Article

pubs.acs.org/JACS

Unraveling the Photoluminescence Response of Light-Switching

Ruthenium(II) Complexes Bound to Amyloid‑β
Nathan P. Cook,† Mehmet Ozbil,∥ Christina Katsampes,† Rajeev Prabhakar,*,∥ and Angel A. Martí*,†,‡,§
†
Department of Chemistry, ‡Department of Bioengineering, and §Institute of Biosciences and Bioengineering, Rice University,
Houston, Texas 77005, United States
∥
Department of Chemistry, University of Miami, Coral Gables, Florida 33146, United States
*
S Supporting Information
See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

ABSTRACT: Photoluminescent molecules are widely used for real-time monitoring of

peptide aggregation. In this Article, we detail both experimental and computational
modeling to elucidate the interaction between [Ru(bpy)2dppz]2+ and amyloid-β (Aβ1−40)
Downloaded via SIMON FRASER UNIV on March 23, 2020 at 19:23:47 (UTC).

aggregates. The transition from monomeric to fibrillar Aβ is of interest in the study of

Alzheimer’s disease. Concentration-dependent experiments allowed the determination of a
dissociation constant of 2.1 μM, while Job plots provided a binding stoichiometry of 2.6
Aβ monomers per [Ru(bpy)2dppz]2+. Our computational approach that combines
molecular docking (both rigid and flexible) and all-atom molecular dynamics (MD)
simulations predicts that the hydrophobic cleft between Val18 and Phe20 is a plausible
binding site, which could also explain the increase in photoluminescence of
[Ru(bpy)2dppz]2+ upon binding. This binding site is parallel to the fibril axis, in marked contrast to the binding site of these
complexes in DNA (perpendicular to the DNA axis). Other binding sites may exist at the edges of the Aβ fibril, but they are
actually of low abundance in an Aβ fibril several micrometers long. The assignment of the binding site was confirmed by binding
studies in an Aβ fragment (Aβ25−35) that lacked the amino acids necessary to form the binding site. The agreement between the
experimental and computational work is remarkable and provides a general model that can be used for studying the interaction of
amyloid-binding molecules to Aβ.

■ INTRODUCTION
Amyloid-β (Aβ) is a 39−43 amino acid-containing byproduct
the Aβ peptide difficult. Recently, structural models of Aβ have
been developed with help from NMR spectroscopy.5−7 These
of the amyloid precursor protein (APP) and is thought to play a models have been used in combination with computational
causative role in the progression of Alzheimer’s disease (AD).1 methods to examine the physical basis for probe binding,8−11
The amyloid cascade hypothesis suggests that the transition of analysis of potential inhibitors,12−16 and structural character-
monomeric Aβ into higher order aggregates is a driving factor istics of the aggregates and prefibrillar forms.17
in the progression of AD.2 Additionally, recent studies have The role of computer simulations regarding the aggregation
revealed that Aβ fibrils can template the formation of of Aβ was recently reviewed by Lemkul and Bevan.18 These
neurotoxic Aβ oligomers.3 Molecules capable of binding on simulations have helped elucidate probe−peptide complexes by
the surface of Aβ fibrils might be able to obstruct the access of identifying key residues and forces that foster these specific
Aβ monomers to its surface, inhibiting the templated formation interactions.8,10,11 Computational methods have been widely
of Aβ oligomeric species. The discovery that ruthenium used for studying the binding modes and interactions of dyes
dipyridophenazine complexes can bind to Aβ fibrils displaying such as Thioflavin-T (ThT)9,19 and Congo Red (CR)10,11
a marked increase in photoluminescence intensity is relevant toward Aβ. Furthermore, a recent publication by one of our
and timely,4 because contrary to most dyes for Aβ detection, laboratories reported the computational modeling of the
these ruthenium dyes are not planar and are easily modifiable. binding of cotinine with Aβ.20 Given that these molecules
These characteristics, combined with their ability to bind to Aβ, have a special affinity for amyloid fibrils, identifying binding
make them potential parent complexes for the production of sites could be helpful for developing drugs to prevent or reverse
compounds capable of inhibiting Aβ aggregation or to quench peptide aggregation.21 On the other hand, concentration-
the production of toxic Aβ oligomeric species induced by Aβ dependent biophysical studies have shed light on the stability of
fibrils. Nonetheless, this will require a profound understanding these interactions and the size of the binding site.22,23
of the interactions of ruthenium dipyridophenazine complexes In this Article, we combine biophysical and computational
and Aβ. studies to elucidate the binding modes of [Ru(bpy)2dppz]2+
It is important to recognize that one of the main challenges (bpy = 2,2′-bipyridine; dppz = dipyrido[3,2-a:2′,3′-c]-
of identifying binding sites on Aβ is the absence of high-
resolution crystalline structures of the Aβ aggregates, which Received: May 14, 2013
makes understanding the action of small molecule binding to Published: July 11, 2013

© 2013 American Chemical Society 10810 dx.doi.org/10.1021/ja404850u | J. Am. Chem. Soc. 2013, 135, 10810−10816
Journal of the American Chemical Society Article

phenazine) to Aβ1−40 fibrils. Ruthenium dipyridophenazine wavelength, and lp is a correction factor. The correction factor lp is
metal complexes have been used in a wide variety of given by:
applications including DNA detection,24 cell viability studies,25 obs
Iem = κA(λex )10−A(λex )l p (2)
solubilization of carbon nanotubes,26 and cell imaging,27−29 but
have rarely been used for studying peptides.30,31 Here, we where κ is a proportionality constant that contains instrument
report studies that have allowed us to generate a general picture parameters as well as the quantum yield of the metal complex. lp
to explain the interactions and light-switching behavior of was calculated to be 0.179 by fitting Figure S1 to:
[Ru(bpy)2dppz]2+ when in the presence of fibrillar Aβ peptides.
A(λex )
Understanding how these metal complexes bind to fibrillar Aβ log obs
= lpA(λex ) − log κ
has general implications in the design of amyloid probes, as well Iem (3)
as potential imaging and therapeutic agents for Alzheimer’s Binding Analysis. Once fibrillar Aβ content was determined,
disease.

■
aliquots of Aβ were diluted into metal complex samples in 20 mM
phosphate buffer, 0.01% NaN3, pH 7.4, and their photoluminescence
EXPERIMENTAL SECTION was determined. Photoluminescence intensities were corrected for
Preparation of Aβ. Bulk Aβ1−40 (lot no. 9596) was purchased inner filter effects. The dissociation constant was determined using the
from 21st Century Biochemicals and purified and stored following equation:37
previously reported methods.4
fAβ‐ Ru − fRu
Preparation of Fibrillar Aggregates. Aβ fibrils were formed PL = fRu Ru tot + (Ru tot + Aβtot + Kd
from a purified lyophilized powder by reconstituting the peptide in a 2
minimal amount of NaOH (2 mM NaOH adjusted to pH 10 with 100
mM NaOH).32 The dissolved peptide was then placed in a bath − (Ru tot + Aβtot + Kd)2 − 4Ru totAβtot ) (4)
sonicator for 2 min and filtered through 0.2 μm centrifuge filters 2+
where PL is the photoluminescence of the [Ru(bpy)2dppz] at
(VWR). After centrifugation, the Aβ solution was diluted with PBS
different concentrations of fibrillar Aβ, f Ru is a proportionality constant
(100 mM phosphate buffer, 300 mM NaCl, pH 7.4) to an approximate
that correlates [Ru(bpy)2dppz]2+ concentration with its photo-
volume of 600 μL, and the concentration was verified with an
luminescence intensity, fAß‑Ru is a proportionality constant that
absorption coefficient of 1280 M−1 cm−1 at 280 nm using a Shimadzu
correlates the concentration of the Aβ-[Ru(bpy)2dppz]2+ complex
2450 UV−vis spectrophotometer. A typical initial concentration was
with its photoluminescence intensity, Aβtot is the sum of the
between 150 and 170 μM. The Aβ solutions were incubated on a
concentration of the Aβ-[Ru(bpy)2dppz]2+ complex plus free Aβ,
Boekel orbital shaker at 37 °C and 700 rpm for 24 h.
Rutot is the sum of the concentration of the Aβ-[Ru(bpy)2dppz]2+
Aβ25−35 was purchased from 21st Century Biochemicals. Each fibril
complex plus free [Ru(bpy)2dppz]2+, and Kd is the dissociation
sample was prepared from a 1 mg/mL sample initially dissolved in a
constant. The reported Kd is an average of five independent
small amount of DMSO and then diluted to 1 mL with PBS buffer as
experiments with the error calculated from student’s t test at 80%
referenced above. Samples were placed in a Boekel orbital shaker and
confidence interval.
incubated at 37 °C and 400 rpm for 24 h. The photoluminescent
Job Plot Analysis.38 Fibrils (prepared as previously described)
experiments were performed by diluting Aβ25−35 and Aβ1−40 to 50 μM
were centrifuged for 30 min at 16 000g, and the supernatant was
with 5 μM [Ru(bpy)2dppz]2+ or 5 μM ThT. Transmission electron
analyzed by UV−vis absorption to obtain the concentration of
microscope (TEM) images were prepared by drop casting the fibril
nonfibrillar Aβ, which was determined to be around 5% of the original
solutions on glow discharged 200 mesh carbon type B coated copper
concentration. After taking into account nonfibrillized Aβ and solvent
grids (Ted Pella 01811) with 3% uranyl acetate applied as a negative
loss due to evaporation, the Aβ fibrils were diluted to a final
stain. Samples were subsequently imaged on a JEOL 1230 High
concentration of either 20, 50, or 100 μM. Photoluminescence was
Contrast TEM operating at 80 kV.
obtained by varying the metal complex−peptide ratio with fixed total
Synthesis of [Ru(bpy)2dppz]2+. [Ru(bpy)2dppz]2+ cis-Ru-
concentrations ([Ru(bpy)2dppz]2+ + Aβ) of 20, 50, or 100 μM. The
(bpy)2Cl2 was used as starting reagent (Strem Chemicals), and the
molar ratio of [Ru(bpy)2dppz]2+ is defined as the moles of
dppz ligand was synthesized following previous methods.33,34 The
[Ru(bpy)2dppz]2+ divided by the total moles in the solution, that is,
dppz ligand and cis-Ru(bpy)2Cl2 were refluxed in 1:1 methanol and
[Ru(bpy)2dppz]2+ + Aβ. The photoluminescence intensities were
water with vigorous stirring for 3 h, as described by Amouyal et al.33
corrected as described in the preceding section. The molar ratio of Aβ
After 3 h, the solution was concentrated under reduced pressure, and
for any point in Figure 1b can be obtained by subtracting the molar
upon the addition of ammonium hexafluorophosphate, orange-red
ratio of [Ru(bpy)2dppz]2+ from 1. The number of Aβ monomers per
crystals precipitated from solution. The crystals were purified by
ruthenium complex is calculated by dividing the molar ratio of Aβ by
column chromatography (4:1 dichloromethane and acetonitrile) on
the molar ratio of [Ru(bpy)2dppz]2+ obtained from the maximum of
SiliaFlash P60 silica gel from SiliCycle. The PF6− salt was used for all
Figure 1b.
binding experiments. Concentrated stock solutions were prepared in
Computational Methods. All three molecular docking proce-
acetonitrile and verified with an absorption coefficient of 16 300 M−1
dures were performed using the Autodock Vina 1.1.2 software39 on
cm−1 at 457 nm.35
Aβ1−40 fibril structures generously provided by Robert Tycko.5 For the
Photoluminescence Experiments. All steady-state photolumi-
docking simulations, the Ru2+ atom is replaced by Fe2+ atom as the
nescence experiments were taken on a Horiba-Jobin Yvon Fluorolog 3.
parameters for the Ru atom are not available in the program. It is a
[Ru(bpy)2dppz]2+ was excited at 440 nm, and right angle emission was
valid approximation, because the [Ru(bpy)2dppz]2+ complex interacts
obtained from 550 to 700 nm with 2 nm slit widths. Both emission and
with fibrils only through its aromatic ligands. Moreover, the Fe2+ and
excitation were corrected for instrument dependent effects. Intensity at
Ru2+ atoms are similar as they are in the same group of the periodic
640 nm was used for subsequent calculations. At high concentrations
table. To treat the complex as a whole by the Autodock Vina program,
of metal complex, corrections for inner filter effects were required.
the coordination between the metal and ligands is defined as a single
This was performed following the methods of Kubista et al.36 The
bond by modifying the structure in the YASARA software.40 The size
photoluminescent intensity was corrected by the inner filter effect
of the grid was chosen to occupy the whole ligand−peptide complex,
using:
and the spacing was kept to 1.00 Å that is a standard value for
corr
Iem obs −A(λex )l p
= Iem 10 (1) Autodock Vina. Each docking trial produced 20 poses with the
exhaustiveness value of 20. In the rigid docking, the flexibility of the
where Iemcorr
is the corrected emission, Iemobs
is the emission obtained receptor (the Aβ fibril), was elusive. To investigate the effect of the Aβ
from the spectrometer, A(λex) is the absorbance at the excitation fibril flexibility on the docked structures, the following two methods

10811 dx.doi.org/10.1021/ja404850u | J. Am. Chem. Soc. 2013, 135, 10810−10816

Journal of the American Chemical Society Article

method.45 A constant pressure of 1 bar was applied with a coupling

constant of 1.0 ps; peptide, water molecules, and ions were coupled
separately to a bath at 300 K with a coupling constant of 0.1 ps. The
equation of motion was integrated at each 2 fs time steps. The tools
available in the GROMACS program package and the YASARA
program46 have been used for analyzing trajectories and simulated
structures.
The MD simulations of [Ru(bpy)2dppz]2+ bound to the Aβ peptide
were performed using AMBER 03 force field47 as implemented in the
YASARA program40,46 in explicit aqueous solution. The box was filled
with single point charge (SPC) water molecules. The sodium and
chloride ions were also added to simulate the ion concentration of 150
mM under physiological conditions. The docked poses provided in the
previous step were used as the starting structures and placed into a
cubic box with dimensions of 101 × 84 × 78 Å. The remaining
parameters used in the simulations have been described above.
Analysis of the trajectories and simulated structures were performed
with the in-built tools of YASARA program. During these MD
simulations C′, Cα atoms of terminal residues and atoms forming
peptide bonds between residues Gln15-Lys16, Val24-Gly25, Lys28-
Gly29, and Leu34-Met35 were fixed to maintain the secondary
structures of Aβ1−40 fibrils. These residues were selected due to their
positions in the fibrils. The Val24-Gly25 residues are located at the end
of the first β sheet, Lys28-Gly29 at the beginning of the second β
sheet, and Gln15-Lys16 and Leu34-Met35 constitute two mid β sheet
residue pairs. These atomic constraints help to maintain the secondary
structure of fibrils without affecting the flexibility of the binding sites.

■ RESULTS AND DISCUSSION

As a first step, we studied the photoluminescence of
[Ru(bpy)2dppz]2+ as a function of the concentration of
fibrillized Aβ1−40 to gain information about the [Ru-
Figure 1. Binding of [Ru(bpy)2dppz]2+ to fibrillar Aβ. (a) Change in
(bpy)2dppz]2+ binding site and to assess the stability of the
photoluminescence as a function [Ru(bpy)2dppz]2+ concentration. Aβ-[Ru(bpy)2dppz]2+ complex. Figure 1a shows a standard
The concentration of Aβ was kept constant at 4 μM. The dissociation saturation curve, which can be fitted to a single-binding site
constant reported in the text is calculated from the fitting of these data with an equilibrium dissociation constant (Kd) of 2.1 ± 0.5 μM
(red line) and is the average of five independent experiments. (b) Job (binding constant, Kb, of 4.8 × 105 M−1). This value is smaller
plot analysis of the photoluminescence of [Ru(bpy)2dppz]2+ in the than the binding constant reported for [Ru(bpy)2dppz]2+
presence of fibrillar Aβ. The mole fraction of the [Ru(bpy)2dppz]2+ bound to DNA (>106 M−1).48 The smaller binding constant
and Aβ was varied, while the total concentration of the two of [Ru(bpy)2dppz]2+ with fibrillar Aβ implies a weaker
components was kept constant at 100 μM. Δ Photoluminescence in interaction in comparison with DNA, which is consistent
the y axis represents the subtraction of the background [Ru-
with the smaller increase in [Ru(bpy)2dppz]2+ photolumines-
(bpy)2dppz]2+ photoluminescence in buffer alone from the [Ru-
(bpy)2dppz]2+ photoluminescence in the presence of fibrillar Aβ. The cence with Aβ (619-fold) than with DNA (1127-fold, see
blue curve is a polynomial fit to better illustrate the concave nature of Supporting Information Figure S2). The Barton group recently
the graph and its maximum. elucidated the X-ray structure of [Ru(bpy)2dppz]2+ bound to
DNA and demonstrated that the dppz ligand is capable of
were utilized: (1) flexible docking and (2) rigid docking on the intercalating between two well-matched stacked base pairs.49
different structures of the Aβ fibrils derived from a short-term (5 ns) This intercalation provides a hydrophobic cavity that
MD simulations in an aqueous solution. The molecular dynamics conveniently shields the dppz ligand from water. Binding
(MD) simulation used in the latter was performed using the sites produced by base stacking are specifically present in
GROMACS program 41,42 utilizing the GROMOS force field oligonucleotides but absent in supramolecular assemblies such
GROMOS96 53A6.42 For all simulations, the starting structures
as Aβ fibrils, which could explain the weaker binding interaction
were placed in a truncated cubic box with dimensions of 100 × 100 ×
100 Å. This dismisses unwanted effects that may arise from the applied of [Ru(bpy)2dppz]2+ to Aβ aggregates. Nonetheless, the
periodic boundary conditions. The box was filled with single point dissociation constant of [Ru(bpy)2dppz]2+ to fibrillar Aβ
charge (SPC) water molecules. Some water molecules were replaced compares with the dissociation constants of ThT and CR
by sodium and chloride ions to neutralize the system and to simulate with Aβ fibrils (0.823 and 1.1 μM, respectively50).
an experimentally used ion concentration of 150 mM. The starting To further characterize the [Ru(bpy)2dppz]2+ binding site,
structures were subsequently energy-minimized with a steepest we used the continuous variation method38 to interrogate the
descent method for 3000 steps. The results of these minimizations system about the number of Aβ monomers associated with the
produced the starting structures for the MD simulations. The MD binding of [Ru(bpy)2dppz]2+. The curve generated, also known
simulations were then carried out with a constant number of particles
(N), pressure (P), and temperature (T), that is, NPT ensemble. The
as a Job plot, can be seen in Figure 1b and shows that the
SETTLE algorithm43 was used to constrain the bond length and angle maximum of the [Ru(bpy)2dppz]2+ molar fraction is between
of the water molecules, while the LINCS algorithm44 was used to the 0.25 and the 0.30 values. Therefore, we have decided to use
constrain the bond length of the peptide. The long-range electrostatic these values as the uncertainty range and taken the 0.275
interactions were calculated by the Particle-Mesh Ewald (PME) middle value as the curve maximum. This allows determining a
10812 dx.doi.org/10.1021/ja404850u | J. Am. Chem. Soc. 2013, 135, 10810−10816
Journal of the American Chemical Society Article

binding stoichiometry of ca. 2.6 ± 0.4 Aβ monomers per every

[Ru(bpy)2dppz]2+ bound to the fibril. Job plots generated using
total concentrations (Aβ+[Ru(bpy)2dppz]2+) of 100, 50, and
20 μM are indistinguishable from one another (see Supporting
Information, Figure S3). Previous studies for ThT have
identified 6.3 Aβ monomers per bound ThT,23 and 1.7 per
bound CR.50 The binding stoichiometry of [Ru(bpy)2dppz]2+
and ThT tends to indicate that multiple Aβ monomers need to
come together to form a binding site. This is consistent with
the strong photoluminescence of [Ru(bpy)2dppz]2+ in the
presence of Aβ fibrils but not when in contact with monomers.
The smaller binding stoichiometry for CR could be attributed
to the association of multiple CR molecules to the same
binding site,10 or to the availability of multiple binding sites.11 It
is important to point out that the maximum of the curve in
Figure 1b is not sharp but rather broad, which suggests that
there could be some slight variation in the binding
stoichiometry. This is rather expected because, contrary to
enzymes, which have well-defined active sites that match the
shape of the target molecule, the binding site of Aβ fibrils will
likely allow the [Ru(bpy)2dppz]2+ to assume slightly different
conformations, resulting in a small continuum of binding Figure 2. Binding modes of [Ru(bpy)2dppz]2+ on Aβ fibrils. (a)
stoichiometries. Binding sites in fibril edges (sites A and B) observed perpendicular
The identification of the exact binding sites between (left) and through (right) to the fibril axis. (b) Binding sites along the
[Ru(bpy)2dppz]2+ and Aβ1−40 fibril would require a high- fibril axis (sites C and D) observed perpendicular (left) and through
(right) the fibril axis. Circular arrow indicates the position of the 2-fold
resolution structure of the Aβ-[Ru(bpy)2dppz]2+ complex, rotational axis.
which is challenging to obtain by either NMR or X-ray
diffraction. To overcome this, we have utilized three different
molecular docking techniques and MD simulations. In the past
few years, computational approaches have emerged as a the Job plot in Figure 1b suggests also a single binding
powerful tool for studying protein−ligand interactions. Here, stoichiometry with ca. 2.6 Aβ monomers per [Ru-
the binding of [Ru(bpy)2dppz]2+ to Aβ1−40 fibril models with (bpy)2dppz]2+. To account for these observations, we propose
either two-fold5 or three-fold7 symmetry was investigated using that site C is mainly responsible for the binding with only
(1) rigid docking, (2) flexible docking, and (3) rigid docking on marginal binding to site D. The two-fold symmetry
different conformations of fibrils derived from short-term MD representation used to model the interactions of [Ru-
simulations. The last two approaches include flexibility of the (bpy)2dppz]2+ with Aβ fibrils only contains amino acids from
receptor, and the details of all of these procedures are provided positions 9−40, which are the ones that form the fibril
in the Supporting Information. The results were similar using backbone. Amino acids 1−8 are in a random coil conformation,
the three methods and are summarized in Figure 2 for fibrils occupying most of the region where site D is found. This
with two-fold symmetry,5 which are likely the most abundant should block the access of [Ru(bpy)2dppz]2+ to site D and
Aβ polymorph in the sample.51 For two-fold symmetry Aβ dramatically reduce its binding. In contrast, site C is right at the
fibril, four major sites were found, with occupation percentages surface of the fibril and accessible for [Ru(bpy)2dppz]2+
of A = 46.9%, B = 16.2%, C = 15.6%, and D = 13.8%. Although binding. Furthermore, molecular modeling shows that even
four different binding sites were located, it is noteworthy that without this random coil region, the binding energy of site C
sites A and B are located at the ends of the fibril (Figure 2a). (−9.1 kcal/mol) is more negative than that of site D (−8.7
Binding sites at the end of the fibrils are scarce given the fibrillar kcal/mol), and the percentage of [Ru(bpy)2dppz]2+ bound to
elongation displayed by Aβ, and therefore their contribution to site C (15.6%) is larger than that to site D (13.8%).52
the binding of [Ru(bpy)2dppz]2+ is negligible. For example, a We have also investigated the stability of site C using 10 ns
fibril several micrometers long with thousands Aβ monomers all-atom classical molecular dynamics (MD) simulations in
will have a maximum of 4 A sites and 4 B sites. In contrast to explicit aqueous solution. These simulations were performed
this, sites C and D run longitudinally to the fibril axis, and their using AMBER force field as implemented in the YASARA
number would increase as more monomers are added and the program.47 The [Ru(bpy)2dppz]2+ complex on the Aβ1−40
fibril is elongated. Therefore, it is plausible that sites C and D fibrils was found to remain intact throughout the simulations.
(Figure 2b) are mostly responsible for the binding of However, a slight change in the orientation of the complex was
[Ru(bpy)2dppz]2+ to Aβ1−40 fibrils. A more detailed discussion observed. The Ru(bpy)2 part of the complex raises slightly from
about the binding sites of [Ru(bpy)2dppz]2+ to Aβ1−40 fibrils the fibril axis in comparison with the previous buried position
can be found in the Supporting Information. obtained by molecular docking (Figure 3, and Supporting
A reliable model for the binding of [Ru(bpy)2dppz]2+ to Aβ Information Figure S6). In the new position, due to a better
should be consistent with both biophysical and computational alignment of two aromatic rings of Phe20 of the fibrils and the
data; however, a few inconsistencies can be found when both dppz ring of the metal complex, the π−π interactions between
sets of data are compared. Especially, the data in Figure 1a fit them become stronger, while the number of CH−π interactions
well to a single site model in contrast with the two sites (sites C between the side chain of Val18 and dppz ring of the complex
and D) found in the molecular docking experiments. Second, decreases (Supporting Information Figure S7).
10813 dx.doi.org/10.1021/ja404850u | J. Am. Chem. Soc. 2013, 135, 10810−10816
Journal of the American Chemical Society Article

fold symmetry.7 Although this kind of fibril is expected to be in

a minimum amount (if any) in our preparations, it is a good
example of a different kind of Aβ1−40 fibril polymorph.
Interestingly, the results tend to indicate that the preferred
binding site would be site K (Supporting Information Figure
S5), which is again the hydrophobic cleft formed between
Val18 and Phe20. The stability of site K was also probed by 10
ns all-atom classical molecular dynamics (MD) simulations in
explicit aqueous solution. These simulations confirmed that the
complex has a preference for this Val18-Phe20 hydrophobic
cleft; although similar to the fibrils with 2-fold symmetry,
changes in the orientation of the complex were observed.
Figure 3. Binding of [Ru(bpy)2dppz]2+ to Aβ by MD simulation. The Nonetheless, it would be reasonable to think that this
binding site is formed between Val18 and Phe20 (previously identified hydrophobic cleft, formed by the self-assembly of several
as site C). The red-marked peptides represent the projection of the monomers to form a fibril structure, is a general biding site for
[Ru(bpy)2dppz]2+ complex on the Aβ axis. This represents ca. 2 Aβ [Ru(bpy)2dppz]2+, which will likely be found in different
monomers per [Ru(bpy)2dppz]2+. polymorphic conformations of Aβ fibrils.
It is important to mention that the computational
The assignment of site C as the dominating binding site of simulations presented were performed using Δ-[Ru-
[Ru(bpy)2dppz]2+ to Aβ is consistent with the biophysical (bpy)2dppz]2+; however, we also investigated the binding of
experiments. The value of 2.6 Aβ monomers per [Ru- the Λ-[Ru(bpy)2dppz]2+ enantiomer. Interestingly, upon
(bpy)2dppz]2+ determined using the Job plot in Figure 1b is docking the isomer to sites C and D, no significant changes
in agreement with the 2.0 monomers per ruthenium complex in the binding poses and binding frequencies were observed,
obtained from MD simulations (Figure 3), as well as the single which contrasts with its behavior in DNA that shows significant
binding site obtained from Figure 1a. The slightly smaller value variations in photoluminescent intensity and lifetime depending
calculated from MD simulation is expected because it considers on the bound enantiomer.31,54 We are currently investigating
the minimum space that the complex could possibly occupy. In the binding of the pure Δ- and Λ-[Ru(bpy) 2 dppz] 2+
reality, the distance between adjacent complexes is expected to enantiomers to Aβ using biophysical techniques, and the
be larger to minimize electrostatic repulsion and steric results will be reported in a future publication.
constrains. Site C is a hydrophobic cleft formed between So far, our experimental results have shown that [Ru-
Val18 and Phe20. At this site, the side chains of Val18 and (bpy)2dppz]2+ binds to a single binding site and that this site is
Phe20 interact with the aromatic ring of dppz through CH−π formed by ca. 2.6 monomers. Our computational simulations
and π−π interactions, respectively (Supporting Information have identified a hydrophobic cleft between Val 18 and Phe 20
Figure S6). The dppz ligand is a hydrophobic extended as the binding site for [Ru(bpy)2dppz]2+. This remarkable
aromatic system, which can efficiently hide from water by agreement between the experimental and computational studies
binding to hydrophobic domains. From Figure 3, it can be seen has led us to propose that this site formed by Val 18 and Phe 20
that the hydrophobic part of [Ru(bpy)2dppz]2+ (the dppz and that runs throughout the fibril axis is responsible for
ligand) is buried into the hydrophobic cleft of site C, while the “turning on the photoluminescence switch” of [Ru-
ionic part (Ru(bpy)22+ fragment) is projected outward, allowing (bpy)2dppz]2+. To further demonstrate this assignment, we
a better solvation by water molecules. Interestingly, this have studied the binding of [Ru(bpy)2dppz]2+ to Aβ25−35, an
observation is consistent with the light-switching properties amyloid peptide fragment that lacks the Val18-Phe20 binding
of [Ru(bpy)2dppz]2+. When [Ru(bpy)2dppz]2+ is in water, its site but that forms amyloid fibrils and binds other Aβ binding
excitation is rapidly transferred to a low-lying dark state, which dyes such as ThT (see Supporting Information Figure S8).55,56
does not display any photoluminescence.53 However, when Figure 4 shows the photoluminescence of [Ru(bpy)2dppz]2+ in
[Ru(bpy)2dppz]2+ is dissolved in organic solvents or interca- the presence of fibrillar Aβ25−35 and Aβ1−40. When [Ru-
lated within the bases of DNA, the microenvironment around (bpy)2dppz]2+ is in contact with Aβ1−40 fibrils, its photo-
the dppz ligand changes, destabilizing the dark state and luminescence increases by more than 65-fold, while when in
promoting the population of an emissive state, which is contact with Aβ25−35 fibrils, the increase in photoluminescence
responsible for the photoluminescence emission. The light- is small (7-fold). The low photoluminescence response
switching behavior of [Ru(bpy)2dppz]2+ in the presence of Aβ indicates a poor interaction between [Ru(bpy)2dppz]2+ and
is consistent with the binding of [Ru(bpy)2dppz]2+ to site C, Aβ25−35, which demonstrates that the hydrophobic cleft formed
which would provide the hydrophobic microenvironment by Val18 and Phe20 is of great importance for the binding and
photoluminescence response of [Ru(bpy)2dppz]2+.

■
around the dppz ligand necessary to promote the population
of the emissive state, and the concomitant increase in
photoluminescence. CONCLUSIONS
The binding site determined for [Ru(bpy)2dppz]2+ is in In summary, our binding and computational experiments have
agreement with those calculated for ThT and Congo Red. led us to conclude that [Ru(bpy)2dppz]2+ binds to a
Molecular modeling has identified that both ThT and Congo hydrophobic cleft, formed on the surface of Aβ fibrils between
red bind to amyloid forming peptides, with the long molecular Val18 and Phe20 which is responsible for the “light-switching”
axis oriented parallel to the fibril axis in agreement with our properties of this ruthenium dipyridophenazine complex. The
docking studies. In fact, site C is similar to the binding site great consistency between the biophysical studies and the
identified by Wu et al. for CR.11 Furthermore, we also analyzed computational simulations found in this study validates the
the binding of [Ru(bpy)2dppz]2+ to Aβ1−40 fibrils with three- proposed model and offers a guide for identifying the binding
10814 dx.doi.org/10.1021/ja404850u | J. Am. Chem. Soc. 2013, 135, 10810−10816
Journal of the American Chemical Society Article

symmetry Aβ fibril models and Kathleen Matthews for helpful

discussions. This material is based upon work supported by the
grants of the Welch Foundation (grant no. C-1743) to A.A.M.,
the National Science Foundation (grant no. 1152846), and the
James and Esther King Biomedical Research Program of the
Florida State Health Department (grant no. 08KN-11) to R.P.

■ REFERENCES
(1) Shoji, M.; Golde, T. E.; Ghiso, J.; Cheung, T. T.; Estus, S.;
Shaffer, L. M.; Cai, X.-D.; McKay, D. M.; Tintner, R.; Frangione, B.;
Younkin, S. G. Science 1992, 258, 126−129.
(2) Hardy, J. A.; Higgins, G. A. Science 1992, 256, 184−185.
(3) Cohen, S. I. A.; Linse, S.; Luheshi, L. M.; Hellstrand, E.; White,
D. A.; Rajah, L.; Otzen, D. E.; Vendruscolo, M.; Dobson, C. M.;
Knowles, T. P. J. Proc. Natl. Acad. Sci. U.S.A. 2013, 110, 9758−9763.
(4) Cook, N. P.; Torres, V.; Jain, D.; Martí, A. A. J. Am. Chem. Soc.
2011, 133, 11121−11123.
Figure 4. Photoluminescence of [Ru(bpy)2dppz]2+ with fibrillar Aβ. (5) Petkova, A. T.; Yau, W.-M.; Tycko, R. Biochemistry 2006, 45,
Photoluminescence spectra of [Ru(bpy)2dppz]2+ incubated with 498−512.
fibrillar Aβ1−40 (red line), fibrillar Aβ25−35 (blue line), and buffer (6) Lührs, T.; Ritter, C.; Adrian, M.; Riek-Loher, D.; Bohrmann, B.;
(black line). Inset shows a TEM image of Aβ25−35 fibrils. Scale bar 200 Döbeli, H.; Schubert, D.; Riek, R. Proc. Natl. Acad. Sci. U.S.A. 2005,
nm. 102, 17342−17347.
(7) Paravastu, A. K.; Leapman, R. D.; Yau, W.-M.; Tycko, R. Proc.
sites of other amyloid binding molecules. Although other Natl. Acad. Sci. U.S.A. 2008, 105, 18349−18354.
binding sites could exist, those do not cause an increase in the (8) Rodríguez-Rodríguez, C.; Rimola, A.; Rodríguez-Santiago, L.;
photoluminescence of [Ru(bpy)2dppz]2+ upon binding, or are Ugliengo, P.; Á lvarez-Larena, Á .; Gutierrez-de-Terán, H.; Sodupe, M.;
Gonzalez-Duarte, P. Chem. Commun. 2010, 46, 1156−1158.
too scarce to produce any measurable effect (such as terminal
(9) Wu, C.; Wang, Z.; Lei, H.; Duan, Y.; Bowers, M. T.; Shea, J.-E. J.
sites A and B). It is important to point out that [Ru- Mol. Biol. 2008, 384, 718−729.
(bpy)2dppz]2+ binds DNA with the dppz ligand perpendicular (10) Wu, C.; Wang, Z.; Lei, H.; Zhang, W.; Duan, Y. J. Am. Chem.
to the DNA strand axis, while the binding to Aβ is parallel to Soc. 2007, 129, 1225−1232.
the fibril axis. This parallel binding mode is unprecedented for (11) Wu, C.; Scott, J.; Shea, J.-E. Biophys. J. 2012, 103, 550−557.
[Ru(bpy)2dppz]2+ and opens a new window of possibilities for (12) Bruce, N. J.; Chen, D.; Dastidar, S. G.; Marks, G. E.; Schein, C.
the binding of this complex to other molecular architectures H.; Bryce, R. A. Peptides 2010, 31, 2100−2108.
containing long hydrophobic domains on their surface. In (13) Lemkul, J. A.; Bevan, D. R. Biochemistry 2010, 49, 3935−3946.
addition, this research provides hard evidence of the ability of a (14) Raman, E. P.; Takeda, T.; Klimov, D. K. Biophys. J. 2009, 97,
hydrophobic cleft in the surface of Aβ fibrils to bind 2070−2079.
hydrophobic molecules with extended aromatic systems. (15) Liu, F.-F.; Dong, X.-Y.; He, L.; Middelberg, A. P. J.; Sun, Y. J.
Therefore, it is reasonable to think that molecules such as Phys. Chem. B 2011, 115, 11879−11887.
(16) Chen, D.; Martin, Z. S.; Soto, C.; Schein, C. H. Biorg. Med.
coumarins and rhodamines, which have extended aromatic
Chem. 2009, 17, 5189−5197.
systems, can potentially bind Aβ in a similar way. This would (17) Buchete, N.-V.; Tycko, R.; Hummer, G. J. Mol. Biol. 2005, 353,
allow the design of a new generation of Aβ binding molecules 804−821.
with potential applications in sensing and inhibition of Aβ (18) Lemkul, J. A.; Bevan, D. R. ACS Chem. Neurosci. 2012, 3, 845−
aggregation.

■
856.
(19) Wu, C.; Bowers, M. T.; Shea, J.-E. Biophys. J. 2011, 100, 1316−
ASSOCIATED CONTENT 1324.
*
S Supporting Information (20) Echeverria, V.; Zeitlin, R.; Burgess, S.; Patel, S.; Barman, A.;
Determination of the photoluminescence correction factor, Thakur, G.; Inouye, H.; Mamcarz, M.; Wang, L.; Buckingham, S. D.;
photoluminescence spectra of [Ru(bpy)2dppz]2+ in the Kirschner, D. A.; Mori, T.; Leblanc, R. M.; Prabhakar, R.; Sattelle, D.;
presence of DNA and fibrillar Aβ, Job Plot with different Aβ Arendash, G. W. J. Alzheimer’s Dis. 2011, 24, 817−835.
(21) Landau, M.; Sawaya, M. R.; Faull, K. F.; Laganowsky, A.; Jiang,
concentrations, expanded discussion of the computational
L.; Sievers, S. A.; Liu, J.; Barrio, J. R.; Eisenberg, D. PLoS Biol. 2011, 9,
binding results, and photoluminescence of Aβ25−35 with ThT. e1001080.
This material is available free of charge via the Internet at (22) Groenning, M. J. Chem. Biol. 2010, 3, 1−18.
http://pubs.acs.org.

■
(23) Levine, H. Amyloid 2005, 12, 5−14.
(24) Erkkila, K. E.; Odom, D. T.; Barton, J. K. Chem. Rev. 1999, 99,
AUTHOR INFORMATION 2777−2796.
Corresponding Author (25) Jiménez-Hernández, M. E.; Orellana, G.; Montero, F.; Portolés,
rpr@miami.edu; amarti@rice.edu M. T. Photochem. Photobiol. 2000, 72, 28−34.
Notes (26) Jain, D.; Saha, A.; Martı ́, A. A. Chem. Commun. 2011, 47, 2246−
2248.
The authors declare no competing financial interest.

■
(27) Puckett, C. A.; Barton, J. K. Biochemistry 2008, 47, 11711−
11716.
ACKNOWLEDGMENTS (28) Puckett, C. A.; Barton, J. K. J. Am. Chem. Soc. 2006, 129, 46−47.
We would like to thank Robert Tycko for generously providing (29) Cook, N. P.; Kilpatrick, K.; Segatori, L.; Martí, A. A. J. Am.
the atomic coordinates for the two-fold and three-fold Chem. Soc. 2012, 134, 20776−20782.

10815 dx.doi.org/10.1021/ja404850u | J. Am. Chem. Soc. 2013, 135, 10810−10816

Journal of the American Chemical Society Article

(30) Murphy, C. J.; Nair, R. B.; Keller, C. E.; Teng, E. S.; Pollard, C.
Proc. SPIE 1997, 2980, 473−478.
(31) Svensson, F. R.; Abrahamsson, M.; Strömberg, N.; Ewing, A. G.;
Lincoln, P. J. Phys. Chem. Lett. 2011, 2, 397−401.
(32) Fezoui, Y.; Hartley, D. M.; Harper, J. D.; Khurana, R.; Walsh, D.
M.; Condron, M. M.; Selkoe, D. J.; Lansbury, P. T.; Fink, A. L.;
Teplow, D. B. Amyloid 2000, 7, 166−178.
(33) Amouyal, E.; Homsi, A.; Chambron, J.-C.; Sauvage, J.-P. Dalton
Trans. 1990, 1841−1845.
(34) Dickeson, J.; Summers, L. Aust. J. Chem. 1970, 23, 1023−1027.
(35) Sun, Y.; Joyce, L. E.; Dickson, N. M.; Turro, C. Chem. Commun.
2010, 46, 2426−2428.
(36) Kubista, M.; Sjöback, R.; Eriksson, S.; Albinsson, B. Analyst
1994, 119, 417−419.
(37) Celej, M. S.; Jares-Erijman, E. A.; Jovin, T. M. Biophys. J. 2008,
94, 4867−4879.
(38) Huang, C. Y. Methods Enzymol. 1982, 87, 509−525.
(39) Trott, O.; Olson, A. J. J. Comput. Chem. 2010, 31, 455−461.
(40) Krieger, E.; Koraimann, G.; Vriend, G. Proteins: Struct., Funct.,
Genet. 2002, 47, 393−402.
(41) Lindahl, E.; Hess, B.; van der Spoel, D. J. Mol. Model. 2001, 7,
306−317.
(42) Oostenbrink, C.; Villa, A.; Mark, A. E.; van Gunsteren, W. F. J.
Comput. Chem. 2004, 25, 1656−1676.
(43) Miyamoto, S.; Kollman, P. A. J. Comput. Chem. 1992, 13, 952−
462.
(44) Hess, B.; Bekker, H.; Berendsen, H. J. C.; Fraaije, J. G. E. M. J.
Comput. Chem. 1997, 18, 1463−1472.
(45) Darden, T. A.; York, D.; Pedersen, L. J. Chem. Phys. 1993, 98,
10089−10092.
(46) Krieger, E.; Vriend, G. Bioinformatics 2002, 18, 315−318.
(47) Duan, Y.; Wu, C.; Chowdhury, S.; Lee, M. C.; Xiong, G.; Zhang,
W.; Yang, R.; Cieplak, P.; Luo, R.; Lee, T.; Caldwell, J.; Wang, J.;
Kollman, P. J. Comput. Chem. 2003, 24, 1999−2012.
(48) Friedman, A. E.; Chambron, J.-C.; Sauvage, J.-P.; Turro, N. J.;
Barton, J. K. J. Am. Chem. Soc. 1990, 112, 4960−4962.
(49) Song, H.; Kaiser, J. T.; Barton, J. K. Nat. Chem. 2012, 4, 615−
620.
(50) Zhen, W.; Han, H.; Anguiano, M.; Lemere, C. A.; Cho, C.-G.;
Lansbury, P. T. J. Med. Chem. 1999, 42, 2805−2815.
(51) The morphology of our fibrils examined by TEM is in
agreement with those identified by the Tycko group as having two-fold
symmetry: Paravastu, A. K.; Leapman, R. D.; Yau, W.-M.; Tycko, R.
Proc. Natl. Acad. Sci. U.S.A. 2008, 105, 18349−18354 Furthermore, it
was also demostrated by Tycko’s group that static incubation tends to
result in three-fold symmetry fibrils, while incubation under agitation
forms two-fold symmetry fibrils. Our fibrils are grown under agitation.
(52) It is important to recognize that the remaining percentage is
attributed to sites A and B, but only in this short fibril are fragments
made of 6 Aβ monomers. Percentage contribution in a real
micrometer long fibril for sites A and B is negligible.
(53) Brennaman, M. K.; Alstrum-Acevedo, J. H.; Fleming, C. N.;
Jang, P.; Meyer, T. J.; Papanikolas, J. M. J. Am. Chem. Soc. 2002, 124,
15094−15098.
(54) Andersson, J.; Fornander, L. H.; Abrahamsson, M.; Tuite, E.;
Nordell, P.; Lincoln, P. Inorg. Chem. 2013, 52, 1151−1159.
(55) Kellermayer, M. S. Z.; Karsai, Á .; Benke, M.; Soós, K.; Penke, B.
Proc. Natl. Acad. Sci. U.S.A. 2008, 105, 141−144.
(56) Grace, E. A.; Rabiner, C. A.; Busciglio, J. Neuroscience 2002, 114,
265−273.

10816 dx.doi.org/10.1021/ja404850u | J. Am. Chem. Soc. 2013, 135, 10810−10816