Fluorescent protein barrel fluctuations and oxygen diffusion pathways in mCherry

Abstract

Fluorescent proteins (FPs) are valuable tools as biochemical markers for studying cellular processes. Red fluorescent proteins (RFPs) are highly desirable for in vivo applications because they absorb and emit light in the red region of the spectrum where cellular autofluorescence is low. The naturally occurring fluorescent proteins with emission peaks in this region of the spectrum occur in dimeric or tetrameric forms. The development of mutant monomeric variants of RFPs has resulted in several novel FPs known as mFruits. Though oxygen is required for maturation of the chromophore, it is known that photobleaching of FPs is oxygen sensitive, and oxygen-free conditions result in improved photostabilities. Therefore, understanding oxygen diffusion pathways in FPs is important for both photostabilites and maturation of the chromophores. In this paper, we use molecular dynamics calculations to investigate the protein barrel fluctuations in mCherry, which is one of the most useful monomeric mFruit variant. We employ implicit ligand sampling to determine oxygen pathways from the bulk solvent into the mCherry chromophore in the interior of the protein. We also show that these pathways can be blocked or altered and barrel fluctuations can be reduced by strategic amino acid substitutions.

Full text

THE JOURNAL OF CHEMICAL PHYSICS 135, 235101 (2011)
Fluorescent protein barrel fluctuations and oxygen diffusion pathways
in mCherry
Prem P. Chapagain,a) Chola K. Regmi, and William Castillo
Department of Physics, Florida International University, Miami, Florida 33199, USA
(Received 1 August 2011; accepted 19 October 2011; published online 15 December 2011)
Fluorescent proteins (FPs) are valuable tools as biochemical markers for studying cellular processes.
Red fluorescent proteins (RFPs) are highly desirable for in vivo applications because they absorb
and emit light in the red region of the spectrum where cellular autofluorescence is low. The naturally
occurring fluorescent proteins with emission peaks in this region of the spectrum occur in dimeric
or tetrameric forms. The development of mutant monomeric variants of RFPs has resulted in sev-
eral novel FPs known as mFruits. Though oxygen is required for maturation of the chromophore, it
is known that photobleaching of FPs is oxygen sensitive, and oxygen-free conditions result in im-
proved photostabilities. Therefore, understanding oxygen diffusion pathways in FPs is important for
both photostabilites and maturation of the chromophores. In this paper, we use molecular dynamics
calculations to investigate the protein barrel fluctuations in mCherry, which is one of the most useful
monomeric mFruit variant. We employ implicit ligand sampling to determine oxygen pathways from
the bulk solvent into the mCherry chromophore in the interior of the protein. We also show that these
pathways can be blocked or altered and barrel fluctuations can be reduced by strategic amino acid
substitutions. © 2011 American Institute of Physics. [doi:10.1063/1.3660197]
I. INTRODUCTION
Fluorescent proteins (FPs) are valuable tools in molec-
ular and cell biology and are used as biochemical markers
for studying cellular processes.1–3Red fluorescent proteins
(RFPs) are highly desirable for in vivo applications because
they absorb and emit light in the red region of the spectrum
where cellular autofluorescence is low.4However, the natu-
rally occurring fluorescent proteins with emission peaks in
this region of the spectrum occur in dimeric or tetrameric
forms,5,6which tend to oligomerize7,8and render them un-
suitable for fusion tagging.9The development of mutant
monomeric variants of RFPs to avoid these issues has re-
sulted in several novel monomeric FPs known as mFruits.10
Some of the most promising mFruits are mCherry, mOrange,
and mStrawberry11 and their names reflect the wavelengths of
their corresponding emission spectra.
Though oxygen is required for maturation of the chro-
mophore, it is known that photobleaching of FPs is oxy-
gen sensitive, and oxygen-free conditions result in improved
photostabilities.12 This poses limitations to the next genera-
tion of single-molecule spectroscopy and low-copy fluores-
cence microscopy experiments and therefore, improving the
photostability of the mFruits is highly desirable. The photo-
bleaching of the monomeric variants of RFPs has been at-
tributed to the lack of proper shielding against oxygen or
other molecules by the beta barrel surface.7Increasing evi-
dence suggests that protein flexibility plays a major role in
gas access into many proteins13–18 and the dynamic fluctu-
ations in the size of transient cavities due to residues’ ther-
mal fluctuations are the determining factor in the pathways
a)Author to whom correspondence should be addressed. Electronic mail:
chapagap@fiu.edu.
of gas diffusion.19–22 Oxygen diffusion in myoglobin’s dis-
tal pocket has been extensively studied, both experimentally
and by simulations, in light of the influence of different pro-
tein conformations or mutations.14,23 In FPs, the interaction
between the chromophore and the surrounding protein has
important implications for both structures.24 The electronic
molecular orbitals of the chromophore that are responsible
for its spectral properties may be modified by the surround-
ing protein. Also, protein barrel fluctuations and the motion
of residues can affect the spectral properties and the lifetime
of the fluorescence.25
Recent developments in efficient computational sampling
methods have allowed thorough scanning of the possible path-
ways for gas diffusion in the interior of proteins.26–29 For ex-
ample, such computational investigations have proved use-
ful in understanding gas diffusion in many protein systems
such as molecular dioxygen pathways via dynamic oxygen
access channels in flavoproteins,30–32 ammonia transport in
carbamoyl phosphate synthetase,33–35 and gas diffusion and
channeling in hemoglobin.28,36 In this paper, we use explicit
solvent all-atom molecular dynamics (MD) simulations to in-
vestigate the protein barrel fluctuations in mCherry, which is
one of the most useful monomeric variants of RFP. We com-
pare the barrel fluctuations in mCherry to those in citrine,
a yellow variant (YFP) of green fluorescent protein (GFP).
Although citrine and mCherry belong to different FP fami-
lies and the photobleaching mechanisms can be different, we
compare the barrel structural integrity of these proteins due
to two main reasons. First, citrine is a natural monomer and
a GFP homologue of mCherry with a similar barrel structure
and second, it is the most useful FP among all YFPs due to
its reduced halide sensitivity and improved photstability.37,38
Other YFPs are very sensitive to halides due to easy ion
0021-9606/2011/135(23)/235101/6/$30.00 © 2011 American Institute of Physics135, 235101-1

235101-2 Chapagain, Regmi, and Castillo J. Chem. Phys. 135, 235101 (2011)
FIG. 1. (a) Superposition of ribbon structures of red (mCherry) and yellow
(citrine) fluorescent proteins. (b) The β7-β10 region is displayed with a space
filling model which shows that the gap in mCherry is larger than in citrine.
access via a solvent channel or cavity formed close to the
dimer interface.39,40 In citrine, this cavity is filled by a mu-
tation Gln69Met preventing the access to the ion.37 A simi-
lar effect is desired in mCherry. We employ implicit ligand
sampling (ILS) to determine oxygen pathways from the bulk
solvent into the mCherry active site. Using these results as
a guide, we show that the barrel fluctuations and the oxygen
pathways can be altered or blocked with strategic amino acid
substitutions.
Following the folding of the protein, the chromophore
formation involves cyclization of tripeptide and oxidation,
which requires molecular oxygen.41 Therefore, the maturation
times can depend on the accessibility of molecular oxygen.
For example, it is shown that a water-filled pore was essen-
tial for fast maturation of TurboGFP chromophore.42 The pore
that leads from outside of the barrel to the inside chromophore
possibly facilitates molecular oxygen entry. Upon comparing
the crystal structures of GFPs and mFruits, structural differ-
ences in the beta barrels are observed. The tetramer subunit
interactions present in the naturally occurring red fluores-
cent protein DsRed are not present in the mFruit monomeric
forms and therefore the latter is expected to have less struc-
tural integrity. The crystal structures show larger openings
in the mFruits’ protein structure, which may be transiently
increased further by more pronounced thermal fluctuations.
These larger openings may allow oxygen to pass more eas-
ily to the chromophore which may have implications to both
chromophore maturation speed as well as photobleaching due
to oxidation.
Figure 1(a) is a superposition of ribbon structures of the
RFP mCherry (PDB code 2H5Q) and its GFP homologue cit-
rine (PDB code 1HUY). Figure 1(b) displays a space-filling
model of the β7-β10 region and shows that the gap between
β7 and β10 is smaller in citrine. We show that differences in
this region create pathways in mCherry for dioxygen diffusion
through the barrel to the chromophore.
II. METHODS
A. Molecular dynamics computations
The time series trajectories were obtained from explicit
solvent, all-atom simulations using the CHARMM27 force
field.43 The initial x-ray crystallographic structures of RFP
mCherry (pdb code 2H5Q) and YFP citrine (pdb code 1HUY)
have a few missing amino acid residues. The missing amino
acid residues were completed using MODELLER.44 For the
chromophore representation, we used pre-cyclized forms that
are composed of three amino acids each, Met66-Tyr67-Gly68
for mCherry and Gly65-Tyr66-Gly67 for citrine. Although
the fluorescent spectral properties are only possible in ma-
ture chromophores, the simpler representative forms of chro-
mophores can be used for the purpose of investigating the
barrel fluctuations.45
The MMTSB toolset46 was used to set up the system
for simulations. The initial structures of mCherry and cit-
rine were separately solvated in octahedral boxes under pe-
riodic boundary conditions with TIP3P water molecules with
a box cut-off of 10 Å. For mCherry, 11 319 water molecules
were used, and for citrine, 9290 water molecules were used.
All water molecules overlapping with the protein were re-
moved. The particle mesh Ewald method47 was used to treat
long range interactions with a 9-Å non-bonded cutoff. Energy
minimization was performed using the adopted basis
Newton–Raphson (ABNR) method.43 Each system was then
neutralized by adding sodium counter ions: six sodium ions
for RFP and 8 sodium ions for YFP. Water molecules that
overlapped with the sodium ions were removed and ABNR
energy minimization was performed again. The systems were
then heated with a linear gradient of 50 K/ps from 50 K to
300 K. At 300 K, the systems were equilibrated for 2 ns with
a 2 fs integration time step in the NVT (constant number, vol-
ume, and temperature) ensemble. The SHAKE algorithm48
was used to constrain the bonds connected with hydrogen
atoms. This was followed by a 10 ns NVT dynamics simu-
lation with 2 fs time steps for each protein that was used for
analysis.
B. Implicit ligand sampling for molecular oxygen
The ILS (Ref. 28) is a computational method, which
computes the potential of mean force (PMF) corresponding
to the placement of a given small ligand such as O2and CO,
everywhere inside the protein. The calculated PMF describes
the Gibb’s free energy cost of having a particle located at
a given position, integrated over all degrees of freedom of
the system, except ligand position. Calculated PMF values
also show the area accessible to the ligand with the associ-
ated free energy cost. We applied PMF/ILS calculations to
the frames from our MD simulations to determine locations
in a protein that are especially important for blocking, or fa-
cilitating oxygen passage, and to quantify the differences at
these locations between FP variants. A total of 5000 protein
conformations from a 10-ns MD trajectory were used for lig-
and sampling. Therefore, the free-energy value at each of the
locations is the average obtained from ILS performed every
2 ps for the 10-ns MD simulation trajectory. For the free en-
ergy calculation at each location, 20 different rotational ori-
entations of molecular dioxygen were sampled at each gird
position with a volume element size of 1 Å3. The free energy
is compared to a dioxygen molecule placed outside the protein
in the surrounding water, where the free energy is defined to
be zero. In the figures, all locations with a free energy below

235101-3 Oxygen diffusion pathways in mCherry J. Chem. Phys. 135, 235101 (2011)
−2.0 kcal/mol are colored red, and all locations with a free
energy above +10.0 kcal/mol are colored blue. The values
of the free energy as a function of reaction coordinate were
calculated for specific positions separated by ∼1 Å distance
along the pathways, extending from outside the protein in
the solvent, into the protein and leading to the chromophore.
The pathways were determined from visual inspection as well
as from the 3D grid data of free-energy values from ILS
simulations.
III. RESULTS AND DISCUSSION
Results from 10 ns MD simulations comparing the
monomeric mCherry RFP to the citrine YFP are shown in
Fig. 2. Figure 2(a) displays the root-mean-square fluctua-
tions (RMSF) of each amino acid. The β7 strand in mCherry
(amino acids 141–151) displays a significantly larger RMSF
than β7 in citrine (amino acids 147–157), whereas for strand
β10 the RMSF is approximately the same for mCherry (193–
204) and citrine (199–208).
To further investigate the possibility of oxygen access in
this region, we next focus specifically on the opening between
strands β7 and β10, which is the region that is postulated to
allow oxygen entry for mCherry. Figure 2(b) is a time series of
the separation between strands β7 and β10. The plots repre-
sent the separation r between a pair of residues, one residue
on β7 and the other on β10. In order to characterize the size
of the gap, an atom is chosen on each residue that is clos-
FIG. 2. Results from 10 ns MD simulations comparing the citrine YFP to
the monomeric mCherry RFP. (a) Root-mean-square fluctuations (RMSF) of
each amino acid. (b) Time series of the separation between strands β7and
β10. Strands β7andβ10 show a large and fluctuating separation in mCherry
as compared to the small, fixed separation in citrine.
est to the other residue across the gap. For mCherry, ris
the distance between the Cαof residues Ala145 and Lys198,
and for citrine, r is the distance between the Cαof residues
Ser147 and Gln204. Not only is the β7-β10 gap always big-
ger in mCherry than citrine throughout the 10 ns, the gap in
mCherry also exhibits larger fluctuations.
A. Dioxygen access routes to the chromophore
in mCherry
The large and fluctuating gap between strands β7 and
β10 displayed in Fig. 2for mCherry but not for citrine, makes
this region a prime candidate for dioxygen access in mCherry.
To determine which regions or pathways allow the molecular
oxygen to enter the protein barrel, we calculated the free en-
ergy of placing molecular oxygen in and around the entire
protein barrel using ILS, which uses PMF calculations to de-
termine the free energy of placing a small molecule such as
dioxygen at a specific location in a protein. Figure 3(a) dis-
plays ensemble-averaged free-energy diagrams for a dioxy-
gen molecule if it is placed in, or around, mCherry and com-
pare that with citrine in Fig. 3(b). We calculated the free en-
ergy of the dioxygen using the ILS routine implemented in
the VMD molecular dynamics package.49 In Fig. 3(a),wedis-
play a slice that includes the β7 region. The color red repre-
sents low free-energy (<−2 kcal/mole) locations, white rep-
resents intermediate free energy locations, and blue represents
high free-energy (>+10 kcal/mol) locations. In order for the
chromophore to have access to molecular oxygen, a pathway
without substantial free-energy barriers must exist from the
region outside the protein, through the protein barrel, to the
chromophore location in the interior. Figure 3(a) shows that
mCherry displays two low free-energy routes through the bar-
rel: one through the β7-β10 gap (R1) and the other through
the β7-β8 gap (R2). These two entry routes for dioxygen
lead all the way to the chromophore. In contrast, citrine has
no easy pathway, including the β7-β10 region (Y1) or the
FIG. 3. Free-energy isosurfaces for molecular dioxygen in (a) citrine and (b)
mCherry. The free-energy slice shown includes the β7-β10 region as well as
the β7-β8 region. The color red represents low free-energy (<−2 kcal/mole)
locations, blue represents high free-energy (>+10 kcal/mol) locations, and
white represents locations for which the oxygen has intermediate free energy.
Neither the β7-β10 region nor the β7-β8 region in citrine offers low free-
energy routes for dioxygen entry, whereas in mCherry both regions display
gaps representing low free-energy access routes.

235101-4 Chapagain, Regmi, and Castillo J. Chem. Phys. 135, 235101 (2011)
FIG. 4. Free-energy values of dioxygen at locations along the pathways
showninFig.3leading from the solvent outside the protein (9 Å) into the
chromophore (0 Å). The mCherry has two easy routes that can be accessed
by entering through either the β7-β10gap(R1)ortheβ7-β8 gap (R2). The
routes for citrine through either the β7-β10 (Y1) region or β7-β8(Y2)region
are blocked by a high free-energy barrier.
β7-β8 region (Y2) both of which involve high free-energy
(blue) barriers.
Figure 4quantifies the value of the free energy along the
pathways shown in Fig. 3. The reaction coordinate is the po-
sition of the oxygen molecule along the route. The oxygen
molecule is placed at steps along the path that are separated
by 1 Å. The coordinate 0 represents a location near the chro-
mophore and 9 Å represents a location outside the protein in
the solvent. It is seen in this figure that both routes in citrine
(Y1, Y2) face large free energy barriers due to the protein
barrel, whereas there is no substantial barrier for either of the
pathways in mCherry (R1, R2). The identification of these
pathways is used later to guide mutations of key residues in
order to create barriers in mCherry to block these routes and
prevent dioxygen access to the chromophore.
B. Importance of sidechains in controlling gap size
As discussed in the Introduction, the β7-β10 gap and the
β7- β8 gap in mCherry are large enough to allow dioxygen to
pass through the protein barrel to the chromophore. An aim of
this work is to determine if site-specific amino acid mutations
can alter these routes. In order to ascertain more details of the
structural fluctuations in the barrel, we determined if the large
fluctuations in the mCherry β7-β10 gap is due to motion of
strand β7 or strand β10, and similarly for the β7-β8 gap.
In comparing the time series of the fluctuations in the
size of the mCherry β7-β10 gap (Fig. 2(b)) and the β7-β8
gap, we found that the openings and closings of the gaps were
out of phase with each other. When the β7-β10 gap is large,
the β7-β8 is small, and vice versa, implying that the fluctu-
ations in the β7-β10 gap and the β7-β8 gap are mostly due
to movement of β7. In addition, to provide more information
for guiding the mutations, we wish to determine why Figs. 3
and 4both show that the dioxygen pathway through the β7-
β8 gap (R2) is not as easy as the pathway through the β7-β10
gap (R1) even though the backbone separations are similar. In
Fig. 5we show that amino acid sidechains play an important
role in closing the β7-β8 gap. Figure 5displays the results
FIG. 5. Fluctuations in the β7-β8 gap in mCherry determined by the distance
between Cαatoms of β7-Ala145 and β8-Gln163 (darker line) compared to
the separation determined by the distance between Cαon β7-Ala145 and
the N on the sidechain of β8-Gln163 (lighter line). The sidechain of Gln163
narrows the gap significantly more than the backbone.
of 10 ns MD simulations for the separation between strands
β7 and β8 determined in two different ways. For both time
series, the separation is measured between Ala145 on β7 and
Gln163 on β8. One plot shows the time series of fluctuations
in the separation between the amino acids’ Cαatoms. The
other time series displays the fluctuating distance between the
Cαof Ala145 (on β7) and the N on the sidechain of Gln163
(on β8). Figure 5shows that the sidechain of Gln163 narrows
the gap significantly more than the backbone.
Figure 6further quantifies the importance of sidechains
for determining gap sizes. In Fig. 6, we present a contour
plot of the free energy of Ala145 on β7 and Gln163 on β8
as a function of their separation, measured in the same two
ways that are used in Fig. 5. The vertical axis is the sepa-
ration between the Cαof residue β7-Ala145 and the Cαof
residue β8-Gln163 (dark line in Fig. 5) and the horizontal
axis is the separation between the Cαof β7-Ala145 and the
N on the sidechain of residue β8-Gln163 (light line in Fig. 5).
Figure 6shows that there are two islands of low free energy,
which implies that β7 is stable at two distinct separations
from β8, with the larger separation meaning that β7iscloser
to β10. Another important feature is that when β7 is further
FIG. 6. Free-energy plot (in kcal/mol) of the β7-β8 strands. The horizon-
tal axis is the separation between the Cαon β7-Ala145 and the N on the
sidechain of β8-Gln163, the vertical axis is the separation between the Cαof
β7-Ala145 and the Cαof β8-Gln163. There are two distinct islands of low
free energy, showing that the β7 strand spends most time in these two distinct
positions. Additionally, when the Cα-Cαseparation is large, the sidechain un-
dergoes larger fluctuations, which restricts the gap from opening widely.

235101-5 Oxygen diffusion pathways in mCherry J. Chem. Phys. 135, 235101 (2011)
from β8, the range of fluctuations in the sidechain of β8is
also larger. This allows the large side chain of β8 to partially
close the gap even when β7 is far away.
C. Amino acid mutations in mCherry to control
dioxygen access to the chromophore
Figures 3and 4show that the easiest pathway for oxygen
access in mCherry is through the gap between β7 and β10,
and Figs. 5and 6show that sidechains play a role in clos-
ing the β7-β8 gap. Therefore, our aim in making amino acid
mutations is to decrease the β7-β10 gap without substantially
increasing the size of the gap between β7 and β8.
On comparing amino acid properties of the residues in
the β7, β8, and β10 strands of mCherry and citrine, it is seen
that there are more charged residues in mCherry as compared
to just two in β8 of citrine. The citrine residues in the region of
interest are mostly polar (Ser, Tyr, Asn, His) or hydrophobic
(Ala, Val, Phe, Ile, Leu). We first attempted a few mutations
in mCherry such that the charged amino acids are replaced
by polar or hydrophobic residues, similar to the pattern in cit-
rine. This change, however, either made the fluctuations worse
or the β7-β10 gap increased even further. Since the β7-β10
gap fluctuates the most, a possible strategy to reduce this fluc-
tuation is to create appropriate ionic interactions. In this re-
gion of mCherry, the inter-strand charged amino acid residues
participate in inter-strand ionic interactions and give rise to
salt-bridges. For example, the attractive interaction between
Glu144(-) in β7 and 164Arg(+)inβ8 swings β7 towards β8
and helps to open the gap between β7 and β10. To reduce
the barrel flexibility in this region, we made two amino acid
replacements (one in β7 and other in β8): we replaced the
polar 143Trp in β7 by a positively charged 143Lys(+), and
the 164Arg(+)inβ8 was replaced by 164Glu(-). The mu-
tations introduce two possible electrostatic interactions that
might close the β7-β10 gap. The attraction between the mu-
tated β7 143Lys(+) and β10 200ASP(-) pulls β7 towards
β10, and the repulsion between β7 Glu144(-) and the mu-
tated β8 164Glu(-) pushes β7 away from β8 and towards β10.
Since the location of the gap is close to the original tetramer-
breaking mutations, the barrel folding in monomeric form can
be sensitive to new mutations such as the one presented here.
A new set of mutations must still allow the barrel to fold and
chromophore to mature. If this is achieved, the marked reduc-
tion in the barrel fluctuations that limit the oxygen entry may
result in a more photostable FP.
The success of the mutations in closing the β7-β10 gap
in the mutated RFP (mut-RFP) is shown in Fig. 7.Thetwo
curves display time series for mCherry (same curve as in
Fig. 2(b)) and mut-RFP for the β7-β10 gap. The β7-β10 gap
in Fig. 7for the mut-RFP starts out at ∼7.5 Å, which is as
large as it ever gets for mCherry. This is expected because
the initial positions for the atoms in our proposed mut-RFP
are given by the PDB file for mCherry. Within a short time,
Fig. 7(a) shows that new interactions in mut-RFP greatly re-
duce the β7-β10 gap compared to mCherry which signifi-
cantly reduces the ease of oxygen permeability as displayed
by the free-energy isosurface in Fig. 7(b).
FIG. 7. (a) Results from the 10 ns MD simulation of the β7-β10 gap for
mCherry (red, same curve as in Fig. 2(b)) compared to its mutant (purple).
The β7-β10 gap in mut-RFP is greatly reduced compared to mCherry. (b)
Free-energy isosurface for molecular dioxygen in mut-RFP. As compared to
the isosurface displayed in Fig. 3(b) for mCherry, mut-RFP isosurface shows
significantly less favorable pathways for oxygen entry with high free-energy
barriers (blue).
We have used MD calculations to determine the pathways
for molecular oxygen entry through the barrel of mCherry. We
have shown that specific point mutations can alter the oxygen
pathways in the RFPs. Blocking or altering these pathways
through the barrel can have an effect on FP maturation as
well as on its photostability. For example, easy oxygen access
may significantly reduce the photostability whereas it may be
useful for chromophore maturation, especially at low oxygen
conditions. The computational approach can provide impor-
tant insights for guiding efficient mutagenesis experiments to
improve the maturation speed and photostability of mFruits.
ACKNOWLEDGMENTS
This work was supported in part by Award No.
SC3GM096903 from the National Institute of General Medi-
cal Sciences. The content of this publication is solely the re-
sponsibility of the authors and does not necessarily represent
the official views of the National Institute of General Medical
Sciences or the National Institutes of Health.
We thank Professor Ralph Jimenez at the University of
Colorado at Boulder for bringing this topic to our attention.
We also thank Professor Jimenez for critical reading of the
paper and many helpful discussions.
1R. Y. Tsien, Annu. Rev. Biochem. 67, 509 (1998).
2M. Zimmer, Chem. Rev. 102(3), 759 (2002).
3M. Chalfie, Y. Tu, G. Euskirchen, W. W. Ward, and D. C. Prasher, Science
263(5148), 802 (1994).
4A. Goulding, S. Shrestha, K. Dria, E. Hunt, and S. K. Deo, Protein Eng.
Des. Sel. 21(2), 101 (2008).
5M. V. Matz, A. F. Fradkov, Y. A. Labas, A. P. Savitsky, A. G. Zaraisky,
M. L. Markelov, and S. A. Lukyanov, Nat. Biotechnol. 17(10), 969 (1999).

235101-6 Chapagain, Regmi, and Castillo J. Chem. Phys. 135, 235101 (2011)
6D. Yarbrough, R. M. Wachter, K. Kallio, M. V. Matz, and S. J. Remington,
Proc. Natl. Acad. Sci. U.S.A. 98(2), 462 (2001).
7R. E. Campbell, O. Tour, A. E. Palmer, P. A. Steinbach, G. S. Baird,
D. A. Zacharias, and R. Y. Tsien, Proc. Natl. Acad. Sci. U.S.A. 99(12),
7877 (2002).
8G. S. Baird, D. A. Zacharias, and R. Y. Tsien, Proc. Natl. Acad. Sci. U.S.A.
97(22), 11984 (2000).
9E. M. Merzlyak, J. Goedhart, D. Shcherbo, M. E. Bulina, A. S. Shcheglov,
A. F. Fradkov, A. Gaintzeva, K. A. Lukyanov, S. Lukyanov, T. W. Gadella,
and D. M. Chudakov, Nat. Methods 4(7), 555 (2007).
10N. C. Shaner, R. E. Campbell, P. A. Steinbach, B. N. Giepmans,
A. E. Palmer, and R. Y. Tsien, Nat. Biotechnol. 22(12), 1567 (2004).
11N. C. Shaner, P. A. Steinbach, and R. Y. Tsien, Nat. Methods 2(12), 905
(2005).
12N. C. Shaner, M. Z. Lin, M. R. McKeown, P. A. Steinbach, K. L. Hazel-
wood, M. W. Davidson, and R. Y. Tsien, Nat. Methods 5(6), 545 (2008).
13P. Amara, P. Andreoletti, H. M. Jouve, and M. J. Field, Protein Sci. 10(10),
1927 (2001).
14M. L. Carlson, R. M. Regan, and Q. H. Gibson, Biochemistry 35(4), 1125
(1996).
15C. Bossa, M. Anselmi, D. Roccatano, A. Amadei, B. Vallone, M. Brunori,
and A. Di Nola, Biophys. J. 86(6), 3855 (2004).
16D. C. Lamb, A. Arcovito, K. Nienhaus, O. Minkow, F. Draghi, M. Brunori,
and G. U. Nienhaus, Biophys. Chem. 109(1), 41 (2004).
17Q. H. Gibson, R. Regan, R. Elber, J. S. Olson, and T. E. Carver, J. Biol.
Chem. 267(31), 22022 (1992).
18M. Brunori and Q. H. Gibson, EMBO Rep. 2(8), 674 (2001).
19J. Cohen, K. Kim, P. King, M. Seibert, and K. Schulten, Structure 13(9),
1321 (2005).
20V. A. Feher, E. Baldwin, and F. W. Dahlquist, Nat. Struct. Biol. 3, 516
(1996).
21J. R. Lakowicz and G. Weber, Biochemistry 12(21), 4171 (1973).
22W. Nadler and D. L. Stein, Proc. Natl. Acad. Sci. U.S.A. 88(15), 6750
(1991).
23D. B. Calhoun, J. M. Vanderkooi, G. V. Woodrow III, and S. W. Englander,
Biochemistry 22(7), 1526 (1983).
24B. R. Branchini, A. R. Nemser, and M. Zimmer, J. Am. Chem. Soc. 120(1),
1 (1998).
25X. Shu, N. C. Shaner, C. A. Yarbrough, R. Y. Tsien, and S. J. Remington,
Biochemistry 45(32), 9639 (2006).
26R. G. Justin, B. Rosemary, and S. Klaus, J. Comput. Phys. 151(1), 190
(1999).
27B. Roux, Comp. Phys. Commun. 91(1–3), 275 (1995).
28J. Cohen, A. Arkhipov, R. Braun, and K. Schulten, Biophys. J. 91(5), 1844
(2006).
29Y. Wang, J. Cohen, W. F. Boron, K. Schulten, and E. Tajkhorshid,
J. Struct. Biol. 157(3), 534 (2007); B. J. Johnson, J. Cohen, R. W. Welford,
A. R. Pearson, K. Schulten, J. P. Klinman, and C. M. Wilmot, J. Biol. Chem.
282(24), 17767 (2007).
30J. Saam, I. Ivanov, M. Walther, H. G. Holzhutter, and H. Kuhn, Proc. Natl.
Acad. Sci. U.S.A. 104(33), 13319 (2007).
31J. Saam, E. Rosini, G. Molla, K. Schulten, L. Pollegioni, and S. Ghisla,
J. Biol. Chem. 285(32), 24439 (2010).
32R. Baron, C. Riley, P. Chenprakhon, K. Thotsaporn, R. T. Winter,
A. Alfieri, F. Forneris, W. J. van Berkel, P. Chaiyen, M. W. Fraaije,
A. Mattevi, and J. A. McCammon, Proc. Natl. Acad. Sci. U.S.A. 106(26),
10603 (2009).
33Y. Fan, L. Lund, Q. Shao, Y. Q. Gao, and F. M. Raushel, J. Am. Chem. Soc.
131(29), 10211 (2009).
34L. Lund, Y. Fan, Q. Shao, Y. Q. Gao, and F. M. Raushel, J. Am. Chem. Soc.
132(11), 3870 (2010).
35X. S. Wang, A. E. Roitberg, and N. G. Richards, Biochemistry 48(51),
12272 (2009).
36R. Daigle, M. Guertin, and P. Lague, Proteins 75(3), 735 (2009).
37O. Griesbeck, G. S. Baird, R. E. Campbell, D. A. Zacharias, and
R. Y. Tsien, J. Biol. Chem. 276(31), 29188 (2001).
38A. Miyawaki, T. Nagai, and H. Mizuno, Adv. Biochem. Eng. Biotechnol.
95, 1 (2005).
39R. M. Wachter, D. Yarbrough, K. Kallio, and S. J. Remington, J. Mol. Biol.
301(1), 157 (2000).
40A. Rekas, J. R. Alattia, T. Nagai, A. Miyawaki, and M. Ikura, J. Biol. Chem.
277(52), 50573 (2002).
41T. D. Craggs, Chem. Soc. Rev. 38(10), 2865 (2009).
42A. G. Evdokimov, M. E. Pokross, N. S. Egorov, A. G. Zaraisky, I. V. Yam-
polsky, E. M. Merzlyak, A. N. Shkoporov, I. Sander, K. A. Lukyanov, and
D. M. Chudakov, EMBO Rep. 7(10), 1006 (2006).
43B. R. Brooks, R. E. Bruccoleri, B. D. Olafson, D. J. States, S. Swaminathan,
and M. Karplus, J. Comp. Chem. 4(2), 187 (1983).
44N. Eswar, B. Webb, M. A. Marti-Renom, M. S. Madhusudhan, D. Eramian,
M. Y. Shen, U. Pieper, and A. Sali, Curr. Protoc. Bioinf. 15, 5.6-1 (2006).
45B. T. Andrews, S. Gosavi, J. M. Finke, J. N. Onuchic, and P. A. Jennings,
Proc. Natl. Acad. Sci. U.S.A. 105(34), 12283 (2008).
46M. Feig, J. Karanicolas, and C. L. Brooks 3rd, J. Mol. Graphics Modell.
22(5), 377 (2004).
47E. Ulrich, P. Lalith, L. B. Max, D. Tom, L. Hsing, and G. P. Lee, J. Chem.
Phys. 103(19), 8577 (1995).
48W. F. van Gunsteren and H. J. C. Berendsen, Mol. Phys. 34(5), 1311
(1977).
49W. Humphrey, A. Dalke, and K. Schulten, J. Mol. Graphics 14(1), 33
(1996).