Photoactivation of the Photoactive Yellow Protein: Why
Photon Absorption Triggers a Trans-to-Cis Isomerization of
the Chromophore in the Protein
Gerrit Groenhof,†Mathieu Bouxin-Cademartory,§Berk Hess,‡Sam P. de Visser,§
Herman J. C. Berendsen,†Massimo Olivucci,|Alan E. Mark,†and
Michael A. Robb*,§
Contribution from the Departments of Biophysical Chemistry and Applied Physics,
UniVersity of Groningen, Nijenborg 4, 9747 AG Groningen, The Netherlands,
Chemistry Department, Imperial College London, London SW7 2AZ, United Kingdom, and
Dipartimento di Chimica, UniVersita ` di Siena, Via Aldo Moro, I-53100 Siena, Italy
Received November 12, 2003; E-mail: email@example.com
Abstract: Atomistic QM/MM simulations have been carried out on the complete photocycle of Photoactive
Yellow Protein, a bacterial photoreceptor, in which blue light triggers isomerization of a covalently bound
chromophore. The “chemical role” of the protein cavity in the control of the photoisomerization step has
been elucidated. Isomerization is facilitated due to preferential electrostatic stabilization of the chromophore’s
excited state by the guanidium group of Arg52, located just above the negatively charged chromophore
ring. In vacuo isomerization does not occur. Isomerization of the double bond is enhanced relative to
isomerization of a single bond due to the steric interactions between the phenyl ring of the chromophore
and the side chains of Arg52 and Phe62. In the isomerized configuration (ground-state cis), a proton transfer
from Glu46 to the chromophore is far more probable than in the initial configuration (ground-state trans).
It is this proton transfer that initiates the conformational changes within the protein, which are believed to
lead to signaling.
A wide variety of organisms have evolved mechanisms to
detect and respond to visible light. In many cases, the biological
response is mediated by structural changes that follow photon
absorption in the protein complex. The initial step in such cases
is normally the photoisomerization of a highly conjugated
prosthetic group. How this leads to large-scale structural changes
of the whole complex is, however, poorly understood. Here,
we report atomistic QM/MM simulations of the photocycle of
the Photoactive Yellow Protein (PYP), a bacterial photoreceptor.
The simulations uncover the detailed sequence of structural
changes that follow photon absorption, including the photo-
isomerization of the covalently bound chromophore and the
intramolecular proton transfer, which leads to the formation of
the signaling state. The results of the simulations (which are
consistent with experimental observations) provide detailed
structural and dynamic information at a “resolution” well beyond
that achievable experimentally. The “chemical role” of the
protein cavity in the control of the photoisomerization has been
elucidated. Thus, the key residues in the protein have been
identified, and the mechanism by which they control the
necessary conformational changes that lead ultimately to signal
transduction in PYP has been documented.
PYP is the primary photoreceptor for the negative photo-
tactic response of Halorhodospira halophila.1,2It has been
extensivelystudied both as a model photoreceptor protein and
as a structural prototype for the PAS class of signal transduc-
tion proteins.3Blue light induces a trans-to-cis isomerization
of a double bond in the covalently bound p-coumaric acid
chromophore4-9(Figure 1). In the resulting metastable state, a
change in the protonation state of the chromophore triggers
major conformational changes in the protein which are believed
to give rise to signal transduction.10-12Structural studies of
†Dept. of Biophysical Chemistry, University of Groningen.
‡Micromechanics of Materials Group, Dept. of Applied Physics,
University of Groningen.
§Imperial College London.
|Universita ` di Siena.
(1) Hellingwerf, K. J.; Hendriks, J.; Gensch, T. J. Phys. Chem. A 2003, 107,
(2) Sprenger, W. W.; Hoff, W. D.; Armitage, J. P.; Hellingwerf, K. J. J.
Bacteriol. 1993, 177, 3096-3104.
(3) Pellequer, J. L.; Wager-Smith, K. A.; Getzoff, E. D. Proc. Natl. Acad. Sci.
U.S.A. 1998, 95, 5884-5890.
(4) Hoff, W. D.; Dux, P.; Hard, K.; Devreese, B.; Nugteren-Roodzant, I. M.;
Crielaard, W.; Boelens, R.; Kaptein, R.; van Beeumen, J.; Hellingwerf, K.
J. Biochemistry 1994, 33, 13959-13962.
(5) Baca, M.; Borgstahl, G. E. O.; Boissinot, M.; Burke, P.; Williams, D. R.;
Slater, K.; Getzoff, E. D. Biochemistry 1994, 33, 14369-14377.
(6) Borgstahl, G. E. O.; Williams, D. R.; Getzoff, E. D. Biochemistry 1995,
(7) Perman, B.; S ˇrajer, V.; Ren, Z.; Teng, T. Y.; Pradervand, C.; Ursby, T.;
Bourgeois, D.; Schotte, F.; Wulff, M.; Kort, R.; Hellingwerf, K. J.; Moffat,
K. Science 1998, 279, 1946-1950.
(8) Genick, U. K.; Soltis, S. M.; Kuhn, P.; Canestrelli, I. L.; Getzoff, E. D.
Nature 1998, 392, 206-209.
(9) Genick, U. K.; Borgstahl, G. E.; Ng, K.; Ren, Z.; Pradervand, C.; Burke,
P. M.; S ˇrajer, V.; Teng, T. Y.; Schildkamp, W.; McRee, D. E.; Moffat, K.;
Getzoff, E. D. Science 1997, 275, 1471-1475.
(10) Xie, A.; Hoff, W. D.; Kroon A. R.; Hellingwerf, K. J. Biochemistry 1996,
(11) Hoff, W. D.; Xie, A.; Van Stokkum, I. H. M.; Tang, X.-J.; Gural, J.; Kroon,
A. R.; Hellingwerf, K. J. Biochemistry 1999, 38, 1009-1017.
Published on Web 03/11/2004
4228 9 J. AM. CHEM. SOC. 2004, 126, 4228-4233
10.1021/ja039557f CCC: $27.50 © 2004 American Chemical Society
kinetically or cryogenically trapped intermediates4-6provide
glimpses of structural changes that follow photoexcitation13but
do not address the crucial question of how the protein achieves
high signaling efficiency and yield no consensus about the
precise sequence of events in the essentially dynamic process.
To model correctly the dynamics of the photoactivated trans-to-cis
isomerization of the chromophore in PYP, the ground- and excited-
state potential energy surfaces must be described accurately. The
reaction starts in the excited state (S1) but ends in the ground state
(S0), so it is essential to model the “hops” from the excited to the ground
state quantum mechanically. Thus, the MD simulations were performed
using a hybrid quantum mechanics/molecular mechanics (QM/MM)
approach14including surface selection at the conical intersection seam
between the excited and ground state.15For these calculations, a special
version of Gromacs 3.016with an interface to Gaussian 9817was used.
The chromophore was described at the CASSCF/3-21G level,18with
an active space of 6 electrons and 6 orbitals. Strictly diabatic surface
hopping was allowed only on the conical intersection hyperline of the
S1and S0surfaces. The remainder of the system, consisting of the apo-
protein, 4845 SPC water molecules,19and 6 Na+ions in a periodic
box, was modeled with the GROMOS96 force field.20The CR-C?bond
of Cys69 connecting the QM and MM subsystems was replaced by a
constraint,21and the QM part was capped with a hydrogen atom. The
force on the cap atom was distributed over the two atoms of the bond.
The QM system experienced the Coulomb field of all MM atoms;
Lennard-Jones interactions between MM and QM atoms were added.
Our treatment resembles that used previously by Warshel and Chu22
to study the photoisomerization of bacteriorhodopsin. Warshel and
Chu22implemented a more sophisticated surface-hopping algorithm but
used a much lower-level semiempirical QM method.23Also, in contrast
to the work of Warshel and Chu,22induced dipoles were not included
in the environment, as this was considered incompatible with the use
of a nonpolarizable MM force field. After the isomerization, the
simulations were continued classically (MM), using the chromophore
force field described in previous work.24
The CASSCF computations are the main computational cost. At each
step in the MD simulation, one must select the appropriate CI
eigenvector to compute the gradient. We have used a simplified surface
hop procedure which we now describe. A diabatic “hop” from S1to S0
was forced when the energy gap was below a given threshold, and the
CI vector indicated that a crossing had been passed. We use the symbol
ground state and K ) 2 for the excited state). Initially, Ci
compute the gradient. When the innerproduct Ci
small and the other innerproduct Ci
trajectory has passed a point of an avoided or a real crossing. Provided
the energy difference is less than a threshold, one assumes that the
step i-1 f i has passed a point of real crossing, and one selects the
is obviously conserved. Of course, such a procedure only allows surface
hopping on the n - 2-dimensional hyperline of the conical intersection.
This procedure could lead to an underestimation of the crossing
probability. However, our experience has shown that, in larger
polyatomic systems, the conical intersection line is almost impossible
to avoid because of its large dimensionality and most surface hops are,
in practice, essentially diabatic.
The affinity of the chromophore for the Glu46 proton (see Figure
1) was computed as follows: In each frame of the classical MM
simulation that was continued after the isomerization, the Glu46 proton
was placed at 20 different positions on the line connecting the phenolate
oxygen of the chromophore and the carboxylate oxygen atom of Glu46.
The energy of each configuration was then calculated using the
following QM/MM scheme. The contribution of the QM subsystem
consisting of the chromophore and the side chains of Tyr42 and Glu46
was computed at the PM3 level.25To this was added the classical
electrostatic contribution of the environment. In this way, the complete
response of the quantum mechanical system to the position of the
proton, including polarization effects, is incorporated. The procedure
Kto refer to the CI vector for state K at MD step i (K ) 1 for the
2is used to
approaches 1, then the
1rather than Ci
2to compute the gradient at step i. Energy
(12) Rubinstenn, G.; Vuister, G. W.; Mulder, F. A. A.; Dux, P. E.; Boelens, R.;
Hellingwerf, K. J.; Kaptein, R. Nat. Struct. Biol. 1998, 5, 568-570.
(13) Ren, Z.; Perman, B.; S ˇrajer, V.; Teng, T.-Y.; Pradervand, C.; Bourgeois,
D.; Schotte, F.; Ursby, T.; Kort, R.; Wulff, M.; Moffat, K. Biochemistry
2001, 40, 13788-13801.
(14) Warshel, A.; Levitt, M. J. Mol. Biol. 1976, 109, 227-249.
(15) Tully, J. C. J. Chem. Phys. 1990, 93, 1061-1071.
(16) Lindahl, E.; Hess, B.; van der Spoel, D. J. Mol. Model. 2001, 7, 306-317.
(17) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M.
A.; Cheeseman, J. R.; Zakrzewski, V. G.; Montgomery, J. A., Jr.; Stratmann,
R. E.; Burant, J. C.; Dapprich, S.; Millam, J. M.; Daniels, A. D.; Kudin,
K. N.; Strain, M. C.; Farkas, O.; Tomasi, J.; Barone, V.; Cossi, M.; Cammi,
R.; Mennucci, B.; Pomelli, C.; Adamo, C.; Clifford, S.; Ochterski, J.;
Petersson, G. A.; Ayala, P. Y.; Cui, Q.; Morokuma, K.; Malick, D. K.;
Rabuck, A. D.; Raghavachari, K.; Foresman, J. B.; Cioslowski, J.; Ortiz,
J. V.; Stefanov, B. B.; Liu, G.; Liashenko, A.; Piskorz, P.; Komaromi, I.;
Gomperts, R.; Martin, R. L.; Fox, D. J.; Keith, T.; Al-Laham, M. A.; Peng,
C. Y.; Nanayakkara, A.; Gonzalez, C.; Challacombe, M.; Gill, P. M. W.;
Johnson, B. G.; Chen, W.; Wong, M. W.; Andres, J. L.; Head-Gordon,
M.; Replogle, E. S.; Pople, J. A. Gaussian 98, revision A.7; Gaussian,
Inc.: Pittsburgh, PA, 1998.
(18) Roos, B. O.; Taylor, P. R.; Siegbahn, P. E. M. Chem. Phys. 1980, 48,
(19) Berendsen, H. J. C.; Postma, J. P. M.; van Gunsteren, W. F.; Hermans J.
In Intermolecular Forces; Pullman, B., Ed.; D. Reidel Publishing Co.:
Dordrecht, 1981; pp 331-342.
(20) Van Gunsteren, W. F.; Billeter, S. R.; Eising, A. A.; Hu ¨nenberger, P. H.;
Kru ¨ger, P.; Mark, A. E.; Scott, W. R. P.; Tironi, I. G. Biomolecular
simulation: GROMOS96 manual and user guide; BIOMOS b.v: Zu ¨rich,
(21) Hess, B.; Bekker, H.; Berendsen, H. J. C.; Fraaije, J. G. E. M. J. Comput.
Chem. 1997, 18, 1463-1472.
(22) Warshel, A.; Chu, Z. T. J. Phys. Chem. B 2001, 105, 9857-9871.
(23) Warshel, A. Nature 1976, 260, 679-683.
(24) Groenhof, G.; Lensink, M. F.; Berendsen, H. J. C.; Mark, A. E. Proteins
2002, 48, 212-219.
(25) Stewart, J. P. P. J. Comput. Chem. 1989, 10, 209-220.
Figure 1. The p-coumaric acid chromophore in PYP: (a) in the resting
state (pG) of the protein;6(b) after the absorption and isomerization (pR,
last frame of a QM/MM MD trajectory). The chromophore is covalently
linked to the side chain of Cys69 through a thioester bond. Its p-hy-
droxyphenyl moiety is deprotonated but stabilized by hydrogen-bonding
interactions with the side chains of Tyr42 and Glu46. The negative charge
is stabilized by the electrostatic interaction with the positively charged
guanidinium group of Arg52. Due to steric interactions, isomerization
involves not only the trans-to-cis conversion of the double bond but also
flipping of the thioester linkage. The images were created with Molscript40
Photoactivation of the Photoactive YellowProteinA R T I C L E S
J. AM. CHEM. SOC. 9 VOL. 126, NO. 13, 2004 4229
was also performed on an extended ground-state trajectory, which
served as a general reference.
In a frame for which the first analysis indicates that proton transfer
would be energetically favorable, the proton was manually removed
from the Glu46 side chain and added to the chromophore. This modified
frame was then used as the starting structure in a next classical MM
simulation, which was performed to measure the effect of the proton
Fluorescence lifetimes were determined as follows. As soon as the
barrier separating the untwisted S1trans minimum and the global twisted
S1 minimum was passed (see Figure 2), a trajectory was no longer
considered to contribute to fluorescence. The time required for passing
the barrier was taken as the fluorescence lifetime of that trajectory.
Results and Discussion
Figure 2 shows a “cartoon” of the potential energy surfaces
of the deprotonated chromophore, in a vacuum (left) and in the
protein (right), for the ground (blue) and excited states (red)
along with the path of a typical MD trajectory (yellow line).
The multi-dimensional surfaces are projected onto the isomer-
ization coordinate and the skeletal deformation coordinate. The
first mode mainly involves a 180° rotation around torsion b
(Figure 1), while the latter describes concerted contractions and
expansions of the bonds between the atoms in the chromophore.
The “shape” of the ground (S0) and excited (S1) state surfaces
differ since these states have different electronic structures and
charge distributions. At the S1global minimum configuration,
where the double bond torsion angle (tosion b in Figure 1) is
90°, the overall negative charge is located near the thioester
group in the ground state but on the phenol ring in the excited
state, so the ground and excited states are affected differently
by the protein environment.
The position of the “seam” of the S1/S0intersection26of the
ground and excited states (see Figure 2) controls the passage
of the trajectory from the excited state to the ground state. The
barrier separating the planar energy minimum, near the FC
region, from the S1global minimum determines the efficiency
of the trans-to-cis isomerization motion. In the protein environ-
ment (Figure 2b), the energy at the global minimum is lowered
relative to that in vacuo (Figure 2a), resulting in (a) a decrease
of the S1-S0energy gap in the region of the twisted intermediate
(from 80 kJ mol-1in vacuo to less than 1 kJ mol-1in the
protein), (b) a displacement of the crossing seam closer to the
global minimum, and (c) a decrease of the energy barrier
separating the early planar S1 minimum and the twisted S1
minimum. Analysis of the protein structure suggests that the
stabilization of S1in the protein is primarily due to the gua-
nidium group of Arg52 that lies just above the chromophore
ring (see Figure 1) and stabilizes the negative charge located
on the ring in the excited state. This conjecture was tested by
recomputing the S0/S1energy difference at the twisted geometry
(minimum on S1), for the mutant Arg52Gly. The S1-S0energy
gap is increased by a factor of 10, highlighting the importance
of Arg52 for the photochemical process. This finding is in
agreement with the experimental observation that the mutant
has a considerably longer fluorescence lifetime.27
In total, 14 QM/MM MD simulations were initiated at various
times from a ground-state trajectory. The results are in listed
Table 1. In the protein, the lifetime of the excited-state ranged
from 129 to 2293 fs (third column in Table 1). The ratio of the
number of successful isomerizations to the number of excited-
state trajectories is ∼0.3, close to the experimental quantum
yield1of 0.35. Fitting a simple exponential (exp[-t/τ]) to the
computed fluorescence lifetime data (second column of Table
1) yields a fluorescence decay time of τ ) 200 fs. This may be
compared to time-resolved fluorescence decay measurements
that show a fast decay time of 430 fs.27Statistically, the number
of trajectories is small but nevertheless yields a consistent
Figure 3 shows a plot of the torsion angles in the tail of the
chromophore during photoisomerization. There are three distinct
phases: (I) the evolution on S0, (II) excitation and evolution
on S1, and (III) decay to S0at the surface crossing and evolution
on the S0surface. Immediately after the excitation (t ) 3.0 ps,
Figure 3), the chromophore decays (via bond length relaxation,
see Figure 2) from the Franck-Condon region into a nearby
local trans minimum on S1. After ca. 0.4 ps (t ) 3.4 ps, Figure
3), a shallow barrier separating this local minimum from the
(26) Robb, M. A.; Garavelli, M.; Olivucci, M.; Bernardi, F. In ReViews in
Computational Chemistry; Lipkowitz, K. B., Boyd., D. B., Eds.; Wiley-
VCH Publishers: New York, 2000; pp 15, 87-146.
(27) Mataga, N.; Chosrowjan, H.; Shibata, Y.; Imamoto, Y.; Tokunaga, F. J.
Phys. Chem. B 2000, 104, 5191-5199.
Figure 2. Potential energy surfaces of the excited and ground states of the deprotonated chromophore in the trans-to-cis isomerization coordinate (torsion
b) and a skeletal deformation of the bonds in vacuo (a) and in the protein (b). The yellow line represents the path sampled in a typical trajectory. Motion
along the torsion b connects the chromophore planar FC geometry to the 90° “twisted minimum”. Nonradiative decay occurs along the “seam” between the
surfaces (conical intersection hyperline26). The electrostatic field of the protein stabilizes the excited state of the twisted chromophore, moving the seam
closer to the twisted S1minimum, and causes a decrease of the excited-state energy barrier (along torsion b), leading to the twisted S1minimum. The main
contribution to the stabilization comes from the Arg52 residue, which is located on top of the chromophore ring (Figure 1).
A R T I C L E S Groenhof et al.
4230 J. AM. CHEM. SOC.9VOL. 126, NO. 13, 2004
global twisted minimum (Figure 2) is crossed and the chro-
mophore relaxes further (via rotation of the double bond,
C2dC3, torsion b, by 90°) to the twisted S1minimum conforma-
tion. The system oscillates around this minimum (typically for
50-500 fs) until the seam is encountered (Figure 2b), and a
surface hop takes the system back to the ground state (t ) 3.776
ps). Nonradiative decay is facilitated in the protein because the
“seam” lies close (Figure 2) to the reaction path sampled in the
trajectory (dashed yellow line in Figure 2b).
Upon returning to the S0state, the chromophore relaxes back
either to the trans or to a cis conformation. In the latter case,
the relaxation also involves flipping of the thioester linkage
(Figure 1), in which the carbonyl oxygen atom rotates 180°
clockwise when viewed from the CR of Cys69. During this
process, the phenyl moiety of the chromophore remains in its
original plane, the local geometry being constrained by the
packing of this ring with nearby protein residues, most notably
Arg52 and Phe62. Figure 1b shows the geometry of the active
site after a successful isomerization. Animations of the trajec-
tories showing the process of isomerization are available as
For comparison, a series of simulations of the deprotonated
chromophore in a vacuum were initiated from the same set of
chromophore configurations (position and velocity) as those in
the protein. In vacuo, the evolution on S1after photoexcitation
involves the single bond C3-C1′ (torsion a), rather than the
double bond C2dC3 (torsion b). The barrier to double bond
torsion is higher than the barrier to single bond torsion, and
single bond torsion leads back to the original configuration
without a net conformational change. Thus, no trans-to-cis
photoproduct would have been observed in the simulations in
vacuo. On the time scale of the simulation, the system remains
trapped in a local S1minimum, associated with a 90° twisted
phenyl moiety (twisted C3-C1′ torsion, Figure 3, right-hand
side), and does not decay (within 5 ps), because in vacuo the
“seam” lies far from the path sampled in the trajectory.
In summary, the isomerization is enhanced in the protein by
altering the stability of the global S1minimum (i.e., by moving
the position of the S1/S0seam and lowering the trans-to-twisted
barrier on S1, see Figure 2), and by sterically constraining the
motion of the chromophore, which allows the isomerization of
the double bond (torsion b) to be favored over isomerization of
the single bond (torsion a). The phenyl ring of the chromophore
lies stacked between the side chains of Arg52 and Phe62. These
interactions significantly restrain the torsional motion of the ring
about the single bond and enhance the torsional motion about
the double bond. The increased population of the twisted global
minimum ultimately ensures a higher quantum yield.
Following isomerization, PYP enters the third stage of the
photocycle. The QM/MM MD trajectories described in the
previous paragraphs are now used as the initial conditions for
extended classical (ns) simulations of the system in the ground
state to study this third stage. Previous classical MD simulations
by Groenhof et al.24showed that isomerization of the chro-
mophore can lead to the transfer of a proton to the chromophore
and that this event stimulates the subsequent conformational
transition (unfolding) of the protein. The validity of the initial
configuration used for these simulations had, however, been
questioned.28In the present simulations, no significant confor-
mational changes occur in the protein within 2.0 ns after the
isomerization. The hydrogen-bond network between the chro-
mophore, Glu46, and Tyr42 (Figure 1) remains intact. The
proton affinity of the chromophore, however, increases sub-
stantially. This affinity is defined as the energy difference
between the protein configuration before and after the proton
is transferred from the Glu46 side chain to the chromophore.
Figure 4 shows the potential energy curve of the system as the
proton migrates from the Glu46 side chain to the phenolate
oxygen atom of the chromophore (O4′, Figure 3) in a typical
protein configuration before (pG) and after (pR) photoisomer-
ization. The difference between the two local minima, which
represent the two different protonation states, is plotted as a
function of time in Figure 5. Positive values imply that the
transfer is energetically unfavorable. In Figures 4a and 5a, we
show that, before isomerization, the energy required to transfer
the proton is around 100 kJ/mol, indicating that the transfer is
highly unlikely. In contrast, after isomerization (Figure 5b), the
energy fluctuates around zero, becoming negative at times (as
in Figure 4b). Although we have not computed the rate of proton
transfer in this case, it is known from quantum dynamical
simulations29-32that proton transfer is dominated by tunneling
during fluctuations that involve equal energies for the reactant
and product state, provided the distance between the donor and
acceptor is sufficently small. Because the hydrogen-bonding
network shown in Figure 1 remains intact throughout the
simulations, the donor-acceptor distance was always optimal
for proton transfer. This suggests a rapid transfer in the
isomerized state. For a complete analysis of the relationship
between the energy gap and the activation energy, we refer the
reader to the works by Warshel.29,30,33,34
The energy of proton transfer after the solvent and protein
had been allowed to relax in either protonation state was also
(28) Thompson, M. J.; Bashford, D.; Noodleman, L.; Getzoff, E. D. J. Am. Chem.
Soc. 2003, 125, 8186-8194.
(29) Warshel, A. Proc. Natl. Acad. Sci. U.S.A. 1984, 81, 444-448.
(30) Warshel, A. J. Phys. Chem. 1982, 86, 2218-2224.
(31) Warshel, A.; Chu, Z. T. J. Chem. Phys. 1990, 93, 4003-4015.
(32) Lensink, M. F.; Mavri, J.; Berendsen, H. J. C. J. Comput. Chem. 1999, 20,
(33) Warshel, A. Acc. Chem. Res. 2002, 35, 385-395.
(34) Warshel, A. Q. ReV. Biophys. 2001, 34, 563-679.
Table 1. Fluorescence Lifetimes, Time of the Surface Hops
(Excited-State Lifetimes), and Resulting Chromophore
Conformations in 14 Different QM/MM Simulationsa
runfluorescencelifetim e(fs) surfacehop(fs) result
aThe fluorescence Lifetime (second column) is defined as the time
interval in which the chromophore samples the local trans S1 minimum
(Figure 2), from where fluorescence can occur. In the Methods section, it
is explained how these fluorescence lifetimes were obtained from a
trajectory. The excited-state lifetime (third column) is defined as the total
amount of time spent in the excited state until a surface hop took place.
The quantum yield was computed by dividing the number of cis results by
the total number of simulations performed.
Photoactivation of the Photoactive YellowProteinA R T I C L E S
J. AM. CHEM. SOC. 9 VOL. 126, NO. 13, 2004 4231
computed (see Supporting Information). Including solvent and
protein relaxation, the energy was strongly positive for transfer
in the trans ground state, and strongly negative for transfer in
the cis ground state (after isomerization). This indicates that
proton transfer from Glu46 to the chromophore becomes
energetically favorable after isomerization. Despite some en-
tropic compensation, the transfer will also be thermodynamically
favored. Indeed, Yoda et al.35recently showed that the sign of
the pKadifference between the two groups changes from plus
to minus upon going from the trans to the cis ground state,
confirming our results. These authors also showed that the
polarization of the environment is an essential factor in this
reversal of proton affinity. We note that the concept of pho-
toisomerization leading to changes in the pKa’s of relevant
proton donors and acceptors in photoreceptor proteins had been
previously proposed by both Schulten,36,37who suggested the
pKa change was due to the twist of the chromophore, and
Warshel,38who suggested that the change is due to the change
in the chromophore environment. In a more quantitative treat-
ment of the third stage, one would want to compute the total
free energy of the proton-transfer reaction. Because the proton
transfer involves bond breaking and formation, one needs to
resort to a QM/MM treatment of the system in the free energy
calculation. Such an approach is far from trivial, although it
has been done by others.36Here, we restrict ourselves to pro-
(35) Yoda, M.; Inoue, Y.; Sakurai, M. J. Phys. Chem. B 2003, 107, 14569-
(36) Strabl, M.; Hong, G.; Warshel, A. J. Phys. Chem. B 2002, 106, 13333-
(37) Schulten, K. Nature 1978, 272, 85-86.
(38) Warshel, A. Photochem. Photobiol. 1979, 30, 285-291.
Figure 3. Behavior of the torsion angles in the tail of the chromophore during a photoisomerization simulation (run b in Table 1): CR-C?(brown), C?-Sγ
(yellow), Sγ-C1(blue), C1-C2(green), C2dC3(red), and C3-C1′(black). Stages: (I) before photoexcitation; (II) after photoexcitation, isomerization of the
double bond (torsion b); and (III) on the ground state. There is no torsional change about the C3-C1′single bond, and ground-state relaxation involves a
rotation of 180° in opposite directions around CR-C?and C?-Sγ, causing the thioester linkage (C?-Sγ-C1) to flip and the O1oxygen atom to rotate 180°
clockwise as viewed from the CRof Cys69.
Figure 4. Potential energy profiles for a proton-transfer reaction between the Glu46 side chain and the chromophore’s phenolate moiety (pCA, see Figure
1) in a typical protein configuration before (a) and after (b) photoisomerization of the chromophore. The reaction coordinate is defined as the distance from
the donor group, the Glu46 side chain. The total distance between the donor and acceptor remained around 2.8 Å with a small fluctuation of around 0.2 Å.
The offset of the curves is chosen such that the reactant minima coincide in both graphs. The green curve represents the reaction profile of the transfer in
the fully hydrated protein environment (QM/MM). The red curve shows the quantum mechanical contribution to the transfer profile (QM). The latter contribution
was computed by repeating the calculations on the donor-acceptor complex in vacuo. The protein stabilizes the product state with respect to the reactant
state both before (pG) and after the isomerization (pR), but only after the isomerization is the proton transfer energetically favorable.
A R T I C L E SGroenhof et al.
4232 J. AM. CHEM. SOC.9VOL. 126, NO. 13, 2004
viding a qualitative picture of the signal transduction mechanism
rather than attempting to predict rate constants. Furthermore,
as we demonstrated previously,24the proton transfer is not likely
to be the rate-limiting step in the formation of the signaling
state of the protein.
Upon transfer of the proton, the protein becomes unstable
and starts to unfold. In the classical (ns) simulations that started
from frames of the previous trajectories in which the proton
was manually transferred, the hydrogen-bonding network be-
tween the chromophore and the side chains of Tyr42 and Glu46
breaks up within a few picoseconds. The loss of these hydrogen
bonds is followed by structural changes in the protein, pre-
dominantly in the N-terminal region. These findings are in line
with the conclusions of the earlier work,24that the transfer of
the proton is the critical first step in the sequence of structural
changes that lead to signaling.
Understanding the dynamics of subpicosecond photoactivated
processes such as those that occur in PYP represents a major
challenge, but is essential to elucidating how proteins have
evolved to regulate such chemical processes. Here, we dem-
onstrated that, by using a QM/MM MD strategy with explicit
surface hopping, it is not only possible to reproduce experi-
mental quantities such as quantum yields and excited-state
lifetimes in PYP, but also to reveal the molecular mechanism
lying at the basis of the “catalytic” activity of the protein
environment on the PYP chromophore. Our results show that
(i) the protein promotes trans-to-cis isomerization of the
chromophore by restricting the rotation of its deprotonated
p-hydroxyphenyl moiety; that (ii) fast nonradiative decay is
enhanced in the protein by reducing the ground-excited state
energy gap in a fully twisted chromophore configuration; that
(iii) proton transfer is considerably facilitated in the ground-
state cis-configuration relative to the initial ground-state trans-
configuration; and that (iv) the proton transfer initiates confor-
mational changes within the protein, which in turn are believed
to lead to signaling. We believe that our work, but also that of
others,22,39-41has opened up the possibility to study a wide range
of photoactivated processes in proteins in atomic detail.
Acknowledgment. We thank Klaas Hellingwerf for raising
our interest in this project. We also thank him and Jocelyne
Vreede for stimulating discussions and their comments on this
manuscript. Frans van Hoesel is acknowledged for his help in
creating the images.
Supporting Information Available: An outline of alternative
approaches for computing the proton-transfer energies (PDF)
and two animations (MPEG) of the photoisomerization trajec-
tories. This material is available free of charge via the Internet
(39) Hayashi, S.; Tajkhorshid, E.; Schulten, K. Biophys. J. 2003, 85, 1440-
(40) Kraulis, P. J. J. Appl. Crystallogr. 1991, 24, 946-950.
(41) Merrit, E. A.; Bacon, D. J. Methods Enzymol. 1997, 277, 505-524.
Figure 5. Affinity of the chromophore for the proton of the Glu46 side chain (Figure 1) before (a) and after (b) the isomerization.19The energy is plotted
that is required to transfer the proton from the Glu46 side chain to the chromophore’s oxygen atom (O4′, Figure 3, right-hand side). Positive values mean
that the transfer costs energy, while a negative value indicates the transfer is energetically favorable. Before the isomerization, proton transfer is always
unfavorable and not likely to take place, but afterward, the energy fluctuates around zero, becoming negative very often, indicating that a proton transfer is
favorable and hence more likely to occur.
Photoactivation of the Photoactive YellowProteinA R T I C L E S
J. AM. CHEM. SOC. 9 VOL. 126, NO. 13, 2004 4233