol. 118, No. 12, 1996 2891
Scheme 1
is oxidation by surplus ground-state sensitizer, which ultimately
leads to a sulfur-containing aldehyde IV or another product,
depending on pH. Relative rates of these two pathways were
determined by evaluating CIDNP intensities at different sensi-
tizer concentrations; by experiments at variable temperature, the
activation energy of fragmentation could be estimated.
Conclusions
Owing to the unique features of CIDNP spectroscopy, it was
possible to obtain some new insight into the mechanism of
photooxidations of cysteine derivatives in this work.
On the one hand, the polarization patterns, which can be
regarded as frozen signatures of short-lived intermediates that
may be difficult to observe directly, provided unambiguous
evidence that the first step of these reactions is electron transfer
from sulfur, even under conditions where electron transfer from
nitrogen could also occur. An order of the rates of the
decarboxylation processes that limit the life of the resulting
sulfur-centered radical cations could be established, and esti-
mates of the rate constants could be given, because the correlated
life of the radical pairs fixes a temporal window.
owing to the unavailable magnetic parameters of the radical
pairs involved. However, the CIDNP effects allow an order-
of-magnitude estimation of the decarboxylation rates. Decarb-
oxylation of the S-carboxymethyl substituent in Id is fast on
the CIDNP time scale but not as fast as to suppress all
polarizations of the starting amino acid, so it must proceed on
a time scale slightly shorter than about 1 ns. In contrast, at pH
•
+
On the other hand, the analytical potential of the detection
method, the signal enhancement by the CIDNP effect, and the
possibility to observe “early” diamagnetic products allowed
determination of relative rates of fragmentation and oxidation
<
pKa2, decarboxylation of the cysteine moieties in our sulfur-
centered radical cations is fairly slow on the CIDNP time scale,
as evidenced by the strong cooperative effects; on the other hand,
an upper limit is given by the life of the resulting R-aminoalkyl
radicals, about 50-70 ns. Hence, this decarboxylation must
occur on a time scale of some 10 ns.37,38 Cysteine decarbox-
ylation in basic medium falls in between these extremes. The
ability of this reaction to compete successfully with decarbox-
•
of the R-aminoalkyl radicals V produced by decarboxylation
of the radical cations. While this would in principle also be
feasible by analysis of the final products once the decay
•
pathways of V are known, the advantages of CIDNP spectros-
copy are that it affords precisely the latter information alongside
with the kinetic data and thus enables one to judge the validity
of the obtained kinetic parameters.
•
+
ylation of the S-carboxymethyl group in Id indicates that it
proceeds on a time scale of about 1 ns.
Reaction Mechanism. All our experimental observations
are in accordance with the mechanism displayed in Scheme 1.
Quenching of electronically excited CB by electron transfer is
the first step of the reaction, as already inferred from other
experiments.2 Electron transfer occurs from sulfur at every
pH; electron transfer from nitrogen can be ruled out. After
intersystem crossing of the resulting triplet radical pairs
Experimental Section
Chemical Substances and Sample Preparation. All amino acids
and the sensitizer CB were obtained commercially in the highest purity
available (>99%) and used as received. Solutions of the reactants in
D O were deoxygenated by bubbling purified nitrogen through them;
2
then the NMR tubes were sealed. pH values were measured with a
glass electrode.
a,b
•
+
•-
I CB , spin-allowed back electron-transfer regenerates the
starting materials, which bear polarizations from the primary
pairs.
CIDNP Measurements. The experiments were performed on a
Bruker WM-250 NMR spectrometer. For data acquisition and experi-
mental control, an 80486-based multitasking workstation equipped with
a Keithley AD converter and a home-made programable pulse generator
were employed. An excimer laser (XeCl, λ ) 308 nm) that was
triggered by the pulse generator served as light source. Samples were
illuminated from the side and within the active volume only; for this,
a modified probe39 was used. A few millijoules were absorbed in the
samples per flash, as determined actinometrically. The pulse sequences
for the pseudo-steady-state CIDNP measurements have been described
previously.9 This technique allows virtual elimination of background
signals and yields CIDNP signals that are undistorted by nuclear spin
relaxation in the diamagnetic reaction products.
The sulfur-centered radicals I•+ of the amino acid undergo
decarboxylation of the cysteine moieties to give R-aminoalkyl
•
37
radicals V ; these processes mostly occur within the correlated
8
-1
life of the radical pairs (k ≈ 10 s ) and become significantly
9
-1
faster (k ≈ 10 s ) when the amino group is deprotonated.
Lastly, two competitive pathways of deactivation are open
â
to the R-aminoalkyl radicals. One is fragmentation of the C -S
bond to give vinylamine VI, which is hydrolyzed to acet-
aldehyde II below pH ≈ 7.25, and a thiyl radical, which then
forms a combination product III with the sensitizer. The other
(
37) As a consequence, decarboxylation may also take place to some
•
+
extent in free radicals I , which then act as random phase precursors to
the pairs V CB•-.
•
Acknowledgment. Financial support by the Deutsche For-
schungsgemeinschaft and the Committee of Scientific Research,
Poland (Grant No. 2 P303 049 06), is gratefully acknowledged
J.R. thanks the Deutscher Akademischer Austauschdienst for a
fellowship.
(
38) Two entirely different effects could be responsible for the influence
•
+
•
of pH on the observed gross rate of the reaction I f V . On the one
hand, the actual decarboxylation step might simply be faster if the amino
function of I is deprotonated. On the other hand, as CB can act as a
base within the radical pair, the overall decarboxylation rate at pH <
pKa2 might be limited by deprotonation of -NH3 by CB , i.e., within the
•
+
•-
1
6
+
•-
cage. In a similar system, decarboxylation has been shown to depend on
JA9536678
3
d
such a deprotonation step. On the basis of our CIDNP data, we cannot
distinguish between these alternatives. We are indebted to a referee for
drawing our attention to this point.
(39) Goez, M. Chem. Phys. 1990, 147, 143-154.