Israel Journal of Chemistry

Vol. 49 2009 pp. 163–173

DOI: 10.1560/IJC.49.2.163

Minireview:
Enzymatic H Transfers: Quantum Tunneling and Coupled Motion from Kinetic
Isotope Effect Studies

Vanja Stojković and Amnon Kohen*

Department of Chemistry, University of Iowa, Iowa City, Iowa 52242, USA

(Received 18 November 2008 and in revised form 3 December 2008)

Abstract. Many hydrogen transfer processes exhibit nonclassical behavior

due to inherent quantum mechanical properties of the hydrogen. Investigation of
various enzymes under physiological conditions indicates that hydrogen transfer
processes often show significant quantum mechanical behavior. Traditionally, this
phenomenon was treated in terms of a tunneling correction to classical or semiclas-
sical models. However, more recently, it has been observed that increasing numbers
of enzymes yield data that cannot be rationalized by tunneling correction models.
Observations such as large kinetic isotope effects (KIEs ) with unusual temperature
dependence, isotope effects on Arrhenius preexponential factors with values differ-
ent from semiclassical predicted ranges, and small temperature-independent KIEs
for processes with significant energy of activation, could only be explained with full
tunneling models for H transfer. Full tunneling models presume that the solvent or
protein fluctuations generate a reactive configuration along the heavy-atom coordi-
nate, from which the hydrogen is transferred through quantum mechanical tunnel-
ing. These models are sometimes denoted as environmentally coupled tunneling
(ECT) or Marcus-like models, and they link protein dynamics to the catalyzed H
transfer. Several enzymatic systems (dihydrofolate reductase, thymidylate synthase,
and soybean lipoxygenase) are presented as case studies of proton, hydrogen, and
hydride transfer.

Introduction tion between the ground state and the TS is presumed

Many important processes that occur in biological sys-
tems involve hydrogen transfer. These include processes
ranging from proton transfers across the cell membranes
to the functional isomerization of intermediary metabo-
lites, and H transfer between substrates and cofactors.
The physical nature of the H transfer in the enzymatic
systems has been the subject of increasing interest in
the past decade. Historically, H transfer in enzymes was
treated in terms of transition-state theory (TST), also
known as absolute reaction rate theory. According to
this theory the rate of a single reaction step is fully de-
termined by the properties of the reactants (free energy
minimum) and the transition state (TS—a free energy
saddle point or a dividing surface between reactants and
products states), and it is independent of the path taken
between these states.1,2 The distribution of the popula-
to be an equilibrium process that follows the Boltzmann
distribution. The rate of the reaction is exponentially
proportional to the height of the energy barrier, which is
closely related to the experimental energy of activation
(Ea) (eq 1).
k = Ae–Ea/RT (1)
where R is the gas constant, T is the absolute tempera-
ture, and A is the Arrhenius preexponential factor, which
also includes the transmission coefficient from the TS
to product. TST has proved useful in explaining a wide
range of chemical and biological reactions. However it
cannot always be applied to electron and/or hydrogen
transfer reactions.3–6
*Author to whom correspondence should be addressed.
E-mail: amnon-kohen@uiowa.edu
164
Fig. 1. An example of the ground-state tunneling along the re-
action coordinate. The picture depicts the probability of a parti-
cle to tunnel, with a higher probability for the lighter isotope to
tunnel. Blue and red lines represent the probability functions for
lighter and heavier isotopes, respectively. Color visible online.
Quantum mechanical (QM) effects that govern elec-
tron transfer processes have been recognized for a long
time, and arise from the particle’s wave-like properties.7
For H transfer, QM effects at the ground state (e.g., zero
point energies—ZPEs) have long been recognized as a
major component of isotope effects (both on equilib-
rium and kinetics)8,9 and the contribution of QM tunnel-
ing to TS and transmission factors has been discussed
for gas phase and organic reaction since the 1960s.10
In general terms, QM tunneling occurs when the prob-
ability of finding a particle in the reactant well overlaps
with the probability of finding the particle in the product
well (Fig. 1.). A more rigorous description would be that
the wave function for the particle at the reactant well
(Ψr) interacts (i.e., non-zero Franck–Condon integral)
with the wave function of the particle at the product
state (Ψp). In a ground state, the two new wave func-
tions (Ψr+Ψp and Ψr – Ψp) would lead to two energy
levels, a phenomenon denoted as tunneling splitting. At
a non-stable state (e.g., TS), this phenomenon will lower
the effective energy barrier (effect on Ea) and critically
affect the transmission coefficient, since the tunneling
probability is governed by different factors than that of
the classical coefficient.
Recognition of the quantum nature of the electron
transfer required the development of a new model ex-
tending beyond the explanation given by classical TST.
Rudolph Marcus won the Nobel Prize for proposing a
new model to describe electron transfer reactions, which
recognizes that the reaction coordinate is largely deter-
mined by the heavy-atom coordinates.3,4 This model
proposed that the free-energy barrier, ∆G*, depends
on λ, which is an environmental distortion that occurs
before the electron transfer (the reorganization energy),
and ∆Gº, the reaction driving force (eq 2).
DG* = (DGº + l)2 / 4l (2)
Israel Journal of Chemistry 49 2009
Table 1. Calculated de Broglie’s wavelengths for different hy-
drogen isotopes from λ = h/(2 mE)1/2, where m is particle mass,
and E is its energy (here assumed to be 10 kJ/mol)
type of isotope de Broglie’s wavelength (Å)
protium, H 0.5
deuterium, D 0.31
tritium, T 0.25
Like electron transfer reactions, hydrogen transfer
exhibits non-classical behavior due to its small mass
and consequent large uncertainty in its location and mo-
mentum. To illustrate the importance of non-classical
behavior of hydrogen, the De Broglie wavelength can
be calculated for each of the three hydrogen isotopes
(Table 1), and these can be compared to a typical dis-
tance traversed during H transfer in solution and bio-
logical systems (~0.5 Å).
Consequently, the quantum nature of hydride transfer
has often been treated as a perturbation to TST. Howev-
er, several examples discussed here indicate that the tun-
neling correction to TST is an oversimplified model.
In this review, we discuss a specific methodol-
ogy used to study the nature of the H transfer and the
potential surface, viz., kinetic isotope effects (KIEs).
Moreover, we present the development of theories for H
transfer, from the simple Bell correction to TST, to the
full-tunneling models applicable to H transfers. Lastly,
we discuss the experimental data obtained for several
model systems.
Kinetic Isotope Effects (KIEs)—an
experimental tool to study H transfer
KIEs are widely used probes for studying chemical
mechanisms due to their ability to provide information
about the reaction coordinate. The KIE is the ratio of
reaction rates of two reactants that differ only in their
isotopic composition (isotopologues). By using differ-
ent isotopologues the mass of the particle of interest
is changed without changing the electronic potential
surface, making it an exquisite tool for studying the
reaction coordinate. Two types of KIEs can be defined:
primary (1º) KIEs, which refer to an isotopic substitu-
tion of an atom that is being transferred during the reac-
tion, and a secondary (2º) KIE, where the labeled atom
is proximal to the atom being transferred. Others and we
have used 2º KIEs extensively in studies of H-tunnel-
ing;11–15 however in this review, we will focus only on 1º
KIEs. 1º hydrogen KIEs proved to be particularly inter-
esting because: (i) the mass ratio of its isotopes is larger
than for any other element, leading to large KIE values,
and (ii) the mass of hydrogen is small enough so that
 165

the quantum effects are more likely to be significant.15–18

Here we will show how KIEs have been used not only
as a tool to probe the nature of the hydride transfer in
biological systems (e.g., enzymes), but also to investi-
gate the role of the whole protein and its dynamics in the
catalysis that occurs at the active site.
Interpretation of data from KIE experiments proves
to be quite challenging due to kinetic complexity, a
feature common to enzymatic systems where the iso-
topically insensitive kinetic steps (substrate binding,
conformational changes of intermediates, product re-
lease, etc.) mask the intrinsic KIE.19,20 Several methods
have been used to unmask intrinsic KIEs from observed
KIEs and to provide direct information about the chemi-
cal step in the context of kinetic cascade that involves
other steps.19–22 The magnitude of the intrinsic KIEs
compared to that expected from a non-tunneling model, Fig. 2. Graphical representation of the semiclassical model
the temperature dependence of the intrinsic KIEs com- indicating that the difference in the energy of activation (∆Ea)
bined with the difference in the activation energy be- for H, D, and T, result from their different zero-point energies
tween isotopes, and the isotope effect on the Arrhenius (ZPE) in the ground state (GS) and transition state (TS). The
GS-ZPE is constituted by all degrees of freedom but mostly
preexponential factors (AL/AH) can be used to examine
by the C–H stretching frequency and the TS-ZPE is consti-
the contribution of tunneling. The temperature depen-
tuted by all degrees of freedom orthogonal to the reaction
dence of KIEs is very useful because it can distinguish coordinate.
between data that can be explained with the tunneling
correction to TST-models, and the data that require full Table 2. Semiclassical limits for the isotope effect on Arrhe-
tunneling models. nius preexponential factors1,10
AH/AT AH/AD AD/AT
upper limit 1.7 1.4 1.2
Semiclassical kinetic isotope effects and
lower limit 0.3 0.5 0.7
tunneling corrections
Semiclassical* models predict that the isotopic rate dif-
ference for H transfer mostly arises from the difference where W represents the energy of the particle, and k is
in the zero point energies (ZPE) between the different Boltzman’s constant. Barrier penetration occurs below
isotopes at the ground state vs. transition state (Fig. 2).10 the classical transition state, and its effect is predicted
This model predicts a primary kH/kD KIE of 6.9 for C–H to be the most significant for the lightest isotope. This
cleavage at 298 K, if we assume a stretching frequency leads both to a primary kH/kD KIE which exceeds the
of ~3000 cm–1 for C–H bond and 2200 cm–1 for C–D semiclassically predicted value of 7 for C–H cleavage,
bond.1 A similar frequency of 1800 cm–1 for C–T gives and to isotopic differences in Ea and AL/AH which are
primary kH/kT = 18 at the same temperature. smaller than limits predicted by the semiclassical model
One of the first models that addressed the phenom- (Table 2).1,15,18 This approach led to a correction to the
enon of hydrogen tunneling, observed initially in or- semiclassical model, which proved useful in explaining
ganic solutions, was the Bell model.10 According to this a range of observed phenomena, such as inflated KIEs,
model, H transfer can be described in the form of the and an inverse isotope effect on Arrhenius prefactor ra-
semiclassical TST with the addition of a multiplier term tios as measured by variable-temperature KIEs.
(Qt) that indicates the probability that a particle will Another indication for the deviation from the clas-
move through an inverse parabolic barrier (eq 3). sical nature of the H transfer is the temperature depen-
kobs = Qt * kSC dence of KIEs. According to the Arrhenius equation the
W (3) rate of the reaction is exponentially proportional to Ea
Qt = kT e kT # G]W ge - kT
1 E
and the reciprocal absolute temperature (1/T) (eq 1).
Since the KIE is the ratio of the reactions’ rates, its tem-
perature dependency will follow:
*In this context, semiclassical means that quantum-mechanical
effects at ground states are taken into consideration, e.g., ZPE, kL AL DERT
a]H - Lg

= e (4)
but, QM effects close to the TS, e.g., tunneling, are ignored. kH AH

Stojković and Kohen / Enzymatic H Transfers

166
(a)
(b)
Fig. 3. An Arrhenius plot of a hydrogen transfer applicable
for the TST-models with tunneling correction. (a) Arrhenius
plot of reaction rates for two isotopes. (b) Arrhenius plot for
their KIEs. The KIE on the Arrhenius preexponential factors
is an intercept of a tangent to a curve at the experimental tem-
perature ranges. Highlighted regions indicate three distinct
systems: (1) no tunneling contribution, where AL/AH is close
to unity; (2) moderate tunneling contribution, where AL/AH is
smaller than 1; (3) extensive tunneling contribution, where
AL/AH is larger than unity. Color visible online.
where H and L are the heavy and the light isotope, re-
spectively. For the majority of the enzymatic systems
only a narrow experimental temperature range is avail-
able (0–80 ºC), thus plots of ln(k) vs. 1/T often appear
linear (Fig. 3). At high temperature regime, the slope
of the Arrhenius plot is exponentially proportional to
∆Ea. At very low temperatures only tunneling contrib-
utes significantly to rates due to the lack of thermal
energy (Fig. 3a, region 3).23,24 Consequently, the rates
are temperature independent, and KIEs are large (over
six orders of magnitude25) and temperature independent,
reflecting the ground-state tunneling. Moreover, AL/AH,
which are intercept values of the tangents to the plot in
Fig. 3b, are expected to be inflated in comparison to the
high temperature limit (Fig. 3b, region 3). Between the
high and low temperature extremes the Arrhenius plot
of the KIEs will be curved, as the lighter isotope has a
higher probability to tunnel at the higher temperature
than the heavier one (Fig. 3b, region 2). As a result the
plot will appear very steep and AL/AH, at region 2, is
expected to be smaller than unity. To summarize, ac-
cording to this model, isotope effects on the Arrhenius
prefactors can be taken as an indication for tunneling,
where different degrees of tunneling will lead to either
inflated or deflated intercept values in comparison to the
Israel Journal of Chemistry 49 2009
high temperature limit. It is important to note that inter-
cept values close to unity do not necessarily indicate no
tunneling, because for systems with data in-between re-
gions 2 and 3, the model suggests significant tunneling
and intercept values close to unity.
TST-like models with or without tunneling correc-
tion, could, in several cases, rationalize temperature-de-
pendent KIEs,10 assuming a 1D rigid potential surface,
and large temperature-independent KIEs for systems
which showed no energy of activation for the isotopical-
ly-sensitive step. These models however fail to explain
temperature-independent small KIEs, with significant
energies of activation for the isotopically sensitive step.
One of the shortcomings of the Bell tunneling correction
and similar TST-like models is that they exclusively fo-
cus on the hydrogen reaction coordinate and ignore the
contributory motions of the heavy-atom environment.
As will be presented in the following section, more
recent models treat H-tunneling as a multidimensional
process.
During the last decade, several enzymatic systems
yielded data that could not be rationalized with the
simple Bell-correction model (Table 3). In several cases
KIEs greatly exceeded the semiclassical limit and in
most cases AL/AH surpassed unity. Moreover, some of
these systems were found to have a nonzero Ea for the
H transfer process, some with temperature-dependent
KIEs and some with temperature-independent KIEs. In
this review, we will specifically discuss three systems
that required a full tunneling (Marcus-like) model in
order to explain experimental data.
Marcus-like models
In an attempt to explain data obtained for systems where
small temperature-independent KIEs with significant
activation energies were observed, several phenomeno-
logical models were developed. These models modified
the Marcus theory for the electron transfer, by adding a
term that accounts for fluctuations of the donor–accep-
tor distance (DAD), and are thus referred to as Marcus-
like models. However, various other terms have been
coined for these kinds of models, such as “vibrationally
enhanced tunneling” model,40 “environmentally coupled
tunneling” model,41 “protein promoting vibration” mod-
el,42 and others. All these models suggest that the protein
dynamics† and H-tunneling enhance the reaction rates.
Even though these different models (e.g., Borgis and
Hynes,43,44 Kuznetzov and Ulstrup,45 Sutcliffe and Scrut-
ton,40 Warshel,46 Caratzoulas and Schwartz,42 Nagel and
Klinman47) originate from different basic principles,
†
By dynamics, we refer to any motion in the system regardless
if it is in thermal equilibrium with the environment or not.
 167

Table 3. Enzymatic systems with properties outside the limits predicted by the “tunneling correction” models
temperature
enzyme type of transfer dependence kH/kD AH/AD ref
aromatic amine dehydrogenase proton no 55 26
methylamine dehydrogenase proton no 16.8 13.3 27
trimethylamine dehydrogenase proton (suspected) no 4.6 (for pH > 9.5) 7.8 28
sarcosine oxidase proton (suspected) no 7.3 5.8 29
pentaerythritol tetranitrate reductase hydride no ~4.1 4.1 30
ethanolamine ammonia lyase H radical 4.4–4.7 31
glutamate mutase H radical 2.4 32
soybean lipooxygenase H radical yes 81 18 33
soybean lipooxygenase, mutants H radical yes 93–112 4–0.12 33
dihydrofolate reductase (DHFR) hydride no 3.5 4.0 34
G121V-M42W DHFR hydride yes 3.5 0.04 35
primitive DHFR hydride yes 3.5 36
thymidylate synthase hydride no 3.7 37
thymidylate synthase proton yes, under 20 ºC 2.2 38
alcohol dehydrogenase (ADH) hydride yes, under 30 ºC 5.6 (kH/kT) 4.3 (>30 ºC) (AH/AT) 39

they all address a similar phenomenological model,

and successfully differentiate between the temperature
dependence on the rate and temperature dependence of
the KIEs.5,6 Even though these models often use dif-
ferent terminology, they all (i) treat hydrogen quantum
mechanically throughout the reaction coordinate and (ii)
consider the reaction coordinate to mostly represent the
dynamics of the environment surrounding the hydrogen
that is being transferred.48 Moreover, they all consider
two requirements for efficient tunneling: degeneracy of
the reactant and product energy levels and narrow bar-
rier width.5,6 A general description for these models can
be expressed by the following equation: Fig. 4. Illustration of the Marcus-like model as expressed
r1 in eq 5, where the two orthogonal coordinates represent:
“Marcus term”, the environmental energy parabolas for the
k = Ce -]DGc + mg
2/]4mRT g
# e F]mg e - E F]mg
/k b T
dDAD (5)
reactant (R) and product (P) states, and the “gating term”,
fluctuation of the DAD along the hydrogen potential surface.
r0

An illustration of a “Marcus-like” model is presented
in Fig. 4.
In this model, the rate of H transfer is determined
by the isotope-independent preexponential term (C =
constant), a Marcus term (first exponential), a Franck–
Condon term (F(m)), and a “gating term” (EF(m)) that
describes the temperature and mass dependence of the
DAD fluctuations. The last two terms are integrated
for all the conformations sampled by the DAD. The
Marcus term is isotopically insensitive, and represents
the heavy-atom reorganization‡ that is required to oc-
‡
Reorganization refers to heavy atom motions that bring the
system from a ground state of the chemical step to a tunneling
ready conformation. The term “preorganization” is sometimes
used to describe the protein motions, of both distal and
proximal residues, that bring the system to the ground state of
the H transfer reaction.
The Franck–Condon nuclear overlap is presented as a red cur-
vature indicating the probability of finding the particle at the
reactant and product state. Figure was modified from ref 36.
Color visible online.
cur before significant tunneling can take place (i.e.,
the reorganization of the donor and acceptor toward a
“tunneling ready” conformation).47,49 The Franck–Con-
don nuclear overlap along the hydrogen coordinate,
which is the integrated hydrogen tunneling probability
(Fig. 1), is an isotopically sensitive but temperature-in-
dependent term. It depends on the DAD, which is both
isotopically and temperature dependent. In eq 5, the
last exponential term describes the relative fluctuations
of DAD, and is graphically illustrated in Fig. 5 as the
“gating term” coordinate. According to this model, the
temperature-independent KIEs indicate a system with
a perfectly reorganized reaction coordinate, where the

Stojković and Kohen / Enzymatic H Transfers

168
Fig. 5. Comparison of the Arrhenius plots of intrinsic H/T
KIEs of wt (red), G121V (green), M42W (blue), and G121V-
M42W (black). Figure was reproduced with permission from
ref 35. Color visible online.
average DAD is ideal for tunneling so thermal activa-
tion of the fluctuations of the DAD does not change
the Franck–Condon integral. In other cases, where the
reorganization is not perfect, sampling of DAD results
in temperature-dependent KIEs.
Several enzymatic systems (dihydrofolate reductase,
thymidylate synthase, and soybean lipoxygenase) and
their mutants are examples of proton, hydrogen, and hy-
dride transfer systems, where Marcus-like models were
implemented in order to explain experimental data.
Experimental examples using intrinsic
1º KIE, isotope effects on Arrhenius
exponential factors, and temperature
dependence of KIEs
Dihydrofolate Reductase
Dihydrofolate reductase (DHFR) catalyzes the re-
duction of 7,8-dihydrofolate (DHF) to 5,6,7,8-tetrahy-
drofolate (THF), with the stereospecific transfer of a
hydride from the pro-R C4 position of NADPH to the si
face of C6 of the pterin ring50 (Scheme 1). DHFR is one
Scheme 1. Mechanism for the hydride transfer in DHFR,
where R = adenine dinucleotide 2¢phosphate and R¢ = (p-ami-
nobenozoyl) glutamate.
Israel Journal of Chemistry 49 2009
of the preferred systems for structure–dynamics–func-
tion studies due to its small size, flexibility, and the
simple chemical transformation it catalyzes. Moreover,
due to its pivotal role in nucleotide biosynthesis in many
organisms, DHFR is the target for many antibiotic and
chemotherapeutic agents51 and an excellent platform
for genetic and evolutionary analysis.52 Consequently,
DHFR has been studied by different experimental meth-
ods in order to gain understanding of its structure,53 dy-
namics,54 and kinetics.55
The coupling of the enzyme’s structure and dynamics
to the chemical step has been studied in wild type (wt)
DHFR as well as for the three remote mutants (G121V,
M42W, and their double mutant G121V-M42W).34,35,56,57
These residues are about 17 Å from the active site and
25 Å from each other, yet, QM/MM simulation sug-
gested that they are dynamically coupled to each other
and to the reaction coordinate.58 The intrinsic 1º H/D
KIE and H/T KIE have been obtained for these enzymes
over a temperature range of 5–45 ºC (Fig. 5). Compara-
tive isotope effects on Arrhenius preexponential factors
together with ∆Ea are presented in Table 4. The tem-
perature-independent KIEs, with large AH/AD, and non-
zero Ea values for the wtDHFR (Fig. 5, red line, color
visible online) have been rationalized in the context of
the Marcus-like models with ideal rearrangement of the
potential surface.34 The average DAD in this system ap-
pears to be ideal for tunneling.
Two single mutants (G121V and M42W) showed
slightly inflated KIEs with weak temperature depen-
dences (Fig. 5, green and blue lines, respectively. Color
visible online.) These observations suggest that both
mutants have less perfectly reorganized reaction coordi-
nates in comparison to wtDHFR. Therefore, weak tem-
perature-dependent KIEs suggest that single mutants
required some thermally activated fluctuations in order
to reduce the DAD and to promote tunneling. In con-
trast to both single mutants, the double mutant (G121V-
M42W) had a very steep temperature dependence of the
KIEs (Fig. 5, black line, color visible online), implying
poor rearrangement, and an average DAD that is too
long to enable efficient tunneling. Consequently, essen-
tial thermally activated fluctuations of the DAD in the
double mutant lead to a large temperature dependence.
These findings indicated that the remote single mutants
mostly affected the prearrangement of the system, re-
sulting in the observed slower reaction (Table 4, first
row). However, each single mutation had a small effect
on the rearrangement of the system (weak temperature
dependence of KIEs). The double mutation, on the other
hand, resulted in a significant change in the nature of
the hydride transfer leading to a substantial temperature
dependence of KIEs, which indicates a poor reorganiza-
 169

Table 4. Comparative KIEs on Arrhenius preexponential factors for wtDHFR and its mutants.
Reproduced using data published in ref 35
parameters wt G121V M42W G121V-M42W
kH* (s )
–1
228 ± 8 1.4 ± 0.2 5.6 ± 0.4 0.03 ± 0.005
AH/AT 7.0 ± 1.5 7.4 ± 1.6 2.8 ± 0.2 0.1 ± 0.1
AH/AD 3.5 ± 0.5 4.7 ± 1.5 2.1 ± 0.2 0.04 ± 0.03
AD/AT 1.70 ± 0.14 1.70 ± 0.07 1.35 ± 0.05 0.25 ± 0.09
∆Ea(T–H), kcal/mol –0.1 ± 0.2 0.23 ± 0.03 0.58 ± 0.04 3.6 ± 0.3
*Pre-steady-state rates of H transfer at 25 ºC and pH 7.

tion of the reaction coordinate. The synergistic effect of
these two remote residues on the active site chemistry
supports the idea of the long-range dynamically coupled
network previously suggested by hybrid quantum/clas-
sical molecular dynamics simulations.58
Another aspect of DHFR catalysis that was investi-
gated was the possible role of evolution in tuning en-
zyme dynamics to enhance tunneling.36 The nature of
the H transfer in primitive R67 DHFR was compared to
that of a mature, highly evolved enzyme, chromosomal
DHFR (cDHFR). The R67 DHFR is genetically unrelat-
ed to the cDHFR; it is an R-plasmid-encoded DHFR and
is present in some bacterial strains59 that are resistant to
the antibiotic drug trimethoprim (TMP—a cDHFR pi-
comolar inhibitor).59 Even though it catalyzes the same
reaction as the cDHFR, they share neither sequence nor
structural similarities (Fig. 6). Furthermore, these two
enzymes exhibit different degrees of flexibility. The ma-
ture cDHFR is flexible and exhibits motions on different
timescales that are crucial for its proper function.60,61 On
the other hand, R67-DHFR shows rigidity in its motions
on all the timescales examined.62 Therefore, these two
systems are a good platform for examination of the role
Fig. 6. Top: Structures of cDHFR (left) and R67-DHFR
(right). Bottom: the reverse images for each active site em-
phasizing the different orientation of the donor (magenta) and
acceptor (green).59 Figure was reproduced with permission
from Wong et al., PNAS 2006, National Academy of Sciences,
U.S.A. Color visible online.
of evolution in optimizing enzyme dynamics to enhance
tunneling. The comparative data indicated that unlike
cDHFR, for which intrinsic KIEs are nearly temperature
independent (∆Ea = –0.1 ± 0.3 kcal mol–1)34, the R67-
DHFR has temperature-dependent KIEs (∆Ea = 0.87 ±
0.03 kcal mol–1). Moreover, AH/AT for cDHFR is signifi-
cantly larger than the semiclassical value (AH/AT = 7.4 ±
4.0)34, while the isotope effect on AH/AT for R67-DHFR
is much smaller and falls in the semiclassical limit (AH/
AT = 1.36 ± 0.07 kcal mol–1).
Interpretation of these results depends on the model
used. According to TST-models with tunneling correc-
tion, the above data would suggest extensive tunneling
in cDHFR, and no tunneling in the primitive DHFR.
Alternatively, Marcus-like models would suggest that
the reaction coordinate in the highly developed enzyme
is perfectly reorganized (temperature-independent
KIEs), and that the primitive enzyme exhibits a poor
reorganization of the reaction coordinate (temperature-
dependent KIEs). These findings are in accordance with
the hypothesis that enzymes evolved to optimize their
reaction coordinate for efficient tunneling.
Thymidylate Synthase (TSase)
Thymidylate synthase catalyzes the reductive meth-
ylation of 2¢-deoxyuridine-5¢-monophosphate (dUMP)
to 2¢-deoxythymidine-5¢-monophosphate (dTMP)
(Scheme 2). The cofactor N5,N10-methylene-5,6,7,8-
tetrahydrofolate (MTHF) serves both as a methy-
lene and hydride donor, yielding 7,8-dihydrofolate
(DHF).63 The hydride transfer step (Step 5 in Scheme
2) was studied using 1º H/T and D/T KIEs, and tem-
perature-independent intrinsic H/T KIEs of 7 were
obtained.37 For the wtTSase, the isotope effects on Ar-
rhenius preexponential factors were much higher than
the semiclassical limit (AH/AT= 6.8 ± 2.8 and AD/AT =
1.9 ± 0.25), and the reaction exhibited nonzero energy
of activation (Ea = 4.0 ± 0.01 kcal/mol). Temperature-
independent KIEs with a significant energy of activation
suggest a perfectly reorganized potential surface for the
hydride transfer, which is in accordance with the envi-
ronmentally coupled tunneling model.

Stojković and Kohen / Enzymatic H Transfers

170

Scheme 2. The chemical mechanism for the TSase catalyzed reaction. The transferred methylene group is in purple, the nu-
cleophilic Cys in yellow, and the hydride in green. R = 2¢-deoxyribose-5¢-phosphate. R¢ = (p-aminobenzoyl)glutamate. Color
visible online.
The method established for the wtTSase was then
used to resolve a controversy regarding the mechanism
of TSase. Until recently, two distinct mechanisms were
proposed for step 5 (Scheme 2). Schultz and cowork-
ers64 studied unnatural amino acid mutations at the
enzyme’s active site, and suggested a two-step radical
mechanism where an electron is first transferred from
THF to form a tryptophan (W80)-stabilized cation radi-
cal, and then an H-radical is transferred from this cation
radical to form product dTMP (Scheme 3, bottom path).
Stroud and coworkers,65,66 on the other hand, studied
crystal structure of wtTSase and several W80 mutants,
and suggested a one-step hydride transfer mechanism
(Scheme 3, top path), which implied that W80 could
not stabilize the H4folate cation radical. Examination
of these mechanisms was performed by comparing the
nature of the H transfer between wtTSase and W80M-
TSase mutant, which cannot stabilize a cation radical.67
It was found that the nature of the H transfer and its
coupling to the environment did not change upon muta- Scheme 3. Two proposed mechanisms for the H transfer from
tion, and that the three-orders-of-magnitude rate reduc- C6 of THF to the exocyclic methylene. (Top) A one-step
tion of this mutant was a result of a smaller percentage hydride transfer mechanism. (Bottom) The two-step radical
of reactive conformations. These findings supported mechanism. Full arrows symbolize transfer of a pair of elec-
the hydride transfer mechanism67 and demonstrated the trons, and half arrows symbolize transfer of a single electron.
ability of these methods to examine the physical nature Figure was reproduced with permission from ref 67.
of H transfer, its coupling to dynamics, and the role of
active site residues in catalysis.
coupled electron transfer is greater than 80 at 298 K.
Soybean Lipoxygenase-1 (SLO) The reaction has a fairly small activation energy (Ea =
Soybean lipooxygenase-1 is a non-heme Fe-depen- 2.1 kcal/mol), and the isotope effect on Arrhenius pref-
dent enzyme that catalyzes the conversion of linoleic actors is close to 20 (Table 2).68–70 An Eyring analysis of
acid to 13(S)-hydroperoxy-9(Z),11(E)-octadecanoic the kinetic data for the wtSLO implies that the reaction
acid (Scheme 4). It exhibits one of the largest ever barrier is largely entropic (∆H‡ = 1.5 kcal/mol, –T∆S‡ =
measured KIEs, where the H/D KIE for the proton 12.8 kcal/mol).70 Significant KIEs indicate a very large

Israel Journal of Chemistry 49 2009


Scheme 4. Consensus mechanism for SLO-1, in which active site Fe(III)-OH abstracts a proton and electron from the C-11 po-
sition of substrate. The structures 2a, 2b, and 2c indicate the three resonance forms, in which the single electron density resides
at positions 11, 13, and 9, respectively (where R = C5H11, and R¢ = C7H14COOH). Figure was reproduced with permission from
Meyer et al., PNAS 2008, National Academy of Sciences, U.S.A.
inherent chemical barrier, suggesting that the transfer of
H-radical occurs mainly through tunneling. However,
semiclassical theory cannot account for such an enor-
mous KIE with inflated AH/AD.
Interestingly, a distal mutant, I553A (I553 is one he-
lix turn away from the active site residue L546), has a
1º /D KIE and Ea parameters similar to wtSLO, however
its AH/AD ratio becomes inverted and its KIEs become
temperature dependent.70 These parameters (kH/kD, Ea,
and AH/AD) could only be explained with the Marcus-
like models suggesting that wtSLO has an active site
structure that is well reorganized to facilitate hydrogen
tunneling, and that the remote I553A mutation perturbs
the structural and dynamic elements that support that
process. The mutation seems to increase the average
DAD. Consequently, the mutant requires thermally ac-
tivated motion in order to facilitate tunneling leading to
temperature-dependent KIEs. Klinman and coworkers
further supported this conclusion while studying a series
of I553 mutants.71
Summary
The focus of this field has drastically changed in the past
ten years. The most frequent question in the late 1990s
concerned the prevalence of H-tunneling in C–H acti-
vation reactions, and if it contributes more to enzyme
catalyzed H transfers relative to uncatalyzed ones. In
recent years it was recognized that H-tunneling contrib-
utes to both catalyzed and uncatalyzed reactions. The
current assessment is that H-tunneling is always part
of the H transfer reactions, whether reactions occur in
the gas phase or the solid phase, solutions or enzymes.
Today most investigators are examining if the dynam-
171
ics of the protein contribute to the enhancement of the
catalyzed chemical transformation. One of the working
hypotheses in this field is that tunneling in enzymatic
reactions occurs from a better pre-/re-arranged con-
formation along the reaction coordinate than a similar
reaction in solution or in primitive enzymes. Therefore,
H-tunneling is used as a probe for the organization of
the system, and to investigate the dynamic contribution
to catalysis.
In this review, we presented several examples of the
way the nature of H-tunneling has been examined. Data
obtained through examination of temperature depen-
dence of KIEs for enzymatic systems such as DHFR,
TSase, and SLO could not be rationalized using the
semiclassical models with or without tunneling correc-
tion. Better analysis and rationale was achieved with
Marcus-like models. With the development of these new
theoretical models we are at a point where we can ex-
amine questions like whether or not enzymes evolved to
optimize their reaction coordinate for tunneling, as well
as other questions that could have a significant impact
on biomimetic catalysis design.
Acknowledgments. This work was supported by NIH R01
GM65368 and NSF CHE- 0715448.
References and notes
(1) Melander, L.; Saunders, W.H. Reaction Rates of Iso-
topic Molecules, 4th ed.; Krieger, R.E.: Malabar, FL,
1987.
(2) Kraut, J. Science 1988, 242, 533–540.
(3) Marcus, R.A.; Sutin, N. Biochem. Biophys. Acta 1985,
811, 265–322.
(4) Marcus, R.A. J. Phys. Chem. B. 2007, 111, 6643–6654.
(5) Kohen, A. In Isotope Effects in Chemistry and Biology;
Stojković and Kohen / Enzymatic H Transfers
172
Kohen, A.; Limbach, H.H., Eds.; Taylor & Francis,
CRC Press: Boca Raton, FL, 2006; Ch. 28, pp 743–
764.
(6) Kohen, A. In Hydrogen Transfer Reactions; Hynes,
J.T.; Limbach, H.H.; Klinman, J.P.; Schowen, R.L.,
Eds.; Wiley-VCH: Weinheim, 2006; Vol. 4, Ch. II-12,
pp 1311–1340.
(7) Steinfeld, J.I.; Francisco, J.S.; Hase, W.L. Chemical
Kinetics and Dynamics, 2nd ed.; Prentice Hall: Upper
Saddle River, NJ, 1998.
(8) Bigeleisen, J.; Mayer, M.G. J. Chem. Phys. 1947, 15,
261–267.
(9) Bigeleisen, J. J. Chem. Phys. 1955, 23, 2264–2267.
(10) Bell, R.P. The Tunnel Effect in Chemistry; Chapman &
Hall: London & New York, 1980.
(11) Cleland, W.W. Methods Enzymol. 1980, 64, 104–125.
(12) Huskey, W.P.; Schowen, R.L. J. Am. Chem. Soc. 1983,
105, 5704–5706.
(13) Kohen, A.; Jensen, J.H. J. Am. Chem. Soc. 2002, 124,
3858–3864.
(14) Cha, Y.; Murray, C.J.; Klinman, J.P. Science 1989, 243,
1325–1330.
(15) Kohen, A. Prog. React. Kinet. Mech. 2003, 28, 119–
156.
(16) Klinman, J.P. Trends Biochem. Sci. 1989, 14, 368.
(17) Kohen, A.; Klinman, J.P. Acc. Chem. Res. 1998, 31,
397–404.
(18) Kohen, A.; Klinman, J.P. Chem. Biol. 1999, 6, R191–
198.
(19) Cleland, W.W. In Enzyme Mechanism from Isotope
Effects; Cook, P.F., Ed.; CRC Press: Boca Raton, FL,
1991, pp 247–268.
(20) Northrop, D.B. In Enzyme Mechanism from Isotope Ef-
fects; Cook, P.F., Ed.; CRC Press: Boca Raton, FL, 1991,
pp 181–202.
(21) Northrop, D.B. Annu. Rev. Biochem. 1981, 50, 103–131.
(22) Cleland, W.W. In Isotope Effects in Chemistry and
Biology; Kohen, A.; Limbach, H.H., Eds.; Taylor &
Francis, CRC Press: Boca Raton, FL, 2006; Ch. 37, pp
915–930.
(23) Kim, T.; Kreevoy, M.M. J. Am. Chem. Soc. 1992, 114,
7116.
(24) Jonsson, T.; Glickman, M.H.; Sun, S.J.; Klinman, J.P. J.
Am. Chem. Soc. 1996, 118, 10319–10320.
(25) Moiseyev, N.; Rucker, J.; Glickman, M.H. J. Am. Chem.
Soc. 1997, 119, 3853–3860.
(26) Masgrau, L.; Roujeinikova, A.; Johannissen, L.O.; Hothi,
P.; Basran, J.; Ranaghan, K.E.; Mulholland, A.J.; Sut-
cliffe, M.J.; Scrutton, N.S.; Leys, D. Science 2006, 312,
237–241.
(27) Basran, J.; Sutcliffe, M.J.; Scrutton, N.S. Biochemistry
1999, 38, 3218–3222.
(28) Basran, J.; Sutcliffe, M.J.; Scrutton, N.S. J. Biol. Chem.
2001, 276, 24581–24587.
(29) Harris, R.J.; Meskys, R.; Sutcliffe, M.J.; Scrutton, N.S.
Biochemistry 2000, 39, 1189–1198.
(30) Basran, J.; Harris, R.J.; Sutcliffe, M.J.; Scrutton, N.S. J.
Biol. Chem. 2003, 278, 43973–43982.
Israel Journal of Chemistry 49 2009
(31) Bandarian, V.; Reed, G.H. Biochemistry 2000, 39,
12069–12075.
(32) Cheng, M.C.; Marsh, E.N. Biochemistry 2005, 44,
2868–2691.
(33) Knapp, M.J.; Klinman, J.P. J. Am. Chem. Soc. 2002,
124, 3865–3874.
(34) Sikorski, R.S.; Wang, L.; Markham, K.A.; Rajagopalan,
P.T.R.; Benkovic, S.J.; Kohen, A. J. Am. Chem. Soc.
2004, 126, 4778–4779.
(35) Wang, L.; Goodey, N.M.; Benkovic, S.J.; Kohen, A. Proc.
Natl. Acad. Sci. U.S.A. 2006, 103, 15753–15758.
(36) Yahashiri, A.; Howell, E.E.; Kohen, A. ChemPhysChem
2008, 9, 980–982.
(37) Agrawal, N.; Hong, B.; Mihai, C.; Kohen, A. Biochemistry
2004, 43, 1998–2006.
(38) Hong, B.; Maley, F.; Kohen, A. Biochemistry 2007, 46,
14188–14197.
(39) Kohen, A.; Cannio, R.; Bartolucci, S.; Klinman, J.P.
Nature 1999, 399, 496–499.
(40) Sutcliffe, M.J.; Scrutton, N.S. Eur. J. Biochem. 2002,
269, 3096–3102.
(41) Francisco, W.A.; Knapp, M.J.; Blackburn, N.J.; Klin-
man, J.P. J. Am. Chem. Soc. 2002, 124, 8194–8195.
(42) Caratzoulas, S.; Schwartz, S.C. J. Chem. Phys. 2001,
114, 2910–2918.
(43) Borgis, D.; Hynes, J.T. J. Chem. Phys. 1991, 94, 3619–
3628.
(44) Borgis, D.; Hynes, J.T. Chem. Phys. 1993, 170, 315–
346.
(45) Kuznetsov, A.M.; Ulstrup, J. Can. J. Chem. 1999, 77,
1085–1096.
(46) Warshel, A. Proc. Natl. Acad. Sci. USA 1984, 81, 444–
448.
(47) Nagel, Z.D.; Klinman, J.P. Chem. Rev. 2006, 106,
3095–3118.
(48) Kiefer, P.M.; Hynes, J.T. In Isotope Effects in Chemistry
and Biology; Kohen, A.; Limbach, H.H., Eds.; Taylor &
Francis, CRC Press: Boca Raton, FL, 2006; Ch. 21, pp
549–578.
(49) Marcus, R.A. J. Phys. Chem. B 2007, 111, 6643–6654.
(50) Miller, G.P.; Benkovic, S.J. Chem. Biol. 1998, 5, R105–
R113.
(51) Warlick, C.A.; Sweeney, C.L.; McIvor, R.S. Biochem.
Pharmacol. 2000, 59, 141–151.
(52) Lee, J.; Natarajan, M.; Nashine, V.C.; Socolich, M.; Vo,
T.; Russ, W.P.; Benkovic, S.J.; Ranganathan, R. Science
2008, 322, 438–442.
(53) Sawaya, M.R.; Kraut, J. Biochemistry 1997, 36, 586–603.
(54) Boehr D.D.; M.D.; Dyson H.J.; Wright, P.E. Science
2006, 313, 1638–1642.
(55) Fierke, C.A.; Johnson, K.A.; Benkovic, S.J. Biochemistry
1987, 26, 4085–4092.
(56) Wang, L.; Goodey, N.M.; Benkovic, S.J.; Kohen, A. Phi-
los. Trans. R. Soc. London B 2006, 361, 1307–1315.
(57) Wang, L.; Tharp, S.; Selzer, T.; Benkovic, S.J.; Kohen, A.
Biochemistry 2006, 45, 1383–1392.
(58) Wong, K.F.; Selzer, T.; Benkovic, S.J.; Hammes-Schiffer,
S. Proc. Natl. Acad. Sci. U.S.A. 2005, 102, 6807–6812.

(59) Howell, E.E. ChemBioChem 2005, 6, 590–600.
(60) Sawaya, M.R.; Kraut, J. Biochemistry 1997, 36, 586–
603.
(61) Boehr, D.D.; McElheny, D.; Dyson, H.J.; Wright, P.E.
Science 2006, 313, 1638–1642.
(62) Pitcher, W.H.R.; DeRose, E.F.; Mueller, G.A.; Howell,
E.E.; London, R.E. Biochemistry 2003, 42, 11150–11160.
(63) Carreras, C.W.; Santi, D.V. Annu. Rev. Biochem. 1995,
64, 721–762.
(64) Barrett, J.E.; Lucero, C.M.; Schultz, P.G. J. Am. Chem.
Soc. 1999, 121, 7965–7966.
(65) Stroud, R.M.; Finer-Moore, J.S. Biochemistry 2003, 42,
239–247.
173
(66) Finer-Moore, J.S.; Santi, D.V.; Stroud, R.M. Biochem-
istry 2003, 42, 248–256.
(67) Hong, B.; Haddad, M.; Maley, F.; Jensen, J.H.; Kohen,
A. J. Am. Chem. Soc. 2006, 128, 5636–5637.
(68) Jonsson, T.; Glickman, M.H.; Sun, S.; Klinman, J.P. J.
Am. Chem. Soc. 1996, 118, 10319–10320.
(69) Rickert, K.W.; Klinman, J.P. Biochemistry 1999, 38,
12218–12228.
(70) Knapp, M.J., Rickert, K.; Klinman, J.P. J. Am. Chem.
Soc. 2002, 124, 3865–3874.
(71) Meyer, M.P.; Tomchick, D.R.; Klinman, J.P. Proc.
Natl. Acad. Sci., USA 2008, 105, 1146–1151.
Stojković and Kohen / Enzymatic H Transfers