Light-enhanced catalysis by pyridoxal phosphate-dependent aspartate aminotransferase

Article abstract of DOI:10.1021/ja107054x

The mechanisms of pyridoxal 5′-phosphate (PLP)-dependent enzymes require substrates to form covalent "external aldimine" intermediates, which absorb light strongly between 410 and 430 nm. Aspartate aminotransferase (AAT) is a prototypical PLP-dependent enzyme that catalyzes the reversible interconversion of aspartate and α-ketoglutarate with oxalacetate and glutamate. From kinetic isotope effects studies, it is known that deprotonation of the aspartate external aldimine C_α-H bond to give a carbanionic quinonoid intermediate is partially rate limiting in the thermal AAT reaction. We show that excitation of the 430-nm external aldimine absorption band increases the steady-state catalytic activity of AAT, which is attributed to the photoenhancement of C_α-H deprotonation on the basis of studies with Schiff bases in solution. Blue light (250 mW) illumination gives an observed 2.3-fold rate enhancement for WT AAT activity, a 530-fold enhancement for the inactive K258A mutant, and a 58600-fold enhancement for the PLP-Asp Schiff base in water. These different levels of enhancement correlate with the intrinsic reactivities of the C_α-H bond in the different environments, with the less reactive Schiff bases exhibiting greater enhancement. Time-resolved spectroscopy, ranging from femtoseconds to minutes, was used to investigate the nature of the photoactivation of C _α-H bond cleavage in PLP-amino acid Schiff bases both in water and bound to AAT. Unlike the thermal pathway, the photoactivation pathway involves a triplet state with a C_α-H pK_a that is estimated to be between 11 and 19 units lower than the ground state for the PLP-Val Schiff base in water.

Full text of DOI:10.1021/ja107054x

Published on Web 11/08/2010
Light-Enhanced Catalysis by Pyridoxal Phosphate-Dependent
Aspartate Aminotransferase
Melissa P. Hill, Elizabeth C. Carroll, Mai C. Vang, Trevor A. Addington,
Michael D. Toney,* and Delmar S. Larsen*
Department of Chemistry, UniVersity of California, DaVis, One Shields AVenue, DaVis,
California 95616, United States
Received August 6, 2010; E-mail: dlarsen@ucdavis.edu; mdtoney@ucdavis.edu
Abstract: The mechanisms of pyridoxal 5′-phosphate (PLP)-dependent enzymes require substrates to form
covalent “external aldimine” intermediates, which absorb light strongly between 410 and 430 nm. Aspartate
aminotransferase (AAT) is a prototypical PLP-dependent enzyme that catalyzes the reversible intercon-
version of aspartate and R-ketoglutarate with oxalacetate and glutamate. From kinetic isotope effects studies,
it is known that deprotonation of the aspartate external aldimine CR-H bond to give a carbanionic quinonoid
intermediate is partially rate limiting in the thermal AAT reaction. We show that excitation of the 430-nm
external aldimine absorption band increases the steady-state catalytic activity of AAT, which is attributed
to the photoenhancement of CR-H deprotonation on the basis of studies with Schiff bases in solution.
Blue light (250 mW) illumination gives an observed 2.3-fold rate enhancement for WT AAT activity, a 530-
fold enhancement for the inactive K258A mutant, and a 58600-fold enhancement for the PLP-Asp Schiff
base in water. These different levels of enhancement correlate with the intrinsic reactivities of the CR-H
bond in the different environments, with the less reactive Schiff bases exhibiting greater enhancement.
Time-resolved spectroscopy, ranging from femtoseconds to minutes, was used to investigate the nature of
the photoactivation of CR-H bond cleavage in PLP-amino acid Schiff bases both in water and bound to
AAT. Unlike the thermal pathway, the photoactivation pathway involves a triplet state with a CR-H pKa that
is estimated to be between 11 and 19 units lower than the ground state for the PLP-Val Schiff base in
water.
1. Introduction
broader biological role might this play for organisms (e.g., plants
and bacteria) that grow under solar radiation?
Although light-activated enzymes such as DNA photolyase1
and protochlorophyllide oxidoreductase (POR)2 do exist, most
enzymes do not require the absorption of light for catalytic
activity. Even though many enzymes require cofactors (e.g.,
hemes, flavins, metal centers, etc.) that absorb light in the
ultraviolet (UV) and visible regions of the spectrum. In some
cases, light can initiate biological activity. For example, CO
dissociation can be initiated in hemoglobin and myoglobin by
blue light absorption,3-5 as can Co-C bond cleavage in
adenosyl cobalamin (vitamin B12).6 This raises the question of
whether light excitation can generally affect the catalytic activity
of chromophoric enzymes by accelerating cofactor-dependent
rate-limiting steps. If so, what molecular mechanisms can couple
light absorption to enzymatic activity, can light enhancement
be a useful tool to explore enzymatic mechanisms, and what
Pyridoxal 5′-phosphate (PLP; vitamin B6) is a chromophoric
cofactor required for catalytic activity by a wide variety of
enzymes.7,8 It typically exhibits absorption bands at ∼430,
∼360, and ∼330 nm with extinction coefficients of ∼5000 M-1
cm-1, depending on its environment (Figure 1). While PLP
enzymes are thermally activated in ViVo, it has been reported
that some PLP enzymes can be activated by UV light.8-11
Previous studies9 with aspartate aminotransferase (AAT) have
suggested that the carbanionic quinonoid intermediate, which
is on the thermal reaction pathway of most PLP enzymes, is
photogenerated by UV laser excitation. 5-Hydroxytryptophan
decarboxylase, which produces serotonin, has also been reported
to have increased steady-state catalytic activity when exposed
to light.10 To date, no detailed mechanism has been proffered
for these interesting observations.
(1) Sancar, A. Chem. ReV. 2003, 103, 2203–2237.
(2) Heyes, D. J.; Hunter, C. N. Trends. Biochem. Sci. 2005, 30, 642–
649.
(7) Toney, M. D. Arch. Biochem. Biophys. 2005, 433, 279–287.
(8) Kallen, R. G.; Korpela, T.; Martell, A. E.; Matsushima, Y.; Metzler,
C. M.; Metzler, D. E.; Morozov, Y. V.; Ralston, I. M.; Savin, F. A.;
Torchinsky, Y. M.; Ueno, H. In Transaminases; Christian, P., Metzler,
D. E., Eds.; John Wiley and Sons: New York, 1985.
(9) Cornish, T. J.; Ledbetter, J. W. IEEE J. Quantum Electron. 1984, 20,
1375–1379.
(3) Cao, W. X.; Ye, X.; Sjodin, T.; Christian, J. F.; Demidov, A. A.;
Berezhna, S.; Wang, W.; Barrick, D.; Sage, J. T.; Champion, P. M.
Biochemistry 2004, 43, 11109–11117.
(4) Sage, J. T.; Morikis, D.; Champion, P. M. Biochemistry 1991, 30,
1227–1237.
(5) Tian, W. D.; Sage, J. T.; Champion, P. M. J. Mol. Biol. 1993, 233,
155–166.
(10) Fraikin, G. Y.; Strakhovskaya, M. G.; Ivanova, E. V.; Rubin, A. B.
Photochem. Photobiol. 1989, 49, 475–477.
(6) Walker, L. A.; Jarrett, J. T.; Anderson, N. A.; Pullen, S. H.; Matthews,
R. G.; Sension, R. J. J. Am. Chem. Soc. 1998, 120, 3597–3603.
(11) Stepuro, I. I.; Konovalova, N. V.; Solodunov, A. A.; Tyshchenko,
A. S. Mol. Biol. 1993, 27, 483–487.
9
10.1021/ja107054x  2010 American Chemical Society
J. AM. CHEM. SOC. 2010, 132, 16953–16961 16953

A R T I C L E S
Hill et al.
Scheme 1. Mechanism of the Transamination Half-Reaction from
the External Aldimine Intermediate Onward
AAT catalyzes the reversible reaction of L-aspartate with
R-ketoglutarate to give oxaloacetate and L-glutamate via two
independent half-reactions:7,8,12
E-PLP + L-Asp h E-PMP + oxaloacetate
E-PMP + R-ketoglutarate h E-PLP + L-Glu
(1a)
(1b)
AAT is central to nitrogen metabolism in all living systems and
is a component of the malate/aspartate shuttle that ushers
reducing equivalents into mitochondria of eukaryotes.8 There
is a large, informative body of literature on AAT,7-9,13,14 making
it a useful prototype for fundamental studies on this class of
enzymes, including light activation.
The reaction mechanism of AAT begins with PLP bound as
a Schiff base, with K258 forming the “internal aldimine”. Amino
acid substrates (i.e., L-aspartate or L-glutamate) bind and form
the covalent “external aldimine” (Scheme 1) by displacing K258
from the internal aldimine. It is generally thought that the L-Asp
external aldimine is deprotonated at CR to form the carbanionic
quinonoid intermediate, which is then protonated at C4′ of the
coenzyme to give the oxalacetate ketimine intermediate. The
ketimine is then hydrolyzed to give the pyridoxamine 5′-
phosphate (PMP) enzyme form and free oxalacetate. The second
half-reaction converts E-PMP back to E-PLP using R-ketoglu-
tarate as the second substrate, generating L-glutamate as the
second product, through the reverse of this sequence of
steps.15,16
Figure 1. Comparison of absorption spectra for (A) PLP (red) and PLP-
amino acid Schiff base in pH 9.0 water (blue), and (B) internal aldimine of
AAT (orange) and external aldimine of AAT with R-methylaspartate (green).
Spectra for excitation sources are shown for comparison: 440-nm LED (light
gray), 400-nm laser (dark gray).
It is demonstrated here that blue light (440 nm) increases
the oVerall catalytic activity of wild-type AAT (2.3 ( 0.3)-
fold (Figure 2). In the K258A mutant, where the lysine general
base catalyst is absent causing a dramatic reduction in activity
under thermal conditions, catalysis is increased (530 ( 15)-
fold at the same excitation power (250 mW), and the rate
constant for deprotonation of the L-Asp Schiff base in water is
increased (58600 ( 200)-fold. The kinetics of the light-induced
processes are characterized with time-resolved transient absorp-
tion spectroscopic techniques that span a time scale from
femtoseconds to minutes. A detailed mechanism is proposed
below, involving formation of a reactive triplet state in which
weakening of the CR-H bond facilitates rapid deprotonation.
In the enzyme, the light-induced deprotonation produces a higher
steady-state concentration of the quinonoid intermediate than
in the thermal reaction, resulting in an increase in the overall
reaction rate.
Figure 2. AAT catalytic activity measured in the dark (b) and under 250
mW of 440-nm irradiation (O). Conditions: 20 nM AAT, 10 × KM Asp
and 5 × KM R-ketoglutarate, pH 7.5, 25 °C. Error bars represent one standard
deviation calculated from three trials. For clarity, only every other data
point is shown.
pathway. Although the quinonoid intermediate is considered
central to most PLP-catalyzed mechanisms, its existence in AAT
has been debated. The present work provides support for the
generally accepted stepwise mechanism, in which the quinonoid
is an obligatory intermediate, rather than an alternative concerted
mechanism in which CR-H deprotonation and C4′ protonation
occur simultaneously.16,17
2. Materials and Methods
All reagents were obtained from Fischer Scientific in the highest
quality available and used without further purification unless
otherwise noted. HEPES (0.1 M, adjusted with NaOH) buffer was
used in the appropriate pH range. PLP-Schiff base samples were
prepared as previously described.18 In several experiments, PLP-
Val was used rather than PLP-Asp. Based on the known equilibrium
Finally, analysis of the dependence of AAT catalytic activity
on excitation power provides insight into the thermal reaction
(12) Jenkins, W. T.; Yphantis, D. A.; Sizer, I. W. J. Bio. Chem. 1959,
234, 51–57.
(13) Toney, M. D.; Kirsch, J. F. Protein Sci. 1992, 1, 107–119.
(14) Dunathan, H. C. Proc. Natl. Acad. Sci. U. S. A. 1966, 55, 712.
(15) Kuramitsu, S.; Hiromi, K.; Hayashi, H.; Morino, Y.; Kagamiyama,
H. Biochemistry 1990, 29, 5469–5476.
(17) Julin, D. A.; Kirsch, J. F. Biochemistry 1989, 28, 3825–3833.
(18) Hill, M. P.; Carroll, E. C.; Toney, M. D.; Larsen, D. S. J. Phys. Chem.
B 2008, 112, 5867–5873.
(16) Goldberg, J. M.; Kirsch, J. F. Biochemistry 1996, 35, 5280–5291.
9
16954 J. AM. CHEM. SOC. VOL. 132, NO. 47, 2010

Light-Enhanced Catalysis by PLP-Dependent AAT
A R T I C L E S
constants and pKa values for the PLP-Asp Schiff base in solution,8,19
it is difficult to form a homogeneous ionization-state population
near neutral pH. In contrast, PLP-Val has a more favorable
equilibrium constant with fewer protonation states, allowing for
the generation of samples of homogeneous populations for study.
The samples were purged with N2 (30 min per mL of sample) prior
to study. KI (1 M) was added as a triplet quencher to some samples,
and KCl was used at the same concentration to control for ionic
strength effects. HPLC was performed with an Agilent 1100
instrument using a Supelco Supelcosil C-18 column (4.6 × 250
mm, 5 µm). Product separation employed an isocratic elution with
2% CH3CN/0.1 M potassium phosphate buffer, pH 7.5, and a flow
rate of 0.5 mL/min, with absorbance monitored at 330 nm.
WT and K258A AAT were expressed and purified as previously
described.20,21 The steady-state transamination of L-Asp and
R-ketoglutarate catalyzed by WT AAT was monitored by coupling
the production of oxaloacetate (OAA) to malate dehydrogenase
(MDH), which converts OAA to malate with concurrent oxidation
of NADH.21 Samples were maintained at 25 °C via a temperature-
controlled water jacket. Steady-state rate measurements were made
with an Ocean Optics S2000 spectrometer. The enzyme assay was
conducted at pH 7.5, 100 mM HEPES with 100 mM KCl, 150 µM
NADH, 20 nM AAT, 10 × KM Asp and 5 × KM R-ketoglutarate,
and 10 units/mL MDH. As a control, the light dependence of MDH
activity was measured using 25 mM OAA in MHP buffer pH 7.5
with 100 mM KCl;22 no light-induced change in activity was
observed. The K258A mutant AAT transamination half-reaction
was monitored by the spectral changes in the coenzyme: the K258A-
PLP external aldimine with L-Asp (λmax ) 430 nm) generates
K258A-PMP (λmax ) 335 nm).21 The reaction was conducted under
saturating conditions, with L-Asp in 100-fold molar excess over
K258A-PLP in 100 mM HEPES buffer pH 8.0 and 100 mM KCl.
The half-reaction maximal rate constant, kmax, was obtained by
fitting a single exponential to the 430-nm absorbance decay.
The light dependence of the activity of glutathione reductase,
which uses the blue-absorbing chromophore flavin adenine dinucleo-
tide (FAD) for activity, was also measured using the same
experimental setup. The enzyme-bound FAD, in both the oxidized
and reduced forms, has strong absorption in the 440-nm region
similar to PLP. The reduction of oxidized glutathione by NADPH
was monitored at 340 nm. Reactions were performed at 25 °C in
100 mM potassium phosphate buffer pH 7.6 with 3 mM oxidized
glutathione, 150 µM NADPH, and 0.3 nM glutathione reductase.
All experiments used a variable-intensity 440-nm LED array
(Roithner Laboratories, Austria L435-66-60-590) as the actinic light
source, since it overlaps well with the external aldimine absorption
(Figure 1).16 The K258A reactions were measured with a Varian
Cary 50 Bio with an orthogonal xenon light source filtered to full
width half maximum (FWHM) of 25 nm centered at 450 nm. The
degree of light enhancement measured with this setup was identical
to that from the LED array light source but showed less noise.
Dispersed femtosecond transient absorption measurements were
conducted as previously described.18,23 Samples were excited with
50 fs, 400-nm pulses produced by frequency doubling the output
of an amplified 800-nm Ti:sapphire laser system. Pump-induced
changes in the transmission of a white-light continuum (320-650
nm) were monitored with 50 fs resolution from 50 fs to ∼8 ns;
measurements on this time scale are referred to in this text as
“ultrafast transient absorption”. The Ti:sapphire laser setup was
modified23 to probe dynamics up to 300 µs with 30 ns resolution,
and the resulting measurements referred to as “microsecond transient
Figure 3. UV-visible absorption spectra of PLP in solution at pH 7.0, 25
°C. (A) Under exposure to 50 mW of 440-nm irradiation. (B) Kinetics
showing the slower rise in absorbance at 392 nm (O) compared to 388 nm
(b) and the decay of 320 nm (4). (C) Structures of PLP and its dominant
photoproducts, PLP acid (pyridoxic acid) and the PLP-dimer.
absorption”. In both transient absorption measurements, the polar-
ization of the probe light was set at the magic angle (54.7°) with
respect to the pump polarization. The pump intensity was limited
to prevent multiphoton processes.24-26 Multidimensional global
fitting27,28 was used to analyze the time- and wavelength-dependent
experimental data.
3. Results and Discussion
Photochemistry of Free PLP in Solution. The photochemical
reactions of free PLP in water under continuous irradiation were
monitored by time-resolved changes in the UV-vis absorbance
spectra. At the onset of irradiation (50 mW, 440 nm), the 388-
nm band of free PLP decreased (k > 0.03 s-1) with a concurrent
increase in absorption at 320 nm (Figure 3A). Within 2 min,
essentially all PLP was converted to the oxidized form (pyri-
doxic acid) identified in previous studies.29-31 Continued
irradiation promoted slow formation of a new product (k ≈ 8
(24) Larsen, D. S.; van Stokkum, I. H. M.; Vengris, M.; van der Horst,
M. A.; de Weerd, F. L.; Hellingwerf, K. J.; van Grondelle, R. Biophys.
J. 2004, 87, 1858–1872.
(25) Blandame, M. J. Chem. ReV. 1970, 70, 59.
(26) Larsen, D. S.; Vengris, M.; van Stokkum, I. H. M.; van der Horst,
M. A.; de Weerd, F. L.; Hellingwerf, K. J.; van Grondelle, R. Biophys.
J. 2004, 86, 2538–2550.
(27) Holzwarth, A. R. In Biophysical Techniques in Photosynthesis; Amesz,
J., Hoff, A. J., Eds.; Kluwer: Dordrecht, The Netherlands, 1996.
(28) van Stokkum, I. H. M.; Larsen, D. S.; van Grondelle, R. Biochim.
Biophys. Acta-Bioenergetics 2004, 1657, 82–104.
(19) Metzler, C. M.; Cahill, A.; Metzler, D. E. J. Am. Chem. Soc. 1980,
102, 6075–6082.
(20) Smith, D. L.; Almo, S. C.; Toney, M. D.; Ringe, D. Biochemistry
1989, 28, 8161–8167.
(29) Bazhulina, N. P.; Kirpichnikov, M. P.; Morozov, Y. V.; Savin, F. A.;
Sinyavina, L. B.; Florentiev, V. L. Mol. Photochem. 1974, 6, 367–
396.
(21) Toney, M. D.; Kirsch, J. F. Biochemistry 1993, 32, 1471–1479.
(22) Geck, M. K.; Kirsch, J. F. Biochemistry 1999, 38, 8032–8037.
(23) Carroll, E. C.; Hill, M. P.; Madsen, D.; Malley, K. R.; Larsen, D. S.
ReV. Sci. Instrum. 2009, 80, 026102-3.
(30) Reiber, H. Biochim. Biophys. Acta 1972, 279, 310.
(31) Morrison, A. L.; Long, R. F. J. Chem. Soc. 1958, 211–215.
9
J. AM. CHEM. SOC. VOL. 132, NO. 47, 2010 16955

A R T I C L E S
Hill et al.
Figure 4. Free PLP and PLP-Valine microsecond transient absorption signals. (A) Kinetics with (4) and without (O) KI at 460 nm for free PLP. (B)
Kinetics with (4) and without (O) KI at 460 nm for PLP-Valine Schiff base. (C) Evolution-associated difference spectra (EADS) for free PLP. (D) EADS
for PLP-Valine.
× 10-5 s-1, Figure 3A,B), which is ascribed to the PLP dimer
with known absorption at 392 nm (ꢀ ) 5300 M-1cm-1, pH >
9; ꢀ ) 700 M-1 cm-1, neutral pH).31 These results agree well
with experiments in which PLP photoproducts isolated after
exposure to UV light revealed that the aldehyde group first
oxidizes to the acid before dimerizing (Figure 3C).30,31
associated difference spectra (EADS)35 (Figure 4C) with the
following rate constants: (7.6 ( 1.2) × 105, (1.9 ( 0.1) × 105,
and (7.7 ( 0.6) × 103 s-1. Introducing KI increases the rate
constant for the initial process by an order of magnitude (Figure
4A, 4) while the two slower rates were unaffected. KCl has no
effect at the same concentration. On the basis of its sensitivity
to KI, the first rate constant is assigned to a triplet state, while
the latter two phases are assigned to formation and decay of a
radical species.8,29,30,32
A number of quenching experiments have suggested that free
PLP forms a triplet state when excited by UV and blue
light.8,18,29,30,32 To determine if the observed PLP photochem-
istry involves a triplet state, the dynamics were measured in
the presence of the triplet quencher KI.32-34 The decay of the
388-nm PLP absorption peak is slower in the presence of KI (k
) 0.0051 ( 0.0008 s-1) than in its absence (data not shown).
Solutions with KCl at the same concentration exhibit nearly
identical kinetics to reactions without added salt, showing that
the effect is specific to iodide. These results indicate that the
free PLP triplet state is a precursor to pyridoxic acid and the
PLP dimer.
In previous flash photolysis experiments, it was concluded
that the aldehyde form of PLP exhibits an n,π* triplet state that
undergoes hydrogen abstraction to form a reactive ketyl
radical.36 The triplet state absorbs at 440 nm and exhibits a decay
rate constant of 106 s-1, comparable to the rate constant observed
here.32,36 The ketyl radical has been reported36 to absorb also
at 440 nm with a decay rate constant of 2.4 × 104 s-1, although
Ledbetter32 reported a rate constant of 8.3 × 103 s-1 for radical
decay after 337-nm excitation of PLP in water at pH 7. The
discrepancy in rates could arise from the variation in the PLP
forms excited in solution at the two excitation wavelengths. Both
the aldehyde and the hydrate forms of PLP in water absorb at
337 nm,37 the excitation wavelength used in ref 36, while only
the aldehyde form of PLP absorbs at 400 nm.37 Both forms of
PLP are expected to undergo triplet and radical formation after
337-nm excitation, though possibly with different kinetics.
Transient absorption experiments allow characterization of
triplet dynamics for free PLP following 400-nm excitation
(Figure 4 A,C). The kinetics monitored at 460 nm, the
absorbance maximum for the triplet state, for free PLP at pH
7.0 are shown in Figure 4A. The data were analyzed using a
three-exponential, sequential model resulting in evolution-
(32) Cornish, T. J.; Ledbetter, J. W. Photochem. Photobiol. 1985, 41, 15–
19.
(35) van Stokkum, I. H. M.; Larsen, D. S.; van Grondelle, R. Biochim.
Biophys. Acta-Bioenergetics 2004, 1658, 262–262.
(36) Ledbetter, J. W.; Schaertel, S. J. Photochem. Photobiol. B 1998, 47,
12–21.
(33) McGlynn, S. P.; Azumi, T.; Kinoshita, M. Molecular Spectroscopy
of the Triplet State; Prentice Hall: Englewood Cliffs, NJ, 1969; Vol.
1.
(34) Porter, G.; Windsor, M. W. Proc. R. Soc. London A-Mater. 1958,
(37) Harris, C. M.; Johnson, R. J.; Metzler, D. E. Biochim. Biophys. Acta
1976, 421, 181–194.
245, 238–258.
9
16956 J. AM. CHEM. SOC. VOL. 132, NO. 47, 2010

Light-Enhanced Catalysis by PLP-Dependent AAT
A R T I C L E S
Table 1. Rate Constants for Conversion of PLP-Schiff Base
Samples in Water to PMP in the Dark (k), in 50 mW of 440-nm
Light (λk), and in the Presence of Heavy Atoms
sample
k, s-1
λk, s-1
λk + KI, s-1
PLP-Asp (6.4 ( 0.1) × 10-7 (5.4 ( 0.8) × 10-3 (1.3 ( 0.1) × 10-4
PLP-Val (5.7 ( 0.1) × 10-7 (1.0 ( 0.2) × 10-3 (2.9 ( 0.3) × 10-4
PLP-AIB (3.9 ( 0.2) × 10-7 (3.7 ( 0.7) × 10-4
but cannot undergo deprotonation. Therefore, these results
suggest that the increased reactivity of the triplet state is due to
a general weakening of bonds to CR and not to factors specific
to deprotonation. These results are congruent with previous
studies.38,39
The microsecond transient absorption following 400-nm
excitation of the PLP-Val Schiff base at pH 7.5 exhibited
kinetics with multiexponential decay similar to free PLP (Figure
4B,D). Global fitting to a sequential model with formation of a
persistent photoproduct that did not decay within the 300 µs
duration of the experiment (Figure 4D) identifies two rate
constants: (3.2 ( 0.4) × 107 and (2.9 ( 0.1) × 105 s-1. When
KI was added, the second rate constant was significantly reduced
(Figure 4B), and these data can be fitted to a single exponential
with a rate constant of (2.7 ( 0.2) × 106 s-1 and an offset of
0.07 ( 0.02 (Supporting Information, Figure S6). The sensitivity
of the reaction rate to KI supports involvement of a triplet in
reactions of PLP Schiff bases. The EADS extracted for PLP-
Val are compared in Figure 4D. Within error, there is no
significant evolution of the transient spectrum (Supporting
Information, Figure S5), suggesting that the transient spectrum
corresponds to a single species. In comparison with the spectra
obtained from the free PLP microsecond kinetic data (Figure
4C), PLP-Val has a greater absorbance at 475 nm, a wavelength
commonly reported as the absorbance maximum for the
quinonoid intermediate in solution.40,41 This species is therefore
assigned as the ground-state quinonoid, though the limited range
of the experimental data does not rule out the possibility of a
nonequilibrium precursor to quinonoid. The observed rate
constant of (2.9 ( 0.1) × 105 s-1 includes both the photochemi-
cal reaction that ultimately yields quinonoid and intersystem
crossing back to the singlet ground state of the external aldimine.
On the basis of additional light-dependent studies (unpub-
lished data), we estimate a value of at most 0.64 for the triplet
to quinonoid yield, corresponding to a rate of k ≈ 1.9 × 105
s-1. This rate sets an upper bound on the rate constant for
deprotonation.
Figure 5. (A) UV-visible absorption spectra of PLP-Asp in solution at
pH 9.0, 25 °C, under exposure to 50 mW of 440-nm excitation. (B) Kinetics
at 412 nm with (O) and without (b) KI. Error bars indicate one standard
deviation calculated from three trials.
Photochemistry of PLP-Amino Acid Schiff Bases in Solu-
tion. The photochemistry of PLP-amino acid Schiff bases was
also investigated. The ketoenamine tautomer of PLP-amino acid
Schiff bases, the dominant form in water,8,37 absorbs maximally
at 410 nm. Transamination of the PLP-Schiff bases is observed
as a decrease in absorbance at 410 nm, with a concurrent
increase at 325 nm attributed to the formation of PMP (Figure
5A). The rate of this reaction with PLP-Asp increases 8400-
fold under 50 mW blue light exposure (Figure 5A) and 58600-
fold at 250 mW. The power dependence of the observed rate
constants is shown in the Supporting Information (Figure S1),
as well as data for PLP-Val and PLP-AIB Schiff bases
(Supporting Information, Figures S2 and S3).
PLP-amino acid Schiff base samples irradiated with 50 mW
light reach the end point for PMP formation in ∼30 min. HPLC
analysis of the PLP-Asp and PLP-Val light-induced reactions
confirmed that PMP is produced (Supporting Information, Figure
S4). For the PLP-Asp Schiff base only, further irradiation leads
to a photoproduct that absorbs at 392 nm. Since the equilibrium
mixture of the PLP-Asp Schiff base has ∼37% of the total PLP
present as free PLP, this 392-nm absorbing photoproduct is
assigned to the PLP dimer that is observed in the reactions of
free PLP in the absence of amino acid (Figure 3C).19 Samples
with more favorable Schiff base equilibrium constants (e.g.,
PLP-Val and PLP-AIB) contain much less free PLP and do not
exhibit this photoproduct (Figures S2 and S3).
The primary photodynamics of triplet-state formation for the
Schiff bases was examined in ultrafast (femtoseconds to
nanoseconds) experiments. The kinetics of the PLP-Asp Schiff
base at 457 nm (Figure 6A) are similar to those obtained in
previous ultrafast studies of PLP-Val in solution.18 An initial
photoproduct with an absorption maximum at 450 nm is formed
within 1 ns (Figure 6B). In the previous work, this species was
incorrectly assigned to the quinonoid intermediate.18 The KI
quenching results reported above instead support triplet assign-
ment for both samples. The pure spectrum for this species was
KI significantly slows the light-dependent reaction under
otherwise identical conditions (Figure 5B), indicating that a
triplet state is involved in the light-induced deprotonation
reaction in solution. The changes in rate constants are sum-
marized for three PLP-Schiff base samples in Table 1. Similar
results were found for all Schiff bases tested. Like PLP-Asp
and PLP-Val, the reaction rate for the PLP-AIB Schiff base in
solution is increased by blue light exposure and is sensitive to
KI (Figure S3). PLP-AIB Schiff base undergoes decarboxylation
(38) Kurauchi, Y.; Ohga, K.; Morita, S.; Nagamura, T.; Matsuo, T. Chem.
Lett. 1983, 349–352.
(39) Kurauchi, Y.; Ohga, K.; Yokoyama, A.; Morita, S. Agric. Biol. Chem.
1980, 44, 2499–2500.
(40) Metzler, C. M.; Harris, A. G.; Metzler, D. E. Biochemistry 1988, 27,
4923–4933.
(41) Zabinski, R. F.; Toney, M. D. J. Am. Chem. Soc. 2001, 123, 193–
198.
9
J. AM. CHEM. SOC. VOL. 132, NO. 47, 2010 16957

A R T I C L E S
Hill et al.
Figure 6. (A) Ultrafast kinetics at 457 nm for the AAT external aldimine with R-methylaspartate and the PLP-Asp Schiff base in solution show intersystem
crossing on a sub-nanosecond time scale. (B) Reconstructed species associated spectra (SAS) were generated by subtracting the ground state and solvated
electron spectra from the 6-ns transient absorbance spectrum.
constructed by decomposing the 6 ns transient absorption
spectrum into ground state bleach and a solvated electron
spectrum (Supporting Information, Figure S7) generated from
multiphoton ionization.26 The triplet is formed within 500 ps
with PLP-Asp and 1 ns with PLP-Val.18
Photochemistry in Enzymes. AAT reactions irradiated with
250 mW of 440-nm light exhibit a 2.3-fold increase in steady-
state catalytic activity compared to reactions in the dark (Figure
2). The coupling enzyme, MDH, was not affected by light
exposure (Supporting Information, Figure S8). Light enhance-
ment of the K258A mutant of AAT, in which the lysine
responsible for CR-H deprotonation is replaced with alanine,42
was also studied (Supporting Information, Figure S9). The half-
reaction kmax for K258A in the dark is (7.0 ( 0.2) × 10-6 s-1
for L-Asp at pH 8.0, 25 °C.21 Excitation with 250 mW of 450-
nm light increases the rate constant 530-fold. The rate of light
enhancement in both WT and K258A AAT is power-dependent
Figure 7. Proposed scheme for PLP Schiff base photodynamics in solution
(Supporting Information, Figure S10), as discussed below.
Although photoexcitation of PLP Schiff bases both in solution
and on AAT generates PMP, the same photochemical mecha-
nisms are not necessarily in effect for both. Unfortunately,
heavy-atom studies are inconclusive with enzymes since the
reactive species are removed from bulk solvent. However,
ultrafast triplet-state formation dynamics of PLP-Asp in solution
can be directly compared to that obtained with AAT. The
ultrafast transient absorption signals of the internal and external
aldimines in AAT at pH 8.0 after 400 nm both resulted in long-
lived photoproduct formation within 55 and 110 ps, respectively
(Figure 6A, O), that have an absorption maximum at 450 nm
(Figure 6B, O). Due to the strong resemblance of the long-
lived intermediate in the transient spectrum of AAT and PLP-
Asp in solution (Supporting Information, Figure S11), the
population at 6 ns is assigned to a triplet state.18 This assignment
suggests that the protein does not significantly alter the
intersystem crossing mechanism that leads to the reactive triplet
species, and that Schiff bases in both water and AAT follow
similar photochemical mechanisms.
and on AAT. Rate constants refer to PLP-Val in water at pH 7.5, 25 °C.
Dashed line schematically represents the thermal reaction.
dependence of the activity of the FAD-dependent enzymes
glutathione reductase and diaphorase (Supporting Information).
Both enzymes absorb in the 410-500-nm region with extinction
coefficients similar to PLP enzymes.44 Ultrafast transient
absorption investigations on free flavins have shown generation
of a triplet state within 1-2 ns, and in numerous light-dependent
flavoenzymes, formation of a triplet is necessary to enzymatic
activity.45 However, the activities of these two enzymes were
not increased by blue light (Supporting Information, Figure S12
for glutathione reductase; data not shown for diaphorase). These
results suggest that local heating is not the mechanism for the
increase in AAT activity.
Incorporating the results of both the steady-state and transient
absorption measurements, we arrive at a scheme for the primary
photochemistry underlying the light-dependent formation of
PMP from Schiff bases in solution and in AAT (Figure 7). In
this scheme, blue light excites the external aldimine from the
ground state to an excited singlet state (λmax ) 465 nm) that
decays within 1 ns (k ) 1.0 × 109 s-1) to an excited triplet
state (λmax ) 455 nm). The triplet state decays in 3.5 µs (k )
2.9 × 105 s-1), during which CR-H cleavage generates the
carbanionic quinonoid intermediate (λmax ) 475 nm) in its
An alternative explanation of the light enhancement of the
enzymatic reactions observed here is that absorption of a photon
simply leads to generation of a vibronically hot chromophore43
that heats the local environment through vibrational relaxation,
thereby increasing the rate of the thermal mechanism. This
alternative hypothesis was tested by measuring the light
(44) Bohme, C. C.; Arscott, L. D.; Becker, K.; Schirmer, R. H.; Williams,
C. H. J. Biol. Chem. 2000, 275, 37317–37323.
(42) Malcolm, B. A.; Kirsch, J. F. Biochem. Biophys. Res. Commun. 1985,
132, 915–921.
(45) Swartz, T. E.; Corchnoy, S. B.; Christie, J. M.; Lewis, J. W.; Szundi,
I.; Briggs, W. R.; Bogomolni, R. A. J. Biol. Chem. 2001, 276, 36493–
36500.
(43) Sibert, E. L.; Reinhardt, W. P.; Hynes, J. T. J. Chem. Phys. 1984, 81,
1115–1134.
9
16958 J. AM. CHEM. SOC. VOL. 132, NO. 47, 2010

Light-Enhanced Catalysis by PLP-Dependent AAT
A R T I C L E S
ground state. The decay of the absorbing species persisting
longer than 300 µs (k ) 3.3 × 104 s-1) is assumed to form the
ketimine intermediate or revert back to the external aldimine
ground state.
The light enhancement of catalytic activity is proposed to
result from an increase in the rate of deprotonation of the
external aldimine to form the quinonoid intermediate due to
higher acidity of the CR-H bond in the triplet state. Therefore,
the ∆pKa between the ground and triplet state for the aldimine
CR-H bond is a quantity of fundamental interest to calculate.
Pearson and Dillon demonstrated a correlation between the rate
constant for deprotonation of carbon acids and their pKa values.46
One can use this correlation to estimate the ∆pK for the aldimine
CR-H bond in the aqueous Schiff bases from the difference in
the rate constant for deprotonation. Although the rate constant
for quinonoid formation was not obtained in the present
experiments, upper and lower bounds on this rate constant are
set by those for triplet decay and PMP formation, respectively.
The triplet decay rate constant obtained from microsecond
transient absorption provides an upper bound on the rate constant
for deprotonation of 2.9 × 105 s-1 for PLP-Val. The rate
constant for deprotonation cannot be slower than that of the
light-dependent transamination reaction, which involves both
deprotonation and quinonoid reprotonation. Triplet formation
is fast (i.e., at equilibrium) compared to product formation under
continuous illumination, therefore a pre-equilibrium treatment
can be applied. The program COmplex PAthway SImulator
(COPASI)47 was used to simulate, at a fixed excitation power,
the steady-state concentration of the triplet state. By dividing
the observed rate constant for the light-dependent transamination
of PLP-Val (1.0 × 10-3 s-1, Table 1) by the fraction of the
population in the triplet state (0.0008 at 50 mW excitation
power), the maximal rate constant for PMP formation and the
lower bound on the rate of deprotonation is estimated to be
1.25 s-1. Comparing these estimates for the rate constant for
triplet deprotonation to the thermal deprotonation of the PLP-
Val ground state (5.7 × 10-7 s-1, Table 1), a decrease of
between 11 and 19 units in the pKa of CR-H in the triplet state
of the PLP-Val Schiff base in water is estimated. Previous
reports have estimated the CR-H pKa in the ground state of the
PLP-glycine Schiff base in water to be ∼17;8,48-50 a ∆pKa in
the range 11-19 would then result in a pKa* between 6 and
-2. These estimates of ∆pKa for the aqueous Schiff bases are
strikingly large but are consistent with the experimental pKa*
value of approximately or greater than -2.5 reported for free
PLP in solution.51
Figure 8. Power dependence of the AAT catalytic activity. Simulations
with two kinetic models are shown. Model 1 has the quinonoid on the
productive pathway, while model 2 has it off the productive pathway. Error
bars indicate one standard deviation calculated from three trials. Conditions:
20 nM AAT, 10 × KM Asp, and 5 × KΜ R-ketoglutarate pH 7.5, 25 °C.
performed by the K258 amino group in the active site. Although
the K258A mutant presumably employs water as the base for
CR-H deprotonation, the steric barrier to water access presented
by the protein contributes significantly to the kinetics and
precludes use of the correlation (which only includes reactions
with insignificant steric effects) in calculating ∆pKa.52 The active
site certainly facilitates the thermal reaction by decreasing the
energy between the external aldimine and quinonoid intermedi-
ates (i.e., by lowering the pKa of the CR-H bond) compared to
reaction in water, and the activation effect of light is mitigated
by this intrinsic protein activation.
Power Dependence of kcat and Implications for the Thermal
Mechanism. The light enhancement of AAT catalyzed tran-
samination exhibits a clear nonlinear dependence on excitation
power (Figure 8). While the rate constant for the microscopic
step that is affected by light (i.e., L-Asp external aldimine CR-H
deprotonation) is expected to have a linear dependence on
excitation power, two factors contribute to the observed
nonlinear dependence of kcat. First, the balance of rate-limiting
steps changes at high excitation power, where CR-H deproto-
nation is no longer rate-limiting. Second, the electronic transition
in the external aldimine optically saturates (discussed below).The
first phenomenon is a result of the multistep nature of the
enzymatic reaction mechanism. The thermal reaction is only
One can propose that the observed trend in optical enhance-
ment (water > K258A > WT) is due to the greater reactivity of
the ground state Schiff base in the active site environment
compared to water. Unfortunately, the correlation of Pearson
and Dillon is for deprotonation of carbon acids by water, and
is therefore inappropriate for AAT, where deprotonation is
(46) Pearson, R. G.; Dillon, R. L. J. Am. Chem. Soc. 1953, 75, 2439–
2443.
(47) Hoops, S.; Sahle, S.; Gauges, R.; Lee, C.; Pahle, J.; Simus, N.; Singhal,
M.; Xu, L.; Mendes, P.; Kummer, U. Bioinformatics 2006, 22, 3067–
3074.
(48) Richard, J. P.; Amyes, T. L.; Crugeiras, J.; Rios, A. Curr. Opin. Chem.
Biol. 2009, 13, 475–483.
(49) Weng, S. H.; Leussing, D. L. J. Am. Chem. Soc. 1983, 105, 4082–
4090.
(50) Crugeiras, J.; Rios, A.; Riveiros, E.; Richard, J. P. J. Am. Chem. Soc.
2009, 131, 15815–15824.
(51) Bridges, J. W.; Davies, D. S.; Williams, R. T. Biochem. J. 1966, 98,
451.
(52) Toney, M. D.; Kirsch, J. F. Science 1989, 243, 1485–1488.
9
J. AM. CHEM. SOC. VOL. 132, NO. 47, 2010 16959

A R T I C L E S
Hill et al.
partially rate-limited by CR-H deprotonation, as judged from
kinetic isotope effects16,53 and this partial rate limitation
disappears as the rate for the deprotonation step increases with
excitation power, with subsequent, light-independent steps
becoming rate-limiting. Equation 2 (see Supporting Information
for derivation) models the light-dependent increase in kcat,
relative to the thermal reaction, for a simplified model of the
AAT reaction.
The existence of the quinonoid intermediate on the productive
thermal pathway in AAT has been contested in the literature.16,53
The quinonoid is observed at low concentration in equilibrium
mixtures of AAT with aspartate or glutamate and their respective
R-keto acids.16,40,54 The alternative substrate erythro-ꢀ-hydroxy-
aspartate gives a higher fraction of AAT in the quinonoid
state.40,55 The quinonoid has been established as a kinetically
competent intermediate in the reaction of AAT with the
alternative substrate L-cysteine sulfinate56 and in numerous other
PLP enzymes including Citrobacter freundii tyrosine phenol-
lyase57 and dialkylglycine decarboxylase.58 An alternative AAT
mechanism was suggested by Goldberg and Kirsch in which
the 1,3-prototropic shift from external aldimine to ketimine is
concerted, with the quinonoid intermediate off the productive
pathway.16 They speculated that the 1,3-prototropic shift
proceeds through a nonlinear transition state with simultaneous
CR-H deprotonation and C4′ protonation catalyzed by K258.
COPASI47 was used to simulate steady-state kinetic data from
which kcat values for AAT catalyzed reactions at various light
intensities were calculated. Simulations include the microscopic
rate constants in model 1 of Figure 8, which has the quinonoid
intermediate on the thermal pathway. Rate constants in black
were obtained from the literature15,16 or obtained from the
analysis above. The values of 11550 and 800 s-1 were estimated
from the percentage population present on AAT saturated with
both aspartate and oxalacetate (20% external aldimine and 1%
quinonoid intermediate).16 The simulated data clearly reproduce
the nonlinear power dependence of the observed value of kcat.
In contrast, model 2, with the quinonoid off the thermal
pathway,16 predicts an inhibition of catalytic activity since
photoexcitation would decrease the population on the productive
pathway. Thus, the photoenhancement results presented here
strongly support the quinonoid being on the productive thermal
pathway, as well as the photochemical pathway. Corroborating
this, it is also found that 440-nm irradiation increases the
population of the quinonoid intermediate in stopped-flow
experiments with AAT (unpublished results).
λk3
k3
k3
k4
λk3
k4
1 +
+
+
λkcat
kcat
)
(2)
k3
k4
λk3
k4
1 +
+
Here, k3 is the net rate constant for conversion of enzyme
bound aspartate (which includes Michaelis complex and external
λ
aldimine) to the quinonoid, k3 is the net rate constant for the
light-dependent pathway for quinonoid formation, and k4 is the
net rate constant for all steps subsequent to quinonoid formation
(i.e., reprotonation at C4′, ketimine hydrolysis, and the second
half-reaction with R-ketoglutarate). The model used here
subsumes the equilibrium constant for Michaelis complex-
external aldimine interconversion into k3 along with external
aldimine deprotonation. This equilibrium constant is known to
be near unity.15
Values for k3 and k4 for the simplified model of the thermal
reaction were determined by fitting eq 2 to the experimental
values of kcat as a function of excitation power (Figure 8). To
do this, values of λk3 were first calculated for each experimental
excitation power, using rate constants obtained directly from
the transient absorption experiments reported above, with k* )
B12I/c, where B12 is the Einstein B coefficient for absorption
(assuming a 0.64 yield for quinonoid from triplet), I is the
irradiance per unit frequency interval in W m-2 Hz-1, and c is
the speed of light (see Supporting Information). The λk3 values
were then used as the independent variable against which the
λ
ratio kcat/kcat was fitted. The fit resulted in values of k3 ) 230
Photoenhancement of the activity of other PLP-dependent
enzymes (dialkylgycine decarboxylase and alanine racemase)
occurs with magnitudes similar to that with AAT (unpublished
results). Thus, the acceleration of PLP enzymes by light may
well be a general phenomenon. If this is the case, then the light
dependence of the activity of PLP-dependent enzymes may play
a heretofore unrecognized role in the physiology of organisms
that live under solar radiation (e.g., plants, fungi, algae, bacteria).
PLP enzymes are ubiquitous in the physiology of all organisms,
being central to amino acid metabolism. In E. coli, PLP enzymes
constitute ∼2% of the proteins encoded in the genome.7 Under
the solar standard AM 1.5, ∼30% of solar spectrum is resonant
with the external aldimine absorption band in AAT. If the full
resonant flux were absorbed, the results reported here suggest
that there would be a generalized 50% increase in the activity
of PLP enzymes. Even a fraction of this calculated increase in
the activity of such a large number of enzymes could have very
significant biological effects. One can speculate the involvement
( 45 s-1 and k4 ) 425 ( 73 s-1. The value of kcat for the overall
thermal reaction calculated from the fitted values of k3 and k4
is 149 ( 23 s-1, in good agreement with the present and
literature values.15,16 These results demonstrate the power of
selectively perturbing a single step by a known amount in a
multistep process such as the AAT reaction with aspartate. Here,
it allows one, from purely steady-state rate measurements, to
define the value of the thermal rate constant for the perturbed
step as well as for subsequent steps.
The rate constants for both K258A and Schiff bases in water
will saturate at high excitation power due to saturation of the
external aldimine optical transition. When the rate of excitation
equals the rate of de-excitation due to stimulated emission
pumping of the excited state (i.e., dEA*/dt ) 0), the steady-
state concentration of excited singlet and the triplet states
becomes independent of excitation power. Because the singlet
excited state is longer lived for K258A (k ≈ 1 × 108 s-1; data
not shown) compared to Schiff bases in solution or on AAT
(k ≈ 9 × 108 s-1), the optical transition saturates at a higher
excitation power (Supporting Information, Figure S13). COPASI
simulations of the AAT reaction indicate that optical saturation
has a significant effect on kcat only at much higher excitation
powers than reported here.
(54) Jenkins, W. T.; Taylor, R. T. J. Biol. Chem. 1965, 240, 2907.
(55) Jenkins, W. T. J. Biol. Chem. 1961, 236, 1121.
(56) Furumo, N. C.; Kirsch, J. F. Arch. Biochem. Biophys. 1995, 319, 49–
54.
(57) Chen, H. Y.; Phillips, R. S. Biochemistry 1993, 32, 11591–11599.
(58) Zhou, X. Z.; Jin, X. G.; Medhekar, R.; Chen, X. Y.; Dieckmann, T.;
Toney, M. D. Biochemistry 2001, 40, 1367–1377.
(53) Gloss, L. M.; Kirsch, J. F. Biochemistry 1995, 34, 3999–4007.
9
16960 J. AM. CHEM. SOC. VOL. 132, NO. 47, 2010

Light-Enhanced Catalysis by PLP-Dependent AAT
A R T I C L E S
of this phenomenon in the coordination of day vs night
metabolism, blue light sensing, and other regulatory phenomena.
synchronized population. If other enzymes are found to show
increased catalytic activity due to increased reactivity of singlet
excited states, then ultrafast photoinitiation of enzymatic reac-
tions that are not intrinsically light-dependent could lead to new
and interesting insights into enzyme catalysis, in a similar vein
to ultrafast studies of ligand binding to heme proteins.4,59
4. Conclusions
Photoexcitation of free PLP, PLP Schiff bases, and PLP bound
to AAT leads to the formation of a reactive triplet state. The triplet
state lowers the pKa of the CR-H bond by 11-19 units for Schiff
bases in water. The increased acidity of this bond in the triplet
state increases the rate of nonenzymatic transamination by 58600-
fold, the rate of transamination catalyzed by the K258A mutant of
AAT by 530-fold, and the rate of the AAT reaction by 2.3-fold
under 250 mW of 440-nm irradiation. The reaction mechanism of
AAT is complex, and the rate of AAT is only partially limited by
CR-H deprotonation,15,16 resulting in a nonlinear dependence
of the rate enhancement on light intensity. Selective perturbation
of the deprotonation step by light is a powerful tool that allows
this individual rate constant in the complex mechanism to be
defined from steady-state kinetics only, and demonstrates for
AAT that the quinonoid intermediate is on the thermal reaction
pathway. Photoinitiation of PLP reactions is possible on the
1-100 µs time scale, with pulsed excitation generating a
Acknowledgment. This work was supported with a Career
Development Award (CDA0016/2007-C) from the Human Frontiers
Science Organization (to DSL) and a grant (GM54779) from the
National Institutes of Health (to MDT). We thank Dr. Nathan C.
Rockwell, Rob Griswold, Justin Foust and Kristin Ziebart for
laboratory assistance and constructive discussions.
Supporting Information Available: Additional data. This
material is available free of charge via the Internet at http://
pubs.acs.org.
JA107054X
(59) Petrich, J. W.; Lambry, J. C.; Kuczera, K.; Karplus, M.; Poyart, C.;
Martin, J. L. Biochemistry 1991, 30, 3975–3987.
9
J. AM. CHEM. SOC. VOL. 132, NO. 47, 2010 16961