Intramolecular H-atom transfer reactions in rigid peptides Correlated solvent and structural effects

Article abstract of DOI:10.1139/V10-110

The rates and the mechanism of intramolecular hydrogen-atom transfer across the rigid diketopiperazine spacer in two epimeric benzophenone-tyrosine dyads were studied by nanosecond laser flash photolysis in 12 non-protic and protic solvents. Effects of th

Full text of DOI:10.1139/V10-110

266
Intramolecular H-atom transfer reactions in rigid
peptides — Correlated solvent and structural
effects
¨
Anna Lewandowska, Gordon L. Hug, Gerald Horner, Dariusz Pogocki,
Franciszek Kazmierczak, and Bronislaw Marciniak
Abstract: The rates and the mechanism of intramolecular hydrogen-atom transfer across the rigid diketopiperazine spacer
in two epimeric benzophenone–tyrosine dyads were studied by nanosecond laser flash photolysis in 12 non-protic and
protic solvents. Effects of the stereochemistry on the ground-state structures and conformational equilibria and dynamics
have been addressed by means of NMR-based population analysis and long-term molecular-dynamics simulations. The ex-
cited triplet state of the benzophenone is found to be quenched intramolecularly by the remote phenol moiety with rates
that are highly dependent on the solvent and on the relative benzophenone–phenol orientation. In agreement with previous
studies, the kinetic diversity, with rates ranging from <105 s–1 to 2 ꢀ 107 s–1, can be attributed to retarding effects of the
bulk viscosity and to effects of specific solvation, which can be both rate retarding and rate accelerating. As an extension
of the previous knowledge, however, the phenomenology of the kinetic solvent effects, that is, the nature of the predomi-
nating solvent parameter and its absolute impact on the reactivity, is found to be highly correlated with structural effects.
Key words: kinetic solvent effect, stereoselectivity, biradicals, tyrosyl radical, molecular dynamics.
´
´
´
´
´
´
´
Resume : Utilisant la photolyse laser eclair, au niveau de la nanoseconde, on a etudie les vitesses et le mecanisme de re-
`
´
`
´
´
action du transfert d’atome d’hydrogene intramoleculaire a travers l’espaceur dicetopiperazine rigide dans deux dyades de
´
`
`
benzophenone–tyrosine dans douze solvants protiques et aprotiques. Faisant appel a l’analyse de populations a l’aide de la
´ ´
´
`
´
´
RMN et de simulations de dynamique moleculaire a long terme, on a etudie les effets de la stereochimie sur les structures
´
´
´
´
´
de l’etat fondamental et sur la dynamique et les equilibres conformationnels. On a trouve que l’etat triplet excite de la ben-
´ ´
´
´
´
´
´
´
zophenone est piege d’une fac¸on intramoleculaire par la portion phenolique eloignee avec des vitesses qui dependent beau-
´
´
´
´
coup du solvant et des orientations relatives de la benzophenone et du phenol. En accord avec les etudes anterieures, la
5
–1
7
–1
´
´
`
ˆ
´
´
diversite cinetique des vitesses qui varient de moins de 10 s a 2 ꢀ 10 s peut etre attribuee aux effets de viscosite glo-
´
´ ´
bale qui les diminuent et aux effets de solvatation specifique qui peuvent soit les accelerer ou les retarder. Toutefois,
´
´
´
´
´
comme extension aux connaissances acquises anterieurement, on a trouve que la phenomenologie des effets cinetiques du
`
`
´
´
´
solvant, c’est-a-dire la nature du parametre predominant du solvant et son impact absolu sur la reactivite, est intimement
´
reliee aux effets structuraux.
´
´
´ ´
´
´
´
Mots-cles : effet cinetique du solvant, stereoselectivite, biradicaux, radical tyrosyle, dynamique moleculaire.
´
[Traduit par la Redaction]
tions of phenols with triplet-excited carbonyl compounds
have stimulated recurrent interest.3–26 Hydrogen-atom trans-
fer (HAT) from the phenol to the excited carbonyl provides,
quite invariantly with respect to the nature of the carbonyl
compound, an efficient source of metastable radicals,
namely the carbonyl-derived ketyl radical and the phenoxyl
Introduction
Photoinduced hydrogen-abstraction reactions by triplet-
excited carbonyl compounds have been a central topic of
photochemical research for many years.1 Starting off with
the seminal study by Das and Scaiano in 1981,2 the reac-
Received 29 April 2010. Accepted 30 June 2010. Published on the NRC Research Press Web site at canjchem.nrc.ca on 15 February
2011.
This article is part of a Special Issue dedicated to Professor J. C. Scaiano.
A. Lewandowska, F. Kazmierczak, and B. Marciniak.2 Faculty of Chemistry, Adam Mickiewicz University, 60-780 Poznan, Poland.
G.L. Hug. Faculty of Chemistry, Adam Mickiewicz University, 60-780 Poznan, Poland; Radiation Laboratory, University of Notre
Dame, Notre Dame, IN 46556, USA.
1
¨
¨
¨
G. Horner. Faculty of Chemistry, Adam Mickiewicz University, 60-780 Poznan, Poland; Institut fur Chemie, Technische Universitat
Berlin, 10623 Berlin, Germany.
D. Pogocki. Faculty of Biology and Agriculture, University of Rzeszow, 35-601 Rzeszow, Poland; Centre of Radiation Research and
Technology, Institute of Nuclear Chemistry and Technology, 03-195 Warsaw, Poland.
1Corresponding author (e-mail: hoerner@amu.edu.pl).
2Corresponding author (e-mail: marcinia@amu.edu.pl).
Can. J. Chem. 89: 266–279 (2011)
doi:10.1139/V10-110
Published by NRC Research Press

Lewandowska et al.
267
the reacting chromophores for epimeric dyads (see
Scheme 1; from left to right, the structures denote the (L-L)
and (L-D) epimers). In addition, the presence of the DKP
ring exerts unique control over the conformations of the re-
spective side chains of the attached amino acids. Previous
studies of the conformations of substituted diketopiperazines
revealed structural preferences that significantly differed
from those of their open-chain analogues.37,38 Such struc-
tural preferences are expected to be reflected in the non-
bonding distances of the side chains and the inter-side-chain
reactivity.
radical. Ample use of these clean radical-affording pathways
has allowed the selective formation and the characterization
of biologically relevant phenoxyl radicals that are derived,
for instance, from the phenolic amino acid tyrosine (Tyr)7
and from the phenolic neurotransmitter adrenaline.6 Selec-
tive generation of fluorinated tyrosyl radicals within biolog-
ical matrices has allowed their use as a mechanistic probe
and as an external trigger of proton-coupled electron transfer
reactions in native proteins.27
This biochemical application was based on an intramolec-
ular variation of the carbonyl–phenol theme within cova-
lently bound dyads. Mechanistic studies of the
intramolecular HAT pathways corroborated the view that
such dyads share the reaction pathways with their bimolecu-
lar congeners and, thus, produce the corresponding biradi-
cals with high selectivity and efficiency as long as the
structural constraints of the dyad allow close contacts be-
tween the reacting moieties. The intramolecular HAT
quenching of, for instance, the benzophenone triplet state
(3bp) by an H-atom donor, RH, implies that >C=OꢁꢁꢁHR dis-
Preliminary work on the HAT between tyrosine and 4-
benzoyl-phenylalanine (bpa) side chains of the epimeric di-
ketopiperazines 1 and 2 (see Scheme 2a) actually showed
significantly enhanced stereoselectivity.39
Herein, we present the results of extended laser-flash pho-
tolysis studies of 1 and 2 that address the medium depend-
ence of the excited-state dynamics in more detail. These
studies are supplemented by the results of molecular-dynamics
simulations and NMR investigations of the ground-state
structures and conformations that establish a quantitative
picture of the conformer distributions of the dyads and de-
fine the molecular prerequisites of the individual quenching
geometries. As will be shown, the presence of the rigid DKP
linker imposes very different structural constraints on the in-
tramolecular mobility of epimers 1 and 2. These structural
differences translate into individual phenomenology of the
solvent effects for both of the dyads.
28
˚
tances below 3.1 A are successful. Studies of the intramo-
lecular HAT of several benzophenone-based dyads have
provided evidence that the chemical nature of the linker be-
tween the chromophores (that is, the flexibility and the di-
rectionality of the linker) is one major source of kinetic
diversity. Remarkable variations in the reaction rates of iso-
meric benzophenone–phenol dyads (regiomers and stereo-
mers) have been attributed to modulations of the distance
distributions and relative orientations of the reacting moi-
eties.10,12,19
Experimental section
In extension of previous work on the solvent effects of the
intramolecular HAT by Scaiano, Miranda, Leigh, and
others,2,12,16,19,23 our recent systematic studies of the HAT
reactions of peptide-linked benzophenone–tyrosine dyads
(bp[Tyr) have shown that this interpretation of structurally
imposed kinetic effects is incomplete. Instead, specific sol-
vent–solute interactions have been identified as the second
important control of the intramolecular HAT dynamics.29,30
Our results imply that the kinetic effects that are imposed
by the solvent and those that are due to the dyad structure
are correlated with each other and that they must not be
treated separately. In non-protic solvents the intramolecular
reactivity was generally found to be linearly correlated with
the solvents’ hydrogen-acceptor strength, which parallels the
phenomenology of the solvent dependence of bimolecular
HAT reactions of phenols.31–36 In particular, the HAT reac-
tion in two epimeric bp[Tyr dyads exhibited significantly
solvent-dependent stereoselectivity, which could be attrib-
uted to specific solvent–solute interactions.29
Synthesis and characterization
Chemicals and solvents (Sigma-Aldrich and BACHEM)
for synthesis were of the highest available analytical grade
and were used without further purification. Products were
identified by means of NMR techniques and combustion
1
analysis. Routine H NMR and 13C NMR spectra were re-
corded at 200/400 MHz and 50.3 MHz, respectively, in per-
deuterated dimethyl sulfoxide (DMSO-d6) and in deuterated
trifluoroacetic acid (TFA-d), with the resonances being re-
ported in parts per million downfield from tetramethylsilane.
1
Assignments of H and 13C resonances were based on two-
dimensional homo- and heteronuclear NMR techniques. An
unambiguous assignment of the strongly overlapping b-
proton resonances of the bpa and the Tyr residues in 1 and
2, as required for the analysis of the ground-state rotamer
distribution, was accomplished by two-dimensional nuclear
Overhauser enhancement spectra (see Figs. S1 and S2 in the
Supplementary data). The assignment of the HbR and HbS in
the Tyr side chain of 1 and 2 as the high-field and the low-
field doublet-of-doublet parts of the AB sub-spectrum, re-
spectively, followed established conventions.40 The same as-
signment was assumed to hold for the bpa side chain. The
synthesis of the cyclic dipeptides (diketopiperazines, DKPs)
1 and 2 followed the method invented by Nitecki et al.41
The fully protected open-chain precursors N-boc-(L/D)-bpa-
(L)-Tyr-OMe were obtained by standard peptide-coupling
methods as recently described in detail.29 The cyclic pepti-
des were obtained from the precursors after acidic deprotec-
tion of the N-terminus in formic acid followed by thermally
In the latter work the chromophores were connected by
3
open-chain peptide structures (see Scheme 1; A denotes bp
as the hydrogen acceptor; HD denotes hydrogen donor
PhOH), wherein the high intrinsic flexibility of the linker
blurs the stereoselectivity. Therefore, limiting the flexibility
of the peptide linker should allow deeper insights into the
relations between the kinetic phenomenology and the molec-
ular structure. The diketopiperazine (DKP) ring system pro-
vides such a peptide environment with reduced flexibility.
First, the DKP ring imposes a distinct orientation between
Published by NRC Research Press

268
Can. J. Chem. Vol. 89, 2011
Scheme 1. Schematic of the structural constraints of the intramolecular H-atom transfer between a hydrogen donor (HD) and an acceptor
(A) within linear and cyclic dipeptide scaffolds (asterisks denote chiral carbon centers; curved arrows (blue in the Web version) denote
additional degrees of rotational freedom in the open-chain arrangements).
Scheme 2. (a) Structures of the diketopiperazines 1 and 2 and notation of atom positions; (b) notation of the side-chain rotamers for the L-
configured side chains in 1 and 2 (R denotes the respective aromatic moiety); the D-configured bpa residue in dyad 1 is accounted for by an
interchange of the >NH and >CO groups.
Published by NRC Research Press

Lewandowska et al.
269
induced cyclization (full synthetic procedures and details of
characterization are given in the Supplementary data).
been taken into account following the approach of Zwanzig
and Ailawadi.55,56
Molecular modeling details
Nanosecond flash photolysis
The conformational space of the DKPs 1 and 2 was
sampled by means of Langevin dynamics (LD)42–48 done in
conjunction with the CHARMM2749 all-atom empirical po-
tential. Since there is no representation of the diphenyl-
ketone functionality in the original CHARMM force field,
the parameters for the atoms of its aromatic rings were taken
from the atoms of the phenylalanine residue. Similarly, the
carbonyl oxygen and carbon atoms of the ketone functional-
ity were calculated with parameters provided for these atoms
in the peptide bond. To achieve consistency of charge
distribution in the molecules, the atomic charges were recal-
culated using the partial equalization of orbital electronega-
tivity method of Gasteiger and Marsili.50 In the LD no
explicit solvent molecules were used, but the usual LD fric-
tional drag of the solvent was factored in using a collision
frequency for all heavy atoms equal to approx. 20 ps–1,
which reflects the viscosity of acetonitrile, fitting molecular
dynamics and experimental data.44,51 The free LD simula-
tions were done with the 2 fs time step and 100 ns propaga-
tion time, preceded by 12 ps of heating from 0 K to 300 K
and 18 ps of equilibration. For the force field calculations
we used version 7.1 of the HyperChem PC molecular mod-
eling package,52 which integrates the Langevin equation of
motion using the method of Allen and Tildesley.53
The setup for the nanosecond laser flash photolysis (LFP)
experiments and the data acquisition system have been pre-
viously described in detail.57,58 LFP experiments employed a
pulsed N2 laser (337 nm, 5 mJ, &8 ns) and a Nd:YAG laser
(266 nm and 355 nm, 5 mJ, 7–9 ns) for excitation. Transient
decays were recorded at individual wavelengths by the step-
scan method with a step distance of 10 nm in the range of
320 to 700 nm as the mean of 10 pulses. Samples for LFP
were deoxygenated with high-purity argon for 15 min prior
to the measurements. Experiments were performed in rectan-
gular quartz cells (1 cm ꢀ 1 cm). All experiments were per-
formed with freshly prepared solutions at room temperature
(295 1 K). In the case of 266 nm photolysis, excitation of
both chromophores (bp and Tyr) cannot be avoided. How-
ever, analysis of the molar absorption coefficients measured
for bp and Tyr59 shows that excitation of the tyrosine moiety
was less than 10%. We used 266 nm excitation (which al-
lowed us to use lower dyad concentrations) to avoid bimo-
lecular hydrogen abstraction by the triplet state from the
ground state of the dyad (self-quenching). Solvents for
time-resolved spectroscopy were of the highest available an-
alytical grade and were used without further purification.
Results
The rate constants of the close-contact formation between
the reacting moieties, here kOO, have been evaluated from
the simulations by means of number correlation functions,
following the routines that have been described in detail
previously.54 If the reactive and non-reactive conformations
are considered as being in equilibrium but the overall H-
abstraction reaction is very fast compared with the relaxa-
tion of equilibrium between the conformations, then the
observed rate constant kH should be comparable to a theoret-
ical rate constant, kOO, computed from the simulation via
eqs. [1–3].54 The numerator, hNOOi, in eq. [1] is the fraction
of the time that the distance between the carbonyl and hy-
NMR study of ground-state conformations
A previous density functional theory (DFT) study of the
DKPs 1 and 2 revealed the structural details of the respec-
tive quenching geometries and a qualitative interrelation
between the inter-side-chain reactivity and structural prefer-
ences, which are reflected in the non-bonding distances of
the side chains.39 The cited work has shown that a variety
of structures that differ in their relative side-chain orienta-
tions should be energetically accessible in solution. There-
fore it was of interest to obtain quantitative information
about the populations of the side-chain rotamers in the dy-
ads’ ground state. Herein we present the results of an
NMR-based analysis of the absolute populations of the side-
chain rotamers of the DKPs 1 and 2. An important result is
that the NMR-derived populations are distinctly different be-
tween the epimeric dyads 1 and 2.
The populations of the rotamers p(i) were extracted from
the vicinal coupling constants of the diastereotopic b-protons
and the a-protons of both of the amino-acid side chains,
3J(Ha–HbR) and 3J(Ha–HbS) (here bpa and Tyr; see
Scheme 2a for the notation of the proton positions),60 as re-
cently used in the analysis of open-chain dyads.29,30 The
acronyms EN, EO, and F denote side-chain rotamers wherein
the aromatic substituent R is extended to the adjacent nitro-
gen atom, extended to the adjacent carbonyl oxygen atom,
and folded back across the DKP ring, respectively. This ro-
tamer notation is based on the orientation of the side-chain
substituent R (see Scheme 2b) rather than on the dihedral
angles, to simplify the comparison between the epimeric
compounds 1 and 2. The computed rotamer populations for
DMSO-d6 and TFA-d are summarized in Table 1. The limited
˚
droxyl oxygens was less than 4 A during an 80 ns LD tra-
jectory. The denominator, trxn, in eq. [1] is the relaxation
time for the equilibrium between the reactive and non-reactive
conformations. It is related to a correlation function by the
fluctuation-dissipation theorem.55 The normalized number
correlation function CN(t) was used for this calculation (see
eq. [2]).
hNOO
i
½1ꢂ
kOO
¼
trxn
Z
þ1
½2ꢂ
trxn
¼
CNðtÞ dt
0
which has the form
hdNOOðtÞdNOOð0Þi
½3ꢂ
CNðtÞ ¼
2
hdNOO
i
Here dNOO(t) = NOO(t) – hNOOi and NOO(t) = 1 for dOO
£
2
˚
˚
4.0 A and NOO(t) = 0 for dOO ‡ 4.0 A. Note that hdNOO i =
2
hNOOi – hNOOi . The statistical error of the evaluation has
Published by NRC Research Press

270
Can. J. Chem. Vol. 89, 2011
Table 1. Summary of the side-chain rotamer populations p(i) of the diketopiperazines 1 and 2
computed from experimental coupling constants (DMSO-d6; 200 MHz).
bpa residue
Tyr residue
p(EN)
p(EO)
p(F)
p(EN)
p(EO)
p(F)
1
2
0.22 (0.40)
0.42 (0.55)
0.10 (0.17)
0.20 (0.16)
0.68 (0.43)
0.38 (0.29)
0.18 (0.21)
0.26 (0.33)
0.07 (0.21)
0.20 (0.16)
0.75 (0.58)
0.54 (0.51)
Note: Estimated uncertainty Dp(i) = 0.02 and 0.04 for p(EN) (p(EO)) and p(F), respectively. Data in
parentheses denote the populations in TFA-d.
Fig. 1. DFT minimized structures of highly populated solution-phase conformations of 1 (F–F) and 2 (F–EN). (Note that the 1 shown here is
the stereoisomer used in the LFP experiments, but the 1 shown in Scheme 2 is the stereoisomer used in the NMR and LD simulations.
These two species, both labeled 1, are identical molecules, being the mirror image of each other, and a limited set of duplicate LFP experi-
ments verified their identical chemical behavior.)
‘‘double-sandwich’’ like structure of 1 also prevails in
DMSO solution, which implies long inter-side-chain distan-
ces (see Fig. 1).
solubility of the DKPs under study interfered with a detailed
investigation of the solvent dependence of the rotamer pop-
ulations. It is noted that the population preferences of 1 and
2 are somewhat blurred in solutions of the very strong H-
bond donor TFA.
For 2, wherein both aromatic residues share the same
half-room above the DKP ring, the side-chain preferences
are less expressed. When compared with dyad 1, signifi-
cantly smaller populations of the F conformation are ob-
tained for 2. In particular, the F–F conformation of 2 would
contribute only with p(F–F) = 0.38 ꢀ 0.54 = 0.20, if statisti-
cally independent conformer populations can be reasonably
assumed. In fact, the molecular-dynamics simulations show
that this latter assumption is not valid for 2 and that even
smaller populations prevail (vide infra). This finding of di-
minished populations of F–F is important for the interpreta-
tion of the excited-state dynamics of 2, because
photoinduced HAT within 2 is limited to the F–F conforma-
tion of the side chains wherein both aromatic moieties are
back-folded across the DKP ring at the same time. It appears
that repulsive interactions between the bulky aromatic
groups favor the population of sterically less crowded con-
formations in the latter structure. In particular, a simultane-
ous population of the F rotamer by both side chains appears
less probable.
The NMR-derived populations for DMSO solutions of 1
and 2 reveal side-chain conformational preferences that dif-
fer significantly from those of open-chain congeners. The ar-
omatic side chains in linear peptides preferably populate the
rotamer EN in polar solvents such as DMSO, with p(EN) ‡
0.7.61–64 In contrast, EN appears to be much less favorable
in the cyclic peptide 1, where both aromatic residues pref-
erably populate the back-folded F conformation instead.
This behavior is quite common for the DKPs that carry aro-
matic amino acids and has been associated with dipole-
induced dipole interactions between the DKP amides and
the aromatic ring.65,66 Based on the assumption that the con-
former populations of the two side chains of 1 are statisti-
cally independent,
a favored population of the F–F
conformation can be expected with p(F–F) = 0.68 ꢀ 0.75 =
0.51. Thus, in full agreement with the outcomes of the pre-
vious DFT studies in an acetonitrile (ACN) continuum, a
Published by NRC Research Press

Lewandowska et al.
271
Fig. 2. Extracts of long-term molecular-dynamics simulations for 2
(Langevin dynamics; implicit-solvent model for ACN): (a) time
variation of the inter-side-chain distance d(O–O) between the tyro-
sine (Tyr) hydroxylic oxygen atom and the 4-benzoyl-phenylalanine
(bpa) carbonyl oxygen atom; (b) time variation of the bpa side-
chain dihedral angle; (c) time variation of the Tyr side-chain dihe-
dral angle.
intramolecular HAT in the range of 107 s–1 (see further dis-
cussion).
As shown by DFT studies, close contacts of <3.2 A
within dyad 1 require the population of conformations that
do not refer to side-chain rotamers. The formation of such
structures does not appear at all during the 100 ns simula-
tion (see Fig. S3 in the Supplementary data).
˚
Laser flash photolysis: spectral and kinetic analyses
The excited-state dynamics of the dyads 1 and 2 were
studied by means of nanosecond laser photolysis in 12 dif-
ferent solvents and solvent mixtures. Transient products
were identified, and the efficiency of their formation was
quantified by means of transient-absorption spectroscopy
and spectral resolutions.
In all cases the triplet state was identified as the predom-
inate intermediate in the early spectra by its characteristic
absorption maxima at 325 nm (not shown) and 520 nm in
addition to a long-wavelength absorption above 600 nm.
The triplet state spectra of both dyads were indistinguishable
and showed little variation with the change in solvent.
For bp[Tyr dyads that are related to 1–2, intramolecular
HAT was found to be the predominate quenching pathway.
The quantum yields for the formation of the respective bir-
adicals were reported to be unity.61 Thus, any significant
spectral evolution with a spectral shift from 525 to 540 nm
and accompanied by a growth at 405 nm would be indica-
tive of the formation of ketyl and tyrosyl radicals, respec-
tively. We found that the two dyads showed distinctly
different spectral evolution during and after the decay of
the triplet state.
This effect of the structure is illustrated by the transient
spectra observed for solutions of 1 and 2 in ACN (Fig. 3).
While the spectrum obtained for 1 in pure ACN after
200 ns largely resembles the original triplet spectrum, with
its distinctive 520 nm absorption, the spectrum obtained for
2 carries significant contributions from a tyrosyl radical and
a ketyl radical after only 100 ns. Note in particular the spec-
tral shift of dyad 2 from the initial maximum of 520 nm to
540 nm, indicating the transition from its triplet to its ketyl
radical. As shown in Fig. 3b, a change in the solvent did not
significantly affect the transient absorption spectra of dyad
2, an observation that holds for all of the studied solvents.
Accordingly, deconvolution of the transient absorption spec-
tra for compound 2 in an ACN–H2O (2:1 v/v) mixture gave
the same results as previously reported for pure ACN.39 The
quenching is found to proceed intramolecularly with high ef-
ficiency. This finding is in accordance with the lack of a
concentration dependence of the triplet decay rate constants
(see below). However, the intermediate (biradical) decays on
the same time scale as it is formed. This behavior interferes
with biradical accumulation and is contrary to the general
observation that biradicals decay on a much longer time
scale than do their parent triplets.12,16,19,61
Conformational dynamics
The interconversion between different molecular confor-
mations has been studied by means of long-term molecular-
dynamics simulations (100 ns duration; implicit-solvent
model for ACN). Figure 2 summarizes the time variation of
3 molecular parameters (both side-chain dihedral angles and
the characteristic inter-side-chain distance) within a 30 ns
simulation window for dyad 2. In Fig. 2a the sum of the res-
ident times at each distance (d(O–O)) should be viewed in
regard to the total period of the simulation (100 ns); i.e.,
when the sums of these resident times are divided by the to-
tal time of the simulation, they basically reflect the relative
existence of the dyad at each O–O distance (probability dis-
tribution). (Respective plots for 1 are given in the Supple-
mentary data.)
In agreement with the NMR results, mainly 3 rotamers
(EN, EO, and F) of each side chain are populated during the
simulation of both compounds. Of particular interest was the
probability of the F–F conformation of 2, which allows for
˚
close contacts of <4.0 A (Fig. 2) between the reacting moi-
eties and closely resembles the quenching geometry of this
dyad. As stated above, the simulations strongly disfavor this
conformation: the calculated probability of the population of
the F–F conformation is only 1.4%. In conclusion, the F–F
conformation, although shown to be a rare conformation
during the simulations for 2, irrespective of the chosen start-
ing geometry, is sufficient to explain the reactivity toward
In contrast to the largely solvent-independent behavior of
2, the evolution of the transient spectra of dyad 1 is strongly
dependent on the nature of the solvent. In particular, the
spectrum obtained after 200 ns in the ACN–H2O (2:1 v/v)
mixture differs dramatically from the one in pure ACN and
is fully consistent with efficient formation of ketyl and tyro-
syl radicals. To clarify the nature of this solvent effect, the
Published by NRC Research Press

272
Can. J. Chem. Vol. 89, 2011
Fig. 3. Transient absorption spectra obtained during laser flash photolysis at 355 nm of deoxygenated solutions (9 ꢀ 10–4 mol L–1) of (a)
compound 1 with a time delay of 200 ns and (b) compound 2 with a time delay of 100 ns; measurements in ACN (filled symbols) and in
ACN–H2O (2:1 v/v) (open symbols).
Fig. 4. (a) Resolutions of transient absorption spectra taken 1400 ns after 355 nm laser pulsing (5 mJ) of a solution of 1 in ACN. (b) Con-
centration profiles for the transients obtained from the deconvolution of transient absorption spectra of 1 in ACN. (c) Resolutions of transi-
ent absorption spectra taken 130 ns after 355 nm laser pulsing (5 mJ) of a solution of 1 in ACN–H2O (1:1 v/v). (d) Concentration profiles
3
for the transients obtained from the deconvolution of transient absorption spectra of 1 in ACN–H2O (1:1 v/v). &, triplet state bp; *, ketyl
radical bpHꢃ; D, tyrosyl radical Tyr(Oꢃ);
^
, bp radical anion bpꢃ–; *, experimental data. Solid curves are the resulting fits from the regres-
sion analysis.
transient spectra obtained for compound 1 in ACN and
ACN–H2O (1:1 v/v) were deconvoluted into their spectral
components, from which concentration–time profiles were
constructed. Exemplary resolutions are given in Figs. 4a
and 4c and the resulting profiles are displayed in Figs. 4b
and 4d. These profiles reveal the distinct difference in the
triplet quenching of dyad 1 in the two solvents. That is, the
reaction is found to be intermolecular in ACN, whereas it is
predominantly intramolecular in the ACN–H2O (1:1 v/v)
mixture. To simulate quantitatively the experimental data
obtained in ACN for dyad 1, only 3 components were
syl radicals (Tyr(Oꢃ)) (see Fig. 4a). The concentrations of
bpHꢃ and Tyr(Oꢃ) were found to be equal within the error
estimation. The triplet decay obtained from the concentra-
tion profile is 1 ms, which is well in agreement with the trip-
let lifetime obtained from the kinetic traces at 630 nm. From
the decomposition of the transient absorption spectra and the
resulting concentration profile, the quantum yield for the
formation of both radicals was calculated to be equal to
25% in ACN. Based on the long lifetimes of the radicals,
which are decaying for tens of microseconds, we can con-
clude that they are formed mainly in self-quenching reac-
tions (see Scheme 3).
3
needed: the triplet state bp, ketyl radicals (bpHꢃ), and tyro-
Published by NRC Research Press

Lewandowska et al.
273
Scheme 3. Deactivation paths (red, bimolecular; blue, unimolecu-
lar) of the triplet-excited state of the dyads 1 and 2.
vents. Thus, in contrast to the previous study of a pair of
open-chain epimeric dyads, the triplet lifetimes of the two
cyclic structures studied here exhibit completely distinct sol-
vent dependencies.
To study the solvent effects on the intramolecular HAT
pathway, the corresponding rate constants, kH, have been
isolated from the observed gross triplet-decay rate constants,
kobs. The latter have been extracted from the decay profiles
at 630 nm (mono-exponential fits), the decay profiles at
520 nm (bi-exponential fits), and, in some cases, the triplet
component of the concentration profiles (as in Figs. 4b and
4d; mono-exponential fits). Agreement between these meth-
ods was within 5%. The absolute value of the triplet-decay
rate constant, kobs, is the sum of the rate constants of the in-
dividual decay processes (Scheme 3). The evaluation proce-
dures take into account (i) the solvent dependence of the
intrinsic triplet lifetime of the bp chromophore by referenc-
ing it to the behavior of unsubstituted bp in the respective
solvent and (ii) the effects of self-quenching (see Supple-
mentary data). Due to the low laser power, effects of triplet–
triplet annihilation were negligible.
A totally different situation was observed for dyad 1 in
ACN–H2O (1:1 v/v). First, the triplet lifetime was reduced
by a factor of 10. Second, to reproduce the experimental
3
data, 4 components had to be used, namely bp, ketyl radi-
cals (bpHꢃ), bp radical anions (bpꢃ–), and tyrosyl radicals
(Tyr(Oꢃ)). The rates of formation of the radical species
bpHꢃ, bpꢃ–, and Tyr(Oꢃ) were, within experimental error,
identical to the triplet-decay rate of dyad 1. The total
amount of bpHꢃ and bpꢃ– together was equal to the Tyr(Oꢃ)
concentration, within experimental error. Formation and de-
cay of both the ketyl radical and the tyrosyl radical appear
to follow the same decay law. This is in accordance with an
intramolecular reaction mechanism. The initial triplet con-
centration, after extrapolation to the end of the pulse, was
7.1 mmol L–1. The maximum concentration of the tyrosyl
radical was determined to be 6.9 mmol L–1. This value was
obtained from fitting the Tyr(Oꢃ) concentration profile to a
function that takes into account the growth and the decay of
the species. From this value and the above triplet concentra-
tion (actinometer), the quantum yield of the tyrosyl radical
is seen to be close to unity. The same procedure was applied
to the bp radical anion, and its quantum yield of formation
was estimated to be 0.18. The results proved that not only
can dyad 1 (which has the benzophenone and phenol moi-
eties on opposite sides of the DKP ring) undergo intramolec-
ular HAT, but this reaction can occur with quantum yields
close to unity. Analogous behavior has been observed in
other solvents such as ACN–H2O mixtures, 2,2,2-trifluoro-
ethanol (TFE), and 1,1,1,2,2,2-hexafluoropropanol (HFIP),
except that the bp radical anion was detected only in ACN–
H2O mixtures.
A detailed kinetics data evaluation and a description of
the way that we dealt with competition between intra- and
intermolecular quenching paths can be found in the Supple-
mentary data.
While the triplet decay of dyad 2 exhibited no significant
concentration dependence, LFP results for dyad 1 in non-
protic solvents, such as ACN or dichloromethane, show that
the values for kobs were concentration dependent. This be-
havior is due to contributions to bimolecular quenching
from self-quenching, which leads to the ketyl radical and
the tyrosyl radical with the rate constant kSQ. The latter spe-
cies decayed by bimolecular processes within tens of micro-
seconds and therefore gave a virtually constant offset in the
transient spectra on the time scale of the experiment. The
self-quenching process was efficient in non-protic solvents
where the reactivity toward intramolecular HAT was small.
Notably, the kH calculated from experiments at different
concentrations, 2 ꢀ 10–5 mol L–1 (lexc = 266 nm) or 9 ꢀ
10–4 mol L–1 (lexc = 337 nm), were in a good agreement
(Table 2).
The rate constants for the intramolecular HAT, kH, calcu-
lated for dyad 1 range from <1 ꢀ 105 s–1 in EtOAc to 2 ꢀ
107 s–1 in HFIP (see Table 2). The rate constants for the in-
tramolecular HAT for dyad 2 varied only from 2.2 ꢀ 107 s–1
in ACN to 8.2 ꢀ 106 s–1 in TFE. The uncertainty in the rate
constant values derived in this manner increases signifi-
cantly with H-abstraction contributions from the solvent.
The results presented in Table 2 show a striking contrast
between the two diastereomers with regard to the depend-
ence of their kH on solvent. The results show that a change
in the configuration of just one chiral center produced com-
pletely distinct changes in the reactivity of the intramolecu-
lar HAT with a change in the solvent. The reaction rates for
dyad 1 were found to be strongly solvent dependent, while
the solvent was found to be of minor importance for the re-
action rates of dyad 2.
The steering effects of the solvents on the efficiency of
the intramolecular triplet quenching of dyads 1 and 2 mani-
fest themselves also in the triplet lifetimes, which exhibit a
remarkably distinct influence of the solvents. Some exem-
plary kinetic profiles, recorded at 630 nm, that illustrate the
distinct solvent dependencies of the triplet decays of 1 and 2
are given in Fig. 5. For dyad 2, where the benzophenone and
tyrosine moieties occupy the same side of the DKP ring, the
kinetic profiles coincide with each other, irrespective of the
solvent. This indicates little impact of the solvent on the
triplet lifetimes. Contrasting behavior was seen in dyad 1,
where the triplet lifetimes were very sensitive to the solvents
and varied between approx. 50 ns in HFIP and >1000 ns in
ACN or EtOAc. In general, the triplet lifetimes of 1 are re-
duced dramatically on going from non-protic to protic sol-
This contrasting solvent behavior toward the intramolecu-
lar HAT observed in 1 and 2 is unusual for a pair of diaster-
eomers. In this regard they are found to be a unique pair of
diastereomers with a unique solvent dependence of the
Published by NRC Research Press

274
Can. J. Chem. Vol. 89, 2011
Fig. 5. Normalized decay profiles observed at 630 nm and mono-exponential fits to the decays for solutions of dyads 1 (a) and 2 (b). a,
ethyl acetate; b, ACN; c, ACN–H2O (2:1 v/v); d, trifluoroethanol.
Table 2. Summary of solvent parameters and the kinetic data of dyads 1 and 2.
Solvent parameters
Rate constants
Ha
Ha
Solvent
Chloroform
Dichloromethane
1,2-Dichloroethane
Acetonitrile
Sa2
Sb2
hb
kH (1)c
(2.2)
1.9 (2.0)
(1.7)
0.3 (0.3)
0.3
0.1
8.5
2.3 (2.8)
21.1
3.0
5.5
8.4
kH (2)c
(11.5)
(14.5)
(13.6)
Sd
1
2
3
4
5
6
7
8
0.15
0.1
0.02
0.05
0.11
0.32
0.33
0.45
0.25
0.47
0.537
0.437
0.77
0.37
1.2
0.423
1.8e
0.544
1.02
—
5.2
7.2
0.1
10.7
0.07
0
20.7 (20.0) 73.7
Benzonitrile
Ethyl acetate
9.5
14.4
8.2
9.2
15.7
—
—
13.9
31.6
131
0.97
0.0
Trifluoroethanol
Methanol
Hexafluoro-2-propanol
Acetonitrile/H2O (9:1 v/v)
Acetonitrile/H2O (4:1 v/v)
Acetonitrile/H2O (2:1 v/v)
0.57
0.43
0.77
0.55f
0.61f
0.62f
3.9
0.74
9
0.10
10a
10b
10c
&0.34g
&0.34g
&0.34g
—
—
—
—
1.7
aTaken from ref. 67.
bAt 25 8C (ref. 68).
cResults from 355 nm and 337 nm LFP (106 s–1); experimental error 5%, 0.7 mmol/L £ [1], [2] £ 2.6 mmol/L. Data in italics:
experimental error 20%. Data in parentheses from 266 nm LFP, 20 mmol/L £ [1], [2] £ 40 mmol/L.
dS = kH (2) / kH (1).
eTaken from ref. 69.
fInterpolated from data in ref. 70.
gLinear interpolation from data in pure solvents, weighted by molar fractions.
stereoselectivity. Both epimers showed reactivity with re-
Discussion
spect to intramolecular HAT, but with a remarkable chiral
In this study the excited-state dynamics of 1 and 2 have
discrimination in their rate constants, and this chiral discrim-
been taken as probes of structural effects on the efficiencies
ination strongly depends on the solvents. We and others pre-
and rates of intramolecular hydrogen-atom transfer reac-
viously reported stereoselection in HAT rates of
tions. We find that the distinct steric constraints of the dy-
diastereomeric ketone–phenol dyads in the range of 2–10
ads, as outlined above, are reflected in distinct dynamics of
when expressed as the ratio of the rate constants of the intra-
the dyads’ triplet states. The dyads particularly differ in their
molecular HAT between remote phenol and ketone moieties
dependencies on the medium, as discussed below.
of these compounds.16,19,29 The stereoselectivity (S = kH (2) /
The intramolecular HAT in dyad 1 was found to be
kH (1)) found in this work varied from 131 to 0.74 (Table 2).
strongly solvent dependent, with the rate constants, kH, rang-
ing from <1 ꢀ 105 s–1 to 2 ꢀ 107 s–1. We recently observed
In particular, the epimers revealed marked stereoselectivity
in some of the non-protic solvents such as EtOAc and ben-
zonitrile, much higher than has ever been reported previ-
ously. Quite unexpectedly, the stereoselectivity was
significantly reduced in protic solvents. At extremes, com-
pound 1, which has the reactive moieties on opposite sides
of the DKP ring, became more reactive than 2 in hexa-
fluoro-2-propanol as a medium.
a similar range of rate constants for the intramolecular HAT
in one open-chain bp[Tyr dyad,30 whereas the HAT reac-
tions in the open-chain congeners of 1 and 2 in a given sol-
vent are generally faster by one to two orders of
magnitude.29 Such solvent-driven modulations of bimolecu-
lar HAT reaction rates of phenols have been generally at-
tributed to effects of specific solvation of the phenol, which
Published by NRC Research Press

Lewandowska et al.
275
Fig. 6. (a) Plot of H-atom transfer rate constants, log kH, against Abraham’s effective H-bond acceptor ability (refs. 71, 73), Sb2H, for protic
solvents (open symbols) and non-protic solvents (filled symbols); (b) plot of H-atom transfer rate constants, log kH, against Abraham’s ef-
H
H
fective H-bond donor ability (refs. 71, 73), Sa2H, for solvents with Sb2 ‡ 0.25 (filled symbols) and solvents with Sb2 £ 0.11 (open
symbols); red symbols represent data for the 1-(D,L) epimer and blue symbols represent data for the 2-(L,L) epimer. (c) Semilogarithmic plot
of stereoselectivity of H-atom transfer in dyads 1 and 2 as a function of the solvents’ H-bond acceptor ability; open symbols represent protic
solvents. (d) Semilogarithmic plot of stereoselectivity of H-atom transfer in dyads 1 and 2 as a function of the solvents’ H-bond donor
H
ability; open symbols represent solvents with Sb2 £ 0.11. Solvent codes as in Table 2.
H
were coined as kinetic solvent effects. In particular, the pre-
vious work on the bimolecular HAT of phenols has shown
that rate constants are correlated with the solvents’ hydrogen-
bond acceptor ability.31–36,71,72 That is, the logarithmic rate
constants depended linearly on a single solvent parameter,
namely Abraham’s H-bond acceptor ability,71,73 irrespective
of the molecularity of the reaction.29 Notably, the solvent
dependence cannot be fully explained by solute–solvent in-
teractions of the tyrosine moiety. As shown here, the single-
parameter dependence of the intramolecular HAT reaction
that sufficed to explain the reactivity of the open-chain con-
geners of 1 and 229 is not sufficient to explain the difference
in reaction rate constants of the cyclic dyad 1. Our results
show that, in contrast to the phenomenology of the kinetic
solvent effect, which necessarily connects specific solvation
with reaction hindrance, specific solvation can also promote
the HAT reactions of phenols.
weak effective H-bond donor abilities, Sa2 (red filled sym-
H
bols in Fig. 6a), have a linear dependence on Sb2 , with a
slope b = –3.1 (r2 = 0.995) and an intercept of 6.5. Thus,
this dyad is another example where the slopes of the loga-
rithmic intramolecular HAT rate constants versus Sb2H plots
are characteristic for each dyad but not characteristic for
each donor.29,30
The second region in Fig. 6a is that formed by the protic
solvents with significant H-bond donor abilities. The protic
solvents seem to allow much higher rate constants than their
non-protic counterparts having similar H-bond acceptor abil-
H
ities (nearly matching Sb2 ). For example, ethyl acetate and
MeOH have almost the same H-bond acceptor ability, but kH
values for 1 differ by a factor of 10. Another example is the
contrasting behavior of pure ACN and the ACN–H2O (2:1 v/
v) mixture, in which kH values for 1 differ by a factor of 25.
The large increase in the intramolecular HAT rate constant
in protic solvents might be due to a switch in the mecha-
nism.
The decrease of the HAT rates with the H-bond acceptor
ability of the non-protic solvents suggests that negligible in-
tramolecular reactivity is expected for dyad 1 in solvents
In the following discussion the focus will first be on the
data for dyad 1 displayed in Figs. 6a and 6b (red symbols).
A plot of the kinetic data for dyad 1 from Table 2 against
the effective H-bond acceptor ability of the solvents
(Fig. 6a) exhibits two different regions that reflect the char-
acter of the solvents. HAT rates in non-protic solvents with
H
with Sb2 > 0.3. However, the data in Table 2 show that
Published by NRC Research Press

276
Can. J. Chem. Vol. 89, 2011
H
protic solvents with high Sb2 values show significant reac-
tivity for dyad 1. In Fig. 6b, data from Table 2 are plotted
Moreover, NMR, molecular-dynamics simulations, and
DFT results39 have shown that conformations with ex-
tremely long distances (12.8 A) between the reactive moi-
H
˚
against Sa2 ; with the exception of the chlorohydrocarbons,
there is a linear correlation for dyad 1 with a positive slope
of 2.3 (r2 = 0.977) and an intercept of 5.2. That is, the reac-
tivity increases with the strength of the solvent–solute inter-
action.
eties are the most favorable ones in solution (Fig. 1;
Table 1). The HAT rates in non-protic solvents are limited
by the accessibility of free (i.e., not H-bonded) tyrosine. In
contrast, the kinetic data obtained in protic media show a
correlation between HAT rates and the H-bond donor ability
of the solvent. Thus, at least for intramolecular HAT reac-
tions, the solvation of the abstracting species should be con-
sidered, a priori, as important as the solvation of the H-
donating species. However, the action of the protic solvent
in this case and its impact on the mechanism of the HAT
step remain under discussion and are the subject of further
work.
In summary, opposite effects were exerted by non-protic
and protic solvents on intramolecular HAT in dyad 1. The
relevant solvent parameters favoring efficient intramolecular
H
HAT for dyad 1 are the solvents’ small Sb2 values and
H
high Sa2 values. This was realized by HFIP, for example.
The opposite behavior was seen in the poor reactivity in
EtOAc, which can be explained on the basis of its high
H
H
Sb2 (0.45) and low Sa2 (0.00).
Several pieces of evidence favor electron transfer involv-
ing pathways that are driven by specific H-bonding interac-
tions of the electron acceptor with the protic solvent. An
electron-transfer pathway in protic solvents has been ob-
served recently for a sterically restricted bp[TyrOH dyad.
In this structure, the directionality of the benzamide linker
of the dyad effectively excludes close intramolecular con-
tacts. Studies of the solvent effects on the nanosecond-
transient dynamics and on the chemically induced dynamic
nuclear polarization following LFP have shown that in
protic solvents the H abstraction reaction is initiated by elec-
In conclusion, the solvent dependence of the cyclic dyad
1 sharply diverges from the behavior of its open-chain con-
gener29 but largely coincides with the phenomenology ob-
served recently for a more rigid open-chain dyad.30
Obviously, the observed phenomenology of the kinetic sol-
vent effects reflects structural parameters of the dyads. This
is most evident from the behavior of the cyclic dyad 2,
which lacks any significant dependence on the H-bonding
properties of the solvent.
H
The plot of the logarithmic rate constants versus Sb2 for
dyad 2 exibits no significant variance with the solvents’ H-
bond accepting or H-bond donating properties (blue symbols
in Figs. 6a and 6b). Actually, the slopes are close to zero in
both cases, which means that the rate constants for HAT for
dyad 2 are independent of these parameters. Accordingly,
the stereoselectivity of the HAT in 1 and 2, expressed as
the ratio of the rate constants and shown in Figs. 6c and 6d,
directly reflects the solvent dependence of the HAT of 1.
The sharp decrease of the HAT reaction rates for dyad 1 in
non-protic solvents with hydrogen-bond acceptor tendencies
and, on the other hand, the sharp increase of the reaction
rates in protic solvents with hydrogen-bond donor tendencies
are the dominant sources of the solvent dependence of the
stereoselectivity (Figs. 6c and 6d). Our findings are in con-
trast with other results where the stereoselectivity of the
¨
tron transfer (G. Horner and colleagues, unpublished data).
Similar conclusions have been drawn from studies of the bi-
3
molecular electron transfer between anisole and bp, where
dramatic activating effects of strongly H-bonding solvents
3
on the one-electron oxidation of anisole by bp show that
this phenomenon is not limited to phenols and might be of
general significance.74
In contrast to the above discussion, neither of the rate-
determining effects of solvation seems to be important for
the HAT in dyad 2, as there was little variation in the kH
rate constant among solvents. The ratio of the highest kH ob-
tained in ACN and the smallest kH obtained in TFE is just
2.4 for the triplet state of dyad 2. The absence of a marked
solvent dependence, which was observed for all other dyads,
therefore deserves further consideration. In particular, it
raises the question, what are the factors determining the rate
constants if there is no impact of specific solvation?
The observation that the triplet state of dyad 2 was
quenched slightly more slowly in solvents with higher vis-
cosity, which led to smaller values of kH (TFE and benzoni-
trile, see Table 2), points to the importance of the
intramolecular dynamics. Similar observations were made
for its open-chain analogue.29 The observation of a greatly
diminished triplet reactivity of 2 in highly viscous tert-
butanol further corroborates viscosity as another source of
kinetic diversity. That is, the dynamics of the molecular mo-
tions bringing dyad 2 to its reactive state appear to be rate
determining in all solvents, which implies that the intrinsic
HAT reaction rate of the dyad, irrespective of its state of
solvation, markedly exceeds the rates of the intramolecular
motions.
H
HAT process was solely dependent on the Sb2 , irrespective
of the protic or non-protic character of the solvents.
In conclusion, we have studied the behavior of the two
cyclic-bridged dyads 1 and 2 in 12 different solvents and
mixtures. It was shown that the properties of a solvent, i.e.,
its ability to act as a hydrogen bond donor or acceptor, play
an important role in determining the HAT rate constants in
only one of the two epimers in this study.
The contrasting solvent dependencies of the HAT rate
constants for dyad 1 in protic and non-protic solvents can
be interpreted as being due to fundamentally different HAT
mechanisms in the respective solvents. In favor of this inter-
pretation are the significant enhancement of the reaction
rates in protic solvents and the presence of the bp radical
anion in an ACN–H2O (1:1 v/v) mixture. In addition, it is
important to note that the intercepts of the single-solvent pa-
rameter regressions in Figs. 6a and 6b (red lines) differ by
almost 2 orders of magnitude. These values represent the
isolated and solvent-independent HAT rate constants and,
thus, they should not vary if there is only a single common
mechanism of HAT in protic and non-protic solvents.30
To check whether molecular dynamics are the rate-
determining steps of the intramolecular HAT of dyad 2, we
have estimated the rate constant for the formation of close-
contact conformations, kOO, based on the LD trajectories de-
Published by NRC Research Press

Lewandowska et al.
277
scribed above. In the calculation of the relaxation time (trxn
from the integration in eq. [2]), we have neglected the nega-
tive values of CN(t), which occur for t > 1.4 ns, since they
were becoming artificially large with increasing delay times.
Neglecting these negative values of CN(t) means that our
computed value of 0.4 ns for the relaxation time is an upper
limit. The average fraction of conformations, hNOOi, with
acid side chains occupy different half rooms of the diketopi-
perazine spacer, two solvent parameters influence the H ab-
straction rates, namely the effective H-bond acceptor ability,
H
H
Sb2 , and the effective H-bond donor ability, Sa2 , as de-
fined by Abraham.67 In sharp contrast, the syn-configured
dyad 2 represents the first example for HAT reactions from
a sterically unhindered phenol without any sizable depend-
ence on specific solvation effects. The reciprocal depend-
ence of the HAT rates of 2 on the bulk viscosity points to
the intramolecular dynamics as the rate-determining step.
In conclusion, the presented experimental and theoretical
work fully confirmed our hypothesis that the solvent-effect
phenomenology of intramolecular HAT reactions of phenols
is dependent on and correlated with the donor-acceptor ge-
ometry. Intramolecular HAT reactions from phenols yield
kinetic solvent effects that are significantly correlated with
structural features; that is, individual kinetic solvent effects
are expected to prevail for every single system. This mecha-
nistic feature will be put to the test in further work.
˚
the O–O distance less than 4 A was determined to be equal
to 0.01. Using this value and the trxn from eq. [2], the esti-
mated value of kOO was calculated to be 2.5 0.1 ꢀ 107 s–1
(from eq. [1]). The resulting good agreement between the
rate constant of formation of close-contact conformations
and the experimental value of the intramolecular HAT rate
constant in ACN (see Table 2) indicates that intramolecular
dynamics is the rate-determining step of the intramolecular
HAT.
As discussed previously,39 DFT calculations suggested the
presence of a number of different conformers of 2 with
nearly degenerate energies. In particular, calculations on
dyad 2 in an ACN continuum highlighted the competition
between a conformer with a short O–O distance (favorable
for the quenching geometry) with other, more extended con-
formers of similar energy (but having unfavorable geome-
tries for HAT). This situation has been put to the test in the
current work using NMR studies and molecular-dynamics
simulations. Taking into account the error in DFT calcula-
tions, we can state that the NMR data and the theoretical
calculations are in quite good agreement. The NMR data
and the simulations both show that conformers that allow
for close contacts between the reacting centers have to over-
come significant steric constraints.
While the energetic differences between all of the side-
chain rotamers appear to be rather small according to the
NMR data (<1.0 kcal mol–1), other intrinsically non-reactive
structures of dyad 2 with only one aromatic side chain
folded over the DKP ring are certainly favored, as evident
from NMR data and the simulations (Fig. 2). Thus, signifi-
cant intramolecular rearrangements must precede the reac-
tion and can even be rate determining as soon as the
intrinsic HAT rate exceeds the rate of formation of the
quenching geometry. This condition actually seems to hold
for dyad 2. As stressed before (see Fig. 2), the long-term dy-
namics simulations indicated that the formation of close-
contact structures occurs for dyad 2. In particular, the rate
of this formation in an ACN continuum with kOO = 2.5
0.1 ꢀ 107 s–1 (computed from eqs. [1–3]) shows very good
agreement with the observed rate constants of the HAT in
this solvent (see Table 2).
Supplementary data
Supplementary data for this article are available on the
journal Web site (canjchem.nrc.ca).
Acknowledgements
Financial support from the Ministry of Science and
Higher Education (Poland) (grant No. N204 143138) and
from the Foundation for Polish Science is gratefully ac-
knowledged. This work was supported by COST Chemistry
CM0603. This is document No. NDRL 4851 from the Notre
Dame Radiation Laboratory, which is supported by the Of-
fice of Basic Energy Sciences at the United States Depart-
ment of Energy. The computation was supported by the
Interdisciplinary Centre for Mathematical and Computa-
tional Modeling of Warsaw University (G28-21). A. L. is a
scholar of Adam Mickiewicz University Foundation at 2010.
A. L. thanks the Director of the Radiation Laboratory for a
stipend that supported part of this work.
References
(1) Turro, N. J.; Ramamurth, V.; Scaiano, J. C. Modern Molecu-
lar Photochemistry of Organic Molecules; University
Science Books: Sausalito, CA, 2010.
(2) Das, P. K.; Encinas, M. V.; Scaiano, J. C. J. Am. Chem. Soc.
1981, 103 (14), 4154. doi:10.1021/ja00404a029.
´
(3) Biczok, L.; Berces, T.; Linschitz, H. J. Am. Chem. Soc.
1997, 119 (45), 11071. doi:10.1021/ja972071c.
(4) Boate, D. R.; Johnston, L.; Scaiano, J. C. Can. J. Chem.
1989, 67 (5), 927. doi:10.1139/v89-142.
Conclusions
(5) Canonica, S.; Hellrung, B.; Wirz, J. J. Phys. Chem. A 2000,
104 (6), 1226. doi:10.1021/jp9930550.
In this work we addressed the solvent effects on unimo-
lecular HAT reactions in two epimeric benzophenone–
tyrosine dyads. In agreement with previous work,29 effects
of specific solvation as well as effects of the bulk viscosity
have been identified as important controls of the intramolec-
ular reactivity. Most interestingly, the limited intramolecular
mobility imposed by the rigid peptide environment of the di-
ketopiperazine spacer translates into a strongly contrasting
phenomenology of the solvent effects between the two epi-
mers. For the anti-configured epimer 1, where both amino
(6) Cosa, G.; Scaiano, J. C. Org. Biomol. Chem. 2008, 6 (24),
4609. doi:10.1039/b810765c. PMID:19039370.
(7) Galian, R. E.; Pastor-Perez, L.; Miranda, M. A.; Perez-
Prieto, J. Chem. Eur. J. 2005, 11 (11), 3443. doi:10.1002/
chem.200401118.
´
(8) Galian, R. E.; Litwinienko, G.; Perez-Prieto, J.; Ingold, K. U.
J. Am. Chem. Soc. 2007, 129 (30), 9280. doi:10.1021/
ja071716y. PMID:17625849.
(9) Koner, A. L.; Pischel, U.; Nau, W. M. Org. Lett. 2007, 9
(15), 2899. doi:10.1021/ol071165g. PMID:17580891.
Published by NRC Research Press

278
(10) Lathioor, E. C.; Leigh, W. J. Can. J. Chem. 2001, 79 (12),
Can. J. Chem. Vol. 89, 2011
(34) Litwinienko, G.; Ingold, K. U. J. Org. Chem. 2004, 69 (18),
5888. doi:10.1021/jo049254j. PMID:15373474.
1851. doi:10.1139/cjc-79-12-1851.
(35) Litwinienko, G.; Ingold, K. U. J. Org. Chem. 2005, 70 (22),
8982. doi:10.1021/jo051474p. PMID:16238337.
(36) Litwinienko, G.; Ingold, K. U. Acc. Chem. Res. 2007, 40 (3),
222. doi:10.1021/ar0682029. PMID:17370994.
(11) Lathioor, E. C.; Leigh, W. J. Photochem. Photobiol. 2006,
82 (1), 291. doi:10.1562/2005-06-20-RA-581. PMID:
16042506.
(12) Lathioor, E. C.; Leigh, W. J.; St. Pierre, M. J. J. Am. Chem.
Soc. 1999, 121 (51), 11984. doi:10.1021/ja991207z.
(13) Leigh, W. J.; Lathioor, E. C.; St. Pierre, M. J. J. Am. Chem.
Soc. 1996, 118 (49), 12339. doi:10.1021/ja961973v.
(14) Wakasa, M.; Hayashi, H. J. Phys. Chem. 1995, 99 (47),
17074. doi:10.1021/j100047a004.
(15) Yamaji, M.; Aoyama, Y.; Furukawa, T.; Itoh, T.; Tobita, S.
Chem. Phys. Lett. 2006, 420 (1–3), 187. doi:10.1016/j.cplett.
2005.12.072.
(16) Miranda, M. A.; Lahoz, A.; Martꢀnez-Manez, R.; Bosca, F.;
Castell, J. V.; Perez-Prieto, J. J. Am. Chem. Soc. 1999, 121
(49), 11569. doi:10.1021/ja990823s.
¨
(37) Hug, G. L.; Bobrowski, K.; Pogocki, D.; Horner, G.; Marci-
niak, B. ChemPhysChem 2007, 8 (15), 2202. doi:10.1002/
cphc.200700369. PMID:17910022.
(38) Kopple, K. D.; Ohnishi, M. J. J. Am. Chem. Soc. 1969, 91
(4), 962. doi:10.1021/ja01032a029. PMID:5778278.
¨
(39) Lewandowska, A.; Hug, G. L.; Horner, G.; Carmichael, I.;
Kazmierczak, F.; Marciniak, B. Chem. Phys. Lett. 2009, 473
(4–6), 348. doi:10.1016/j.cplett.2009.03.085.
(40) Kirby, G. W.; Michael, J. J. Chem. Soc. Chem. Commun.
1971, 187, 415.
(41) Nitecki, D. E.; Halpern, B.; Westley, J. W. J. Org. Chem.
1968, 33 (2), 864. doi:10.1021/jo01266a091.
(42) Leach, A. R. Molecular Modelling: Principles and Applica-
´
´
(17) Perez-Prieto, J.; Bosca, F.; Galian, R. E.; Lahoz, A.; Dom-
ingo, L. R.; Miranda, M. A. J. Org. Chem. 2003, 68 (13),
5104. doi:10.1021/jo034225e. PMID:12816463.
tions; Longman: Harlow, England, 1996.
(43) Frenkel, D.; Smit, B. Understanding Molecular Simulation.
From Algorithms to Applications; Academic Press: San
Diego, 1996.
´
(18) Perez-Prieto, J.; Galian, R. E.; Morant-Min˜ana, M. C.; Mir-
anda, M. A. Chem. Commun. (Camb.) 2005, (25): 3180.
doi:10.1039/b500697j. PMID:15968364.
(44) Widmalm, G.; Pastor, R. W. J. Chem. Soc., Faraday Trans.
´
´
(19) Perez-Prieto, J.; Lahoz, A.; Bosca, F.; Martinez-Man˜ez, R.;
Miranda, M. A. J. Org. Chem. 2004, 69 (2), 374. doi:10.
1021/jo0356199. PMID:14725449.
1992, 88 (13), 1747. doi:10.1039/ft9928801747.
(45) Levy, R. M.; Karplus, M.; McCammon, J. A.. Chem. Phys.
Lett. 1979, 65 (1), 4. doi:10.1016/0009-2614(79)80114-1.
(46) He, S.; Scheraga, H. A. J. Chem. Phys. 1998, 108 (1), 287.
doi:10.1063/1.475379.
(47) Field, M. J. A Practical Introduction to the Simulation of
Molecular Systems; Cambridge University Press: Cambridge,
1999.
´
´
(20) Perez-Prieto, J.; Stiriba, S. E.; Bosca, F.; Lahoz, A.; Dom-
ingo, L. R.; Mourabit, F.; Monti, S.; Miranda, M. A. J. Org.
Chem. 2004, 69 (25), 8618. doi:10.1021/jo048973v. PMID:
15575737.
(21) Pischel, U.; Nau, W. M. Photochem. Photobiol. Sci. 2002, 1
(2), 141. doi:10.1039/b110108k. PMID:12659131.
(22) Pischel, U.; Patra, D.; Koner, A. L.; Nau, W. M. Photochem.
Photobiol. 2006, 82 (1), 310. doi:10.1562/2005-02-07-RA-
434. PMID:15882091.
(23) Scaiano, J. C.; McGimpsey, W. G.; Leigh, W. J.; Jakobs, S.
J. Org. Chem. 1987, 52 (20), 4540. doi:10.1021/
jo00229a021.
(48) Hinchliffe, A. Chemical Modeling. From Atoms to Liquids;
John Wiley & Sons: Chichester, UK, 1999.
(49) MacKarell, A. D., Jr.; Bashford, D.; Bellott, M.; Dunbrack,
R. L., Jr.; Evanseck, J. D.; Field, M. J.; Fischer, S.; Gao, J.;
Guo, H.; Ha, S.; Joseph-McCarthy, D.; Kuchnir, L.; Kuczera,
K.; Lau, F. T. K.; Mattos, C.; Michnick, S.; Ngo, T.;
Nguyen, D. T.; Prodhom, B.; Reiher, W. E., III; Roux, B.;
Schlenkrich, M.; Smith, J. C.; Stote, R.; Straub, J.; Wata-
(24) Seyedsayamdost, M. R.; Reece, S. Y.; Nocera, D. G.;
Stubbe, J. J. Am. Chem. Soc. 2006, 128 (5), 1569. doi:10.
1021/ja055926r. PMID:16448128.
´
nabe, M.; Wiorkiewicz-Kuczera, J.; Yin, D.; Karplus, M. J.
Phys. Chem. B 1998, 102 (18), 3586. doi:10.1021/jp973084f.
(50) Gasteiger, J.; Marsili, M. Tetrahedron 1980, 36 (22), 3219.
doi:10.1016/0040-4020(80)80168-2.
(25) Sul’timova, N. B.; Levin, P. P.; Chaikovskayab, O. N. Russ.
Chem. Bull., Int. Ed. 2005, 56, 1397.
(26) Yamaji, M.; Oshima, J.; Hidaka, M. Chem. Phys. Lett. 2009,
475 (4–6), 235. doi:10.1016/j.cplett.2009.05.063.
(27) Stubbe, J.; van der Donk, W. A. Chem. Rev. 1998, 98 (2),
705. doi:10.1021/cr9400875.
(28) Wagner, P. J.; Pabon, R.; Park, B. S.; Zand, A. R.; Ward, D.
L. J. Am. Chem. Soc. 1994, 116 (2), 589. doi:10.1021/
ja00081a020.
(51) Venable, R. M.; Zhang, Y.; Hardy, B. J.; Pastor, R. W.
Science 1993, 262 (5131), 223. doi:10.1126/science.
8211140. PMID:8211140.
(52) Hypercube. HyperChem1 Computational Chemistry; Hyper-
cube, Inc.: Gainesville, FL, 2002.
(53) Allen, M. P.; Tildesley, D. J. Computer Simulation of Li-
quids; Oxford University Press: New York, 1987.
¨
(29) Horner, G.; Hug, G. L.; Lewandowska, A.; Kazmierczak, F.;
¨
(54) Pogocki, D.; Schoneich, C. Chem. Res. Toxicol. 2002, 15
Marciniak, B. Chem. Eur. J. 2009, 15 (13), 3061. doi:10.
1002/chem.200802674.
(3), 408. doi:10.1021/tx0101550. PMID:11896689.
(55) Chandler, D. Introduction to Modern Statistical Mechanics;
Oxford University Press: Oxford, 1987.
¨
(30) Horner, G.; Lewandowska, A.; Hug, G. L.; Marciniak, B. J.
Phys. Chem.
jp901740z.
C 2009, 113 (27), 11695. doi:10.1021/
(56) Zwanzig, R.; Ailawadi, N. K. Phys. Rev. 1969, 182 (1), 280.
doi:10.1103/PhysRev.182.280.
(31) Avila, D. V.; Ingold, K. U.; Lusztyk, J.; Green, W. H.; Pro-
copio, D. R. J. Am. Chem. Soc. 1995, 117 (10), 2929. doi:10.
1021/ja00115a029.
(57) Pedzinski, T.; Markiewicz, A.; Marciniak, B. Res. Chem. In-
termed. 2009, 35 (4), 497. doi:10.1007/s11164-009-0046-4.
(58) Thomas, M. D.; Hug, G. L. Comput. Chem. 1988, 22 (6),
491. doi:10.1016/S0097-8485(98)00022-9.
(32) Banks, J. T.; Ingold, K. U.; Lusztyk, J. J. Am. Chem. Soc.
1996, 118 (28), 6790. doi:10.1021/ja961032i.
(33) Litwinienko, G.; Ingold, K. U. J. Org. Chem. 2003, 68 (9),
3433. doi:10.1021/jo026917t. PMID:12713343.
(59) Ludwig, M.; Asher, S. A. J. Am. Chem. Soc. 1988, 110 (4),
1005. doi:10.1021/ja00212a004.
(60) Laws, W. R.; Ross, J. B. A.; Wyssbrod, H. R.; Beechem, J.
Published by NRC Research Press

Lewandowska et al.
279
M.; Brand, L.; Sutherland, J. C. Biochemistry 1986, 25 (3),
599. doi:10.1021/bi00351a013.
(67) Abraham, M. H. Chem. Soc. Rev. 1993, 22 (2), 73. doi:10.
1039/cs9932200073.
¨
(61) Horner, G.; Hug, G. L.; Pogocki, D.; Filipiak, P.; Bauer, W.;
(68) CRC Handbook of Chemistry and Physics, 76th ed.; D. R.
Lide (Ed.); CRC Press: Boca Raton, 1995.
¨
Grohmann, A.; Lammermann, A.; Pedzinski, T.; Marciniak,
B. Chem. Eur. J. 2008, 14 (26), 7913. doi:10.1002/chem.
200800315.
(69) Sierra, P. S.; Tejuca, C. C.; Garcia-Blanca, F. Helv. Chim.
Acta 2005, 88, 312.
(62) Kobayashi, J.; Higashijima, T.; Miyazawa, T. Int. J. Pept.
Protein Res. 1984, 24 (1), 40. doi:10.1111/j.1399-3011.
1984.tb00925.x.
(63) Kobayashi, J.; Higashijima, T.; Sekido, S.; Miyazawa, T. Int.
J. Pept. Protein Res. 1981, 17 (4), 486. doi:10.1111/j.1399-
3011.1981.tb02018.x.
(70) Marcus, Y.; Migron, Y. J. Phys. Chem. 1991, 95, 400.
(71) Snelgrove, D. W.; Lusztyk, J.; Banks, J. T.; Mulder, P.; In-
gold, K. U. J. Am. Chem. Soc. 2001, 123 (3), 469. doi:10.
1021/ja002301e.
(72) Valgimigli, L.; Banks, J. T.; Ingold, K. U.; Lusztyk, J. J.
Am. Chem. Soc. 1995, 117 (40), 9966. doi:10.1021/
ja00145a005.
(64) Kobayashi, J.; Nagai, U. Biopolymers 1978, 17 (9), 2265.
doi:10.1002/bip.1978.360170918.
(73) Abraham, M. H.; Grellier, P. L.; Prior, D. V.; Morris, J. J.;
(65) Donzel, B.; Gauduchon, P.; Wahl, Ph. J. Am. Chem. Soc.
1974, 96 (3), 801. doi:10.1021/ja00810a027.
Taylor, P. J. J. Chem. Soc., Dalton Trans. 1990, 20, 521.
¨
(74) Lewandowska, A.; Hug, G. L.; Horner, G.; Pedzinski, T.; Fi-
(66) Chandrasekaran, R.; Lakshminarayanan, A. V.; Mohanakrishnan,
P.; Ramachandran, G. N. Biopolymers 1973, 12 (6), 1421.
doi:10.1002/bip.1973.360120615.
lipiak, P.; Marciniak, B. ChemPhysChem 2010, 11 (10),
2108. doi:10.1002/cphc.201000196. PMID:20540143.
Published by NRC Research Press