Computational study on donor-acceptor optical markers for Alzheimer's disease: a game of charge transfer and electron delocalization.

View Article Online View Journal

PCCP Accepted Manuscript

This article can be cited before page numbers have been issued, to do this please use: M. Sodupe, F. Peccati, X. Solans and M. D. Wisniewska, Phys. Chem. Chem. Phys., 2016, DOI: 10.1039/C5CP07274C.

This is an Accepted Manuscript, which has been through the Royal Society of Chemistry peer review process and has been accepted for publication. Accepted Manuscripts are published online shortly after acceptance, before technical editing, formatting and proof reading. Using this free service, authors can make their results available to the community, in citable form, before we publish the edited article. We will replace this Accepted Manuscript with the edited and formatted Advance Article as soon as it is available. You can find more information about Accepted Manuscripts in the Information for Authors. Please note that technical editing may introduce minor changes to the text and/or graphics, which may alter content. The journal’s standard Terms & Conditions and the Ethical guidelines still apply. In no event shall the Royal Society of Chemistry be held responsible for any errors or omissions in this Accepted Manuscript or any consequences arising from the use of any information it contains.

www.rsc.org/pccp

Page 1 of 33

Physical Chemistry Chemical Physics

Computational study on donor‐acceptor optical

View Article Online

markers for Alzheimer’s disease: a game of charge transfer and electron delocalization. Francesca Peccati,a Marta Wiśniewska,b Xavier Solans‐Monfort,a,* Mariona Sodupea,*

Published on 20 January 2016. Downloaded on 25/01/2016 14:55:59.

a Departament de Química, Universitat Autònoma de Barcelona, Edifici Cn, 08193

– Bellaterra Spain. b Centre of New Technologies, University of Warsaw, Banacha 2c Street, 02‐097

Warsaw, Poland. [email protected] [email protected]

1

Physical Chemistry Chemical Physics Accepted Manuscript

DOI: 10.1039/C5CP07274C


Abstract

Page 2 of 33

View Article Online

DOI: 10.1039/C5CP07274C

central event in the Alzheimer’s disease and thus, detection of these deposits is crucial to monitor the progression of the pathology. Despite its low tissue penetration, fluorescence imaging may become an alternative technique for identifying these deposits because it is less toxic and less costly than positron


emission tomography. Suitable dyes, however, should emit in the Near Infrared region (NIR), cross the blood‐brain barrier and target A aggregates. In this work, we use TD‐DFT, AIMD simulations and protein energy landscape exploration (PELE) to analyze the photophysical properties of a family of donor‐ acceptor markers and their binding to amyloid fibrils. These markers are formed by a N,N‐dimethylaniline donor and a propanedinitrile acceptor groups separated by a spacer consisting of one, two or three conjugated double bonds. The smallest compound has a low emission wavelength, can deactivate through a non‐radiative process involving a conical intersection and binds weakly to A fibrils. In contrast, the largest dye is a suitable compound as it shows an emission wavelength in the NIR, does not seem to relax through conical intersection processes and binds to A fibrils strongly entering hydrophobic voids. Analysis of electronic excitations shows that the transition has an important charge transfer character that increases with the length of the spacer, the  bridge being an active participant in the transition. Therefore, adding double bonds to the dye skeleton has two beneficial effects: i) it increases the emission wavelength as it enlarges the  system and ii) it increases the charge transfer character of the transition, which increases the red‐shifting of the emission wavelength in polar solvents.

2


According to the amyloid cascade hypothesis, amyloid‐ (A) deposition is a


Introduction

View Article Online

DOI: 10.1039/C5CP07274C

Alzheimer’s disease (AD) is the most common form of dementia that gradually worsens over time. It is associated with language impairment, irreversible memory loss, and disorientation. It presents several pathological hallmarks such as aggregation of the amyloid‐ (A) peptide into fibrils and plaques; tangles associated with the irregular phosphorylation of the tau protein and the


generation of reactive oxygen species (ROS) that increases the oxidative stress.1– 3 An early diagnosis of the disease is crucial for monitoring the progression of

neuropathological disorders as well as treating the symptoms of patients from the early stages. The amyloid cascade hypothesis states that amyloid‐ deposition in the brain is a fundamental cause of Alzheimer’s disease.4–7 Therefore, determination of the amounts of A deposits can serve as a good diagnostic for the AD progression and patient response to therapy.8 It is for this reason that several in vivo imaging techniques to identify the formation of A fibrils have been developed.9–17 The standard technique today is the Positron Emission Tomography (PET),9–11,18 which requires radiolabeled imaging probes, high cost equipment and involves a time‐consuming data acquisition process. On the other hand, optical imaging is associated with lower costs, faster acquisition processes and avoids the patient’s exposure to radioactivity.17 Unfortunately, within other limitations, optical imaging suffers from low tissue penetration that prevents its application in vivo. This can be at least partially overcome by using light from the Near Infrared (NIR) region (600÷800 nm), which would decreases the biological photodamage, the autofluorescence from biological matter or by using multi‐photon

3


Page 3 of 33


Page 4 of 33

View Article Online microscopy‐based imaging techniques. Moreover, these dyes have to cross the DOI: 10.1039/C5CP07274C

Scheme 1

Several fluorescent probes able to target A deposits have been synthetized (Scheme 1). Thioflavin T (ThT)19 is a widely used probe in post mortem or in vitro assays. However, its emission wavelength, which is far from the NIR region (approximately 445 nm), and the limited penetration through the BBB due to its positive charge prevent its use in vivo. With the aim of overcoming the limitations of ThT, other probes have been proposed such as: i) the boron dipyrromethane derivatives known as BODIPYs, BAP‐1 being one example;20,21 ii) the curcumin derivatives among which some of the most promising compounds are from the CRANAD series;22–24 or iii) compounds with a dithiophene central skeleton, among which the {[5’‐(p‐hydroxyphenyl)‐2,2’‐ bithienyl‐5‐yl]‐methylidene}‐propanedinitrile commonly known as NIAD‐4 is the most representative example.25–27 In 2014, M. Cui et al.28 reported a series of probes with a donor‐acceptor architecture and small molecular weight that can favor the penetration through the BBB. These probes were called DANIRs (Scheme 2) and we will use this terminology hereafter. Noteworthy, while

4



Blood‐Brain Barrier and target specifically the amyloid‐ deposits.

Page 5 of 33


View Article Online compound DANIR‐2a presents no fluorescence emission and low affinity for A DOI: 10.1039/C5CP07274C

emission wavelength of 665 nm in PBS and a strong affinity for A aggregates.28


Scheme 2

In the present contribution, we focus on the photophysical properties of DANIR compounds shown in Scheme 2. We use time‐dependent density functional theory (TD‐DFT) calculations to determine the absorption and emission transitions. TD‐DFT calculations are nowadays widely used and the potentialities and drawbacks are well established.29–32 In particular, we have recently applied them to the study of the photophysical properties of some dyes for A deposits detection.33,34 The present study also explores the potential role of the trans to cis isomerization along C=C double bonds in both the ground and the first excited singlet state. It has been shown that this rotation is associated with a conical intersection involving the ground state, and thus represents a potential non‐ radiative relaxation pathway in several dyes with C=C double bonds such as retinal.35–38 Moreover, the search for binding sites of these probes on three different fibril structures, that have been proposed experimentally by means of NMR measurements and electron microscopy,39–41 has also been carried out to determine the influence of dye size in the affinity for the amyloid‐ fibrils. Only aggregates containing 1‐40 amyloid‐ peptides were considered. Finally, with

5


aggregates, compound 2c (Scheme 2) is a promising probe. It presents an


Page 6 of 33

View Article Online the aim of better understanding the nature of the electronic transition involved DOI: 10.1039/C5CP07274C

doubles bonds in the dye skeleton, we have computed the dipole moments of S0 and S1 states and the spatial extent of the charge transfer associated to the S0 to S1 transition by computing the DCT index,42 which has been applied to push‐pull systems and qualitatively evaluates the charge transfer extent in the electronic excitation.43 Moreover, we have determined the electron delocalization in both


states with Harmonic Oscillator Model of Aromaticity (HOMA) index.44,45 This index is a fast way for evaluating the aromatic character of cyclic compounds that has also been recently applied to acyclic species such as polyene chains; 45 results for this acyclic compounds show that HOMA converges to an asymptotic value and that is a fair measure of the ‐electron delocalization. Note that DANIR compounds have recently been studied computationally by Arul Murugan et al.,46 who focused their attention on analyzing the potential role of these probes in two‐photon microscopy‐based imaging and the relevance of charge transfer in this process, determined by computing the r index developed by Guido, Adamo, and co‐workers.47 In the present contribution we focus on the fluorescence emission and how solvent effects tune this process. Additionally, we analyze the potential role of non‐radiative deactivation pathways as well as markers binding sites to  fibril models. That is, the aim of this work is to understand the photophysical properties of these dyes in solution and their interaction with fibrils, with the ultimate goal of getting insights that may help in the development of new dyes with improved Aβ detection capabilities.

6


in the absorption and fluorescence emission, as well as the role of the number of


Computational Details

View Article Online

DOI: 10.1039/C5CP07274C

Three sets of calculations (static DFT and TD‐DFT calculations, ab initio molecular dynamics simulations, and protein energy landscape exploration) have been performed in order to understand the photophysics of DANIRs dyes and their interaction with A‐fibrils. Moreover, the nature of the excitation has been analyzed by evaluating the dipole moment of S0 and S1 states, the DCT index


that qualitatively evaluates the charge transfer associated with the excitation42 and the electron delocalization in the aromatic ring and the conjugated double bond chain for the ground and excited state optimal geometries by means of the HOMA index.44 Static DFT calculations. The ground and excited states geometries of DANIRs dyes as well as the absorption and emission wavelengths were computed with density functional theory (DFT) and time‐dependent density functional theory (TD‐DFT) calculations using the CAM‐B3LYP functional48 as implemented in Gaussian09 package.49 All atoms were represented with a 6‐31+G(d,p) basis set.50 Solvation effects were included using the SMD continuum model51 and the non‐equilibrium linear response (LRNE) solvation framework52 for the absorption and the linear response equilibrium solvation (LRE)53 for the excited state optimizations. Otherwise stated, optimizations were performed in solution and the nature of the stationary points was verified by vibrational analysis. Thermal corrections were evaluated at 298.15 K and 1 atm using the harmonic oscillator and rigid rotor models as implemented in the Gaussian package. Note that, this methodology is identical to the one used in our study of NIAD‐4, for which a previous calibration study showed that it was a cost‐effective method.33

7


Page 7 of 33


Page 8 of 33

View Article Online Ab initio molecular dynamics (AIMD) simulations. The effect of dye flexibility on DOI: 10.1039/C5CP07274C

molecular dynamics (AIMD) calculations within the NVT ensemble at T = 300 K with the CP2K package.54 For that, DANIRs dyes were included in a 30 (compounds 2b and 2c) or 20 (compound 2a) Å cubic box. AIMD simulations of 10 ps were performed at the PBE‐D3 level of theory55–57 using a time step of 0.5 fs. We employed the Goedecker‐Teter‐Hutter (GTH) pseudopotentials58,59 and


the associated DZVP basis sets for representing all atoms.60 UV‐Vis absorption spectra simulations were performed taking 40 equally distributed snapshots for the most representative isomer (see below). Solvent effects were included with the same SMD continuum model51 used in the static calculations on 40 equally distributed snapshots resulting from the AIMD simulation in gas phase. Although this approach does not allow taking into account the influence of the solvent in the dye flexibility, it includes the influence of solvent polarity on the absorption spectrum. This is expected to be a reasonable approach for solvents with low viscosity as those considered here. For the case of 2a and 2b longer simulations of 20 ps were performed. In these cases 80 instead of 40 snapshots were considered and the resulting spectra were compared with those obtained from the 10 ps simulation (Figure S1). All spectra are very similar and, in particular, the maximum wavelengths are in the same region for the 10 and 20 ps simulations and the bandwidths do not differ significantly. Protein Energy Landscape Exploration (PELE) simulations. The binding site search and refinement of DANIRs dyes on three experimentally proposed A fibrils was performed with the PELE program.61,62 This program combines a Monte Carlo stochastic approach and protein prediction algorithms for identifying the most

8


the absorption wavelength was taken into account by performing ab initio

Page 9 of 33


View Article Online favorable binding sites. It includes partial protein relaxation allowing a better DOI: 10.1039/C5CP07274C

superior to conventional rigid receptor docking algorithms.62 The search was based on 90 trajectories for each fibril model (see below) and only the most favorable structures are reported. Energies are computed with the OPLS force field63 and solvation effects are included with the generalized Born model.64 With the aim of getting more accurate interaction energies a model including a portion Published on 20 January 2016. Downloaded on 25/01/2016 14:55:59.

of the fibril around the dye was constructed (Figure S2 of the supporting information shows the selected models) and computed at the PBE‐D2 level of theory using the 6‐31+G(d,p) basis sets for representing all atoms and the Gaussian09 package. DCT index. The DCT index was developed by Le Bahers, Adamo, and Ciofini.42 It qualitatively evaluates the spatial extent of the charge transfer transition between two electronic states and it corresponds to the distance between the barycenters of the regions in which electron density increases or decreases as a result of the electronic transition. It is based on the difference of electron densities between the excited and ground states ∆ρ r that is used to define the regions in which the electron density increases (ρ r ) and decreases (ρ r ) as follows:

  r    ES  r    GS  r 

  r  if   r   0  r    0 if   r   0 

0 if   r   0  r      r  if   r   0 

9

(1)

(2)


fitting between the amyloid‐ fibril and the marker, which has been shown to be


Page 10 of 33

View Article Online Further details can be found in the original contributions of Le Bahers, Adamo, DOI: 10.1039/C5CP07274C

index is applied.42,43 The values reported here are obtained with a custom developed program. HOMA index. The HOMA index was defined by Krygowski in the seventies as:44

HOMA 1



n

 R  R  n i

ref

2

(3)


i1

In which and

are the i‐th C‐C bond length and the C‐C bond length

reference value respectively, is the number of C‐C bonds and is a normalization factor that guarantees that HOMA index vanishes for fully localized systems. Consequently, HOMA index is close to 1 for highly aromatic systems and tends to 0 when there is no electron delocalization. Here, we have used 1.388 Å as Rref value as proposed by Krygowski and the  value is set to 257.7 according to the same bibliography.65 Although HOMA index is usually used to evaluate the degree of aromaticity of cyclic compound, it has also been applied to acyclic systems, and shown to be a fair measure of the ‐electron delocalization.45

Results and Discussion Ground and excited state properties DANIR probes are composed of an N,N’‐dimethylaniline and a propanedinitrile group connected by a variable number of conjugated double bonds: one in the case of derivative 2a, two in the case of 2b and three in the case of 2c, resulting

10


Ciofini and co‐workers as well as some paradigmatic examples in which this

Page 11 of 33


View Article Online in a donor‐acceptor linear architecture (Scheme 2 and Figure 1). The DOI: donor‐ 10.1039/C5CP07274C

bonds, is aimed at red‐shifting of the absorption and emission wavelengths of the molecule, as well as favoring the interaction with Aβ aggregates.28 While compound 2a has only one conformer, 2b and 2c exhibit two and four conformers, respectively, arising from torsion around dihedral ϕ1, ϕ2 and ϕ3. These isomers are reported in Figure 1, along with the corresponding relative


Gibbs free energies with respect to the most stable isomer. In the two cases that have more than one isomer, the linear one is by far the most stable (more than 17.4 kJ mol‐1). Moreover, the Gibbs energy barriers associated with rotations around ϕ2 and ϕ3 dihedral angles (see Table 1) are relatively high (larger than 35 kJ mol‐1), as expected for these kind of conjugated systems and thus, only the linear conformation will be important for describing the absorption of the ground state. The linear isomer presents a flat structure with all dihedral angles being essentially 0 or 180o and bond lengths typical of substituted aromatic rings and conjugated double bonds (Table S1).

11


acceptor character of these probes, along with the presence of conjugated double


Page 12 of 33

View Article Online

Figure 1 Conformational isomers of DANIR molecules 2a, 2b and 2c with the corresponding values of ΔG (kJ mol‐1).

The minima of the S1 state (Figure S3) are also flat for 2b and 2c. However, for the case of 2a, the dihedral angle between the acceptor propanedinitrile group and the aromatic ring assumes a value of 8.2 degrees (Figure S3). Moreover, in these minima and regardless of the dye, four C‐C bonds of the aromatic ring are longer by 0.002  0.026 Å than those of the ground state, while the other two are shorter by 0.004 – 0.012 Å. The C‐C bond distances of the double bond conjugated chain also suffer significant variations, becoming much more similar in the case of the S1 minima (Table S1). This has an important effect on the HOMA index (see below).

12



DOI: 10.1039/C5CP07274C

Page 13 of 33



2a

2b

2c

ΔG‡ (TS ϕ1)

43.9 (42.3)

47.4 (41.8)

35.1 (34.9)

ΔG‡ (TS ϕ2)

‐

73.5 (58.3)

52.6 (49.2)

ΔG‡ (TS ϕ3)

‐

‐

59.1 (52.7)

λabs gas phase [f]a

336 [0.9]

375 [1.3]

406 [1.8]

λabs CH2Cl2 [f]a

376 [1.1]

431 [1.6]

473 [2.0]

λmax,abs CH2Cl2b

385

455

505

exp λmax,abs,CH2Cl2 c

433

489

519

λabs water [f]a

377 [1.1]

425 [1.6]

470 [2.0]

λmax,abs , waterb

385

435

505

λem gas phase [f]

376 [0.8]

411 [1.3]

450 [1.9]

λem CH2Cl2 [f]

433 [1.3]

525 [1.7]

619 [2.2]

λem water [f]

450 [1.3]

549 [1.8]

655 [2.3]

exp λem,PBC c

487

577

665

LogBB

‐0.96

‐1.04

‐1.12

a Absorption wavelength from optimized ground state structures

b Maximum absorption wavelength from simulated spectra (see Figure 3) c Experimental values taken from ref. 25

Figure 2 shows the main orbitals involved in the transition to the first excited state. In all three molecules, this transition has a strong HOMO‐LUMO character and an oscillator strength > 1, indicating that state S1 is responsible for the fluorescence emission. Noteworthy, the HOMO has a larger contribution from the orbitals of the dimethylaniline moiety and the LUMO is more centered in the

13


View Article Online

DOI: 10.1039/C5CP07274C Table 1 Computed rotational barriers (in kJ mol‐1) in gas phase and in aqueous solution (in parenthesis), absorption and emission wavelengths (in nm) of the most stable isomer with corresponding oscillator strengths of DANIR probes 2a, 2b and 2c in gas phase, dichloromethane and water and LogBB. Experimental emission wavelengths in water and absorption wavelength in dichloromethane are shown for comparison.


Page 14 of 33

View Article Online conjugated double bond chain and cyanide groups as expected for these DOI:push‐ 10.1039/C5CP07274C


Figure 2 Representation of HOMO and LUMO orbitals of 2a, 2b and 2c.

The computed absorption wavelengths in CH2Cl2 of the most stable isomers are 376, 431 and 473 nm for 2a, 2b and 2c respectively. These values are very similar to those obtained by Arul Murugan and co‐workers, the larger difference being 19 nm for 2c.46 Changing the solvent from dichloromethane to water has very little effect on the absorption wavelength but both the values in dichloromethane and water are between 30 and 68 nm larger than the values computed in gas phase. The large preference for the linear isomer and the high rotational energy barriers (ΔG‡ > 30 kJ mol‐1) indicate that the molecule will exhibit little flexibility in aqueous solution. Therefore, dye flexibility is expected to affect the absorption properties only to a small extent. In any case, we decided to run a 10 ps ab initio molecular dynamics simulation for each molecule to reproduce the experimental spectra. Note that, the spectra was constructed from TD‐DFT calculations of the vertical excitation energy on 40 equally distributed snapshots for each trajectory.66 The histograms obtained from the excitation energies in dichloromethane and water are reported in Figure 3, and the corresponding maximum absorption wavelengths reported in Table 1.

14


pull complexes.


View Article Online


DOI: 10.1039/C5CP07274C

Figure 3 Simulated absorption spectra of 2a, 2b and 2c in water (blue) and dichloromethane (orange) from TD‐DFT CAM‐B3LYP calculations on PBE‐D3 molecular dynamics snapshots. Maximum absorption wavelengths in dichloromethane(water) are 385 (385) nm, 455 (435) nm and 505 (505) nm for 2a, 2b and 2c, respectively.

The simulated absorption spectra for the two solvents are very similar and the maximum computed wavelengths are in fair agreement with experimental values. The maximum absorption wavelength is about 10 nm red‐shifted with

15


Page 15 of 33


Page 16 of 33

View Article Online respect to the vertical excitation of the CAM‐B3LYP optimized linear isomer in DOI: 10.1039/C5CP07274C

composed of two factors: the different level of theory used for determining the ground state geometries and the flexibility of the molecule. The latter accounts for the bandwidth, which becomes larger as the number of double bonds increases. Noteworthy the bandwidth is always smaller than in the case of NIAD‐ 4, a much flexible dye in which several isomers contribute to the light


absorption.30 The computed emission wavelengths in gas phase, dichloromethane and water are also reported in Table 1. The computed values are in agreement with experimental data, differences between computed and experimental values being smaller than 40 nm. These values indicate a red‐shifting of the emission wavelength when increasing solvent polarity (bathocromic shift). This effect is more pronounced for compound 2c, which presents the larger number of double bonds between the aromatic ring and the acceptor group. All these data suggest that the S0 to S1 transition has an important charge transfer character as expected for these donor–acceptor dyes (see below for further analysis). Note that increasing the number of double bonds in the spacer chain has two beneficial effects on the emission wavelength: i) the larger number of ‐bonds moves the em to the infrared region as evidenced by the values in gas phase and ii) it increases the charge transfer character of the transition which also moves the em to the infrared region when increasing solvent polarity. Overall, the emission wavelength of 655 nm for compound 2c, suggests that this compound may be a good candidate for in vivo application. Moreover, the

16


the case of 2a and 2b, and 30 nm in the case of 2c (Table 1). This effect is

Page 17 of 33


View Article Online computed values of logBB,67 which is an indicator of the propensity DOI: of 10.1039/C5CP07274C the

(octanol‐water partition) and TPSA (Topological Polar Surface Area),68 show that 2c as well as 2a and 2b have a logBB close to the optimal range ‐1 < logBB < 0,67 suggesting that they can rapidly cross the BBB after intravenous injection. trans to cis isomerization in S0 and S1 states.


As already mentioned in the introduction, the trans to cis or cis to trans isomerization in the S1 excited state is related to a non‐radiative relaxation process involving a conical intersection at a geometry where the double bond is twisted by about 90o in many different dyes presenting C=C double bonds.35–37,69 The spacer between the donor and acceptor groups in DANIR probes contains a conjugated double bond chain28 that can undergo trans to cis isomerization in the S1 state and thus offer a non‐radiative decay pathway back to the ground state. Interestingly, 2a presents no fluorescence emission in solution, in contrast to 2b and 2c.28 With the aim of evaluating the feasibility of achieving a conical intersection by isomerization in the S1 state, we have performed constrained optimizations along all C=C double bond. In these scans the torsional angles defining the rotation around the double bond are kept fixed, while all other geometrical parameters are allowed to relax. This methodology was adopted before for Retinal dye.69 Since TD‐DFT is not suitable for describing the CI region due to its multireference nature,36 we limited our exploration between 0 < ϕ1 < 50 degrees. This allows us having a brute estimation of the energy barrier for achieving the conical intersection in the excited states of 2a, 2b and 2c and more importantly a qualitative estimation of the effect of enlarging the spacer.

17


molecule to cross the Blood‐Brain Barrier (BBB) based on the values of logP


Page 18 of 33

View Article Online Figure 4 shows the energetics associated with easiest rotation around one C=C DOI: 10.1039/C5CP07274C

the energetics associated to the rotation around all C=C bonds.


Figure 4 Potential energy (kJ mol‐1) of the S1 state associated with the rotation around double bonds.

In the S0 ground state, the rotations around the C=C double bonds produce a significant destabilization of the system. The same rotations in the S1 state produce initially a similar destabilization of this state. However, for the case of 2a, when the torsional angle reaches a value between 50 degrees, the energy of the S1 state starts decreasing. The energy difference between the full trans optimized geometry for the S1 state and the highest point of the scan is very low (less than 1 kJ mol‐1), indicating that the conical intersection can be reached through an essentially barrierless process for 2a. Extending the conjugated double bond chain produces a dramatic change in the feasibility of the C=C double bond rotations. For 2b and 2c, the energy of the S1 state keeps increasing until at least 50 degrees. This destabilization of the S1 state is larger when increasing the number of double bonds, and this is regardless of the considered double bond. This suggests that the non‐radiative decay through isomerization is much less efficient for 2b and 2c and this correlates with the experimentally

18


double bond at the S1 electronic states for dyes 2a, 2b and 2c. Figure S4 shows

Page 19 of 33


View Article Online determined fluorescence intensity in solution: There is no fluorescence emission DOI: 10.1039/C5CP07274C

responsible of 2a non‐radiative relaxation. Moreover, interaction with the A fibrils may prevent this process due to a decrease in dye flexibility. This could then explain at least in part the fluorescence enhancement observed upon interaction with A deposits. For that, DANIR probes should bind the fibril in small channels. It is for this reason that we explored the preferred sites of the Published on 20 January 2016. Downloaded on 25/01/2016 14:55:59.

smallest 2a and largest 2c probes in the experimentally proposed fibrils I, II and III.39,40,70 Binding to amyloid fibrils The interaction of derivative 2a and 2c with Aβ fibrils has been studied by considering three different models of Aβ aggregates reported by Tycko.39,40,70 Structure I70 has a two‐fold symmetry and was obtained from fibrils grown in‐ vitro. Structures II40 and III have a three‐fold symmetry, but different origins. While structure II was obtained from fibrils grown in vitro, structure III39 was obtained ex vivo from a human brain affected by Alzheimer’s disease. Although other models exist,71 we considered those derived from Aβ(1‐40) only. All three models share the same organization into protofilaments, which are the basic structural unit of Aβ fibrils and are composed of two β‐strand segments (residues 10‐22 and 30‐40) forming two parallel β‐sheets (cross‐β unit) separated by a loop in residues 23‐29. The protofilaments can be arranged into structures with an overall two‐fold (I) or three‐fold symmetry (II,III). The search for the binding sites has been conducted using the program PELE.61,62 The most

19


for 2a but this emission is remarkable for 2b and 2c.28 This process can thus be


Page 20 of 33

View Article Online stable binding poses for compound 2a and 2c with each fibril model are reported DOI: 10.1039/C5CP07274C

(p‐hydroxyphenyl)‐2,2’‐bithienyl‐5‐yl]‐methylidene}‐ propanedinitrile)33,25 are shown for comparison. It is interesting to compare the results for the three markers because even though they share a hydrophobic linear structure and the same acceptor group, they have different π bridges and donor groups.


Results indicate that all markers are accommodated in the same binding pocket in fibrils I and II. These pockets are formed by 8 Å x 11 Å and 10 Å x 11 Å section channels, whose size suggest a decrease in the probe flexibility. Moreover, in all cases the interaction is dominated by dispersion forces and involves the π bridge and MET35 residues. While NIAD‐4 is able to form a hydrogen bond involving the hydroxyl group and a carbonyl group of the backbone, compounds 2a and 2c are bound exclusively by dispersion forces. Despite the hydrogen bond formed in the interaction of NIAD‐4 with the fibril, PBE‐D2 interaction energies suggest that NIAD‐4 and DANIR‐2c interact similarly and more strongly than DANIR‐2a, due to smaller contribution of dispersion forces. (Table S2). Therefore, assuming a similar stabilization of all dyes in solution, calculations suggest similar affinities for A deposits for NIAD‐4 and 2c probes and a much weaker affinity for 2a. Interestingly, it has been suggested72 that fiber II may accommodate small organic molecules at the hollow core defined by methionine residues of all three protofilaments. This cavity is located at the very center of the fibril and runs parallel to its axis, so that it could accommodate the molecule providing interaction with all three protofilaments, resulting in an important stabilization. However, our simulations indicate that it is more favorable to accommodate the

20


in Figure 5, and the corresponding results obtained for the marker NIAD‐4 ({[5’‐

Page 21 of 33


View Article Online molecule at the corner of the triangular fibrils, where the smaller cavity defined DOI: 10.1039/C5CP07274C

larger stabilizing dispersive interactions. This preference may depend on a variety of factors, including the size and nature of the ligand and the flexibility of the side chains that define the channel domain and thus, present results may not


be general.

Figure 5 Graphical representation of the binding poses of DANIR derivative 2a, 2c and NIAD‐4 on three different models of Aβ fibrils reported by Tycko et. al.39,40,70 The view is along the axis of the fiber. PDB codes: I 2LMN; II 2LMQ; III 2M4J.

Finally, in structure III, molecule 2c is accommodated within the loop of a single protofilament and is interacting with PHE20, ILE31 and LYS28 side chains. On the other hand, NIAD‐4 and 2a are preferably accommodated at the juncture of two protofilaments and interacting with HIS13, GLU11, GLY38, and VAL39. In all cases the linear structure of the molecule interacts with the hydrophobic

21


by methionines and glycines of only two of the three protofilaments leads to


Page 22 of 33

View Article Online channels composed by regular rows of nonpolar side chains arising from the DOI: 10.1039/C5CP07274C

or the other is not clear and may be related to a variety of factors. In summary, DANIR‐2c28 presents very promising structural and photophysical properties that make it a good candidate for A aggregate detection. It presents a Log BB value in the desired range, an emitting wavelength in the NIR region and


interacts with A fibrils through dispersion forces showing preferential binding sites characterized by the presence of small pockets in which its flexibility is reduced. This, in turn, can be responsible of the fluorescence signal enhancement (about 10‐fold) by making the conical intersection geometries less accessible. Since the large emission wavelength is strongly related to the charge transfer character of the S0 to S1 transition and the presence of the conjugated double bond chain, we decided to further analyze this transition. Electronic structure analysis of S0 and S1 states. Charge transfer and electron delocalization The nature of the S0 to S1 excitation has been analyzed by computing the dipole moments of the ground and S1 excited state as well as the DCT index, which qualitatively measures the spatial extension of the charge transfer transition.42 Furthermore, we have evaluated the electron delocalization for both the aromatic ring and the double bond conjugated chain of 2b and 2c in both S0 and S1 states by means of the HOMA index.44 It is worth mentioning at this point that HOMA index has been recently used in conjugated acyclic molecules showing an asymptotic behavior.45

22


cross‐β architecture of amyloid fibrils. The preference of one probe for one site

Page 23 of 33


Article Online Table 2 reports the dipole moments of the S0 and S1 states at the S0DOI: and SView 1 10.1039/C5CP07274C

smaller than those of the excited state, regardless of the considered dye. This is indicative of a charge transfer transition and is in agreement with the nature of the molecular orbitals mainly involved in the transition (HOMO and LUMO, Figure 2). Both orbitals are delocalized over the π system, but while the HOMO is more polarized on the donor group, LUMO is more polarized toward the


electron‐withdrawing end of the molecule. The difference between the S0 and S1 dipole moments is larger for the longer 2c compound, which shows the importance of the spacer length. Noteworthy, optimization of the excited state geometry produces a decrease of the dipole moment showing that geometry relaxation partially compensates the charge transfer. The DCT index is defined as the distance between the barycenters of charges associated to the zones of increase and depletion of the electronic density upon excitation.42 The larger the value of the index the more separated the barycenters of charge is and thus, the more pronounced charge transfer character of the transition. The values of DCT computed for compounds 2a, 2b and 2c at the S0 and S1 optimized geometries are reported in Table 2. Figure 5 shows the representation of the density difference ρ(S1)‐ρ(S0) and the barycenters of charge depletion and increase. Table 2 Dipole moments of S0 and S1 state at the S0 and S1 optimized geometries, HOMA and DCT index (in Å).

2a

2b

2c

μS0//S0

11.0

12.4

13.8

μS1//S0

15.5

18.9

22.5

23


optimized geometries. The dipole moments of the ground state are always


Page 24 of 33

View Article Online

13.6

DOI: 10.1039/C5CP07274C 15.6

μS1//S1

14.4

17.6

21.0

HOMA S0 aromatic ring

0.89

0.91

0.92

HOMA S1 aromatic ring

0.67

0.73

0.77

HOMA S0 double bonds

‐

0.72

0.67

HOMA S1 double bonds

‐

0.95

0.97

DCT S0 geometry

2.0

2.7

3.4

DCT S1 geometry

1.3

2.1

2.6

Figure 5 Representation of the density difference S1‐S0, isovalue 0.00005. Red indicates increase and blue depletion of the electron density. Barycenters of charge depletion and increase are also shown.

Results indicate that, at the S0 optimized geometry, the value of the DCT index increases steadily with the number of double bonds, confirming that the spacer composed by the conjugated double bonds is effective in red‐shifting the energy of the transition not only by increasing the number of double bonds but also by enlarging the charge transfer character of the transition. The representation of the barycenters of charge depletion (blue) and increase (red) shows that the

24


11.9


μS0//S1

Page 25 of 33


View Article Online electron density is transferred from the aromatic ring to the second DOI: closest 10.1039/C5CP07274C

spatial extent of the charge transfer is inferior to the distance between the donor and acceptor groups43 and show that the π bridge is an active participant in the transition. On the other hand, Table 2 and Figure 5b indicate that the geometry relaxation of


the S1 state results in a significant reduction of the spatial extent of the charge transfer. That is, both the DCT index and the dipole moment of the excited state show a partial compensation of charge separation at the optimized S1 state. Since excitation makes the electron density of the aromatic ring decrease while that of the conjugated double bond chain increases, we decided to compute the geometric HOMA index for the 6‐membered ring moiety as well as for the conjugated double bond chain at the optimal S0 and S1 geometries. This will allow us understanding how the excitation process modifies the electron delocalization of the system. Results are summarized in Table 2. At the ground state geometry the six membered ring presents a strong aromatic character (HOMA values between 0.89 and 0.92), while the spacer behaves essentially as a normal conjugated double bond chain, the computed values lying between 0.67 and 0.72.45 In contrast, relaxation of the S1 excited state leads to a structure in which the 6‐membered ring loses part of its aromatic character (HOMA values decrease to around 0.67–0.77), while the conjugated bonds of the spacer show larger electron delocalization (HOMA between 0.95 and 0.97). Therefore, the loss of aromaticity in the benzene ring due to electronic excitation is compensated by the delocalization of the electron density in the double bonds of the spacer that stabilizes the S1 electronic state. That is, at the S1 optimized geometry the

25


double bond to the electron‐withdrawing group. These results confirm that the


Page 26 of 33

View Article Online conjugated double bond chain exhibits larger electron delocalization than the 6‐ DOI: 10.1039/C5CP07274C

Overall, the dipole moment and the charge transfer transition and HOMA indexes illustrate that the π bridge is an active participant in the transition and not just a passive spacer between the donor and acceptor groups. This implies that structural modifications of the π bridge, even those keeping the spatial


separation between the donor and acceptor constant, may greatly affect the charge transfer transition and the final emission wavelength.

Conclusions The photophysical properties of DANIR 2a, 2b and 2c markers for amyloid aggregates have been studied by a combination of computational techniques (DFT and TD‐DFT static calculations, ab initio molecular dynamics simulations and protein energy landscape exploration). These three dyes differ only in the number (one, two or three) of conjugated double bonds between the N,N’‐ dimethylaniline and a propanedinitrilene donor and acceptor groups. Results show that the linear isomers of 2b and 2c are by far the most stable and thus, the ones that determine the absorption and emission wavelength. Moreover, in the excited state, these all trans isomers can undergo trans to cis isomerization through a conical intersection, which involves a non‐emitting relaxation process when the double bond is twisted by 90 degrees. Noteworthy, this rotation is barrierless for the smallest 2a marker and presents a significant barrier for 2b and 2c, which rapidly increases with the number of double bonds. This non‐ radiative deactivation can thus, be responsible for the absence of fluorescence

26


membered ring.

Page 27 of 33


View Article Online signal for 2a in solution. Regarding the interaction with A fibrils, DOI:PELE 10.1039/C5CP07274C

voids of these aggregates, where the molecule’s flexibility is reduced. Since the interaction is essentially by dispersion forces the bigger 2c dye shows larger affinity for the A fibrils. Overall, all these computational data confirms that DANIR‐2c presents promising properties as A aggregates marker: an emitting


wavelength in the Near Infrared Region, a good penetration through the Blood Brain barrier according to the Log BB parameter and a good affinity for the amyloid‐ fibrils in which it interacts by dispersion forces. Analysis of the S0  S1 excitation reveals that the transition is mainly of charge transfer nature. This charge transfer character increases with the number of double bonds in the spacer. This is confirmed by the dipole moments of the S0 and S1 electronic states and the DCT index, which are significantly larger for 2c. Therefore, increasing the number of double bonds has two beneficial contributions: i) it increases the em wavelength per se as the  system becomes larger and ii) it increases the effect of polarity in red‐shifting the emission wavelength. Interestingly, the spatial extension of the charge transfer is inferior to the distance between the donor and acceptor groups. Moreover, at the S1 optimized geometry, the conjugated C=C double bond chain exhibits larger HOMA index than that of the ring suggesting a larger electron delocalization in the former. These two factors indicate that the π bridge is an active participant in the transition and, consequently, modifications on the conjugated C=C double bond chain can affect the charge transfer transition and the final emission wavelength.

27


explorations show that DANIR molecules are accommodated in the hydrophobic


Acknowledgements

Page 28 of 33

View Article Online

DOI: 10.1039/C5CP07274C

59544‐P) and the Generalitat de Catalunya (2014SGR‐482) and Barcelona Supercomputing Center (QCM‐2015‐2‐0030). MS acknowledges the Generalitat de Catalunya for the 2011 ICREA Academia award and XSM for a Professor Agregat Serra Húnter position.


28


The authors gratefully acknowledge financial support from MINECO (CTQ2014‐



References

View Article Online

DOI: 10.1039/C5CP07274C

1

K. Blennow, M. J. de Leon and H. Zetterberg, Lancet, 2006, 368, 387–403.

2

H. W. Querfurth and F. M. Laferla, N. Engl. J. Med., 2010, 326, 329–344.

3

D. J. Selkoe, Physiol Rev, 2001, 81, 741–766.

4

J. Hardy and D. J. Selkoe, Science (80‐. )., 2002, 297, 353–357.

5

R. E. Tanzi and L. Bertram, Cell, 2005, 120, 545–555.

6

H. F. Kung, ACS Med. Chem. Lett, 2012, 3, 265–267.

7

C. A. Mathis, N. S. Mason, B. J. Lopresti and W. E. Klunk, Semin. Nucl. Med., 2012, 42, 423–432.

8

E. Karran, M. Mercken and B. De Strooper, Nat. Rev. Drug Discov., 2011, 10, 698–712.

9

M. E. Phelps, Proc. Natl. Acad. Sci. U. S. A., 2000, 97, 9226–9233.

10

W. E. Klunk, B. J. Lopresti, M. D. Ikonomovic, I. M. Lefterov, R. P. Koldamova, E. E. Abrahamson, M. L. Debnath, D. P. Holt, G. F. Huang, L. Shao, S. T. DeKosky, J. C. Price and C. A. Mathis, J Neurosci, 2005, 25, 10598–10606.

11

A. Nordberg, Lancet Neurol., 2004, 3, 519–527.

12

D. Yanagisawa, T. Amatsubo, S. Morikawa, H. Taguchi, M. Urushitani, N. Shirai, K. Hirao, A. Shiino, T. Inubushi and I. Tooyama, Neuroscience, 2011, 184, 120–127.

13

14

H. Hampel, S. Lista, S. J. Teipel, F. Garaci, R. Nisticò, K. Blennow, H. Zetterberg, L. Bertram, C. Duyckaerts, H. Bakardjian, A. Drzezga, O. Colliot, S. Epelbaum, K. Broich, S. Lehéricy, A. Brice, Z. S. Khachaturian, P. S. Aisen and B. Dubois, Biochem. Pharmacol., 2014, 88, 426–49. A. Schmidt and J. Pahnke, J. Alzheimers. Dis., 2012, 30, 651–64.

29


Page 29 of 33


15

Page 30 of 33

View Article Online M. Staderini, M. A. Martin, M. L. Bolognesi and J. C. Menendez, Chem. Soc. DOI: 10.1039/C5CP07274C Rev., 2015, 44, 1807–1819.

16

V. Ntziachristos, Annu. Rev., 2006, 8, 1–33.

17

H. Tong, K. Lou and W. Wang, Acta Pharm. Sin. B, 2015, 5, 25–33.

18

M. D. Ikonomovic, W. E. Klunk, E. E. Abrahamson, C. A. Mathis, J. C. Price, N. D. Tsopelas, B. J. Lopresti, S. Ziolko, W. Bi, W. R. Paljug, M. L. Debnath, C. E. Hope, B. A. Isanski, R. L. Hamilton and S. T. DeKosky, Brain, 2008, 131, 1630–1645.


19

N. Amdursky, Y. Erez and D. Huppert, Acc. Chem. Res., 2012, 45, 1548– 1557.

20 21 22 23 24 25

M. Ono, M. Ishikawa, H. Kimura, S. Hayashi, K. Matsumura, H. Watanabe, Y. Shimizu, Y. Cheng, M. Cui, H. Kawashima and H. Saji, Bioorg. Med. Chem. Lett., 2010, 20, 3885–8. M. Ono, H. Watanabe, H. Kimura and H. Saji, ACS Chem. Neurosci., 2012, 3, 319–324. X. Zhang, Y. Tian, P. Yuan, Y. Li, M. a Yaseen, J. Grutzendler, A. Moore and C. Ran, Chem. Commun. (Camb)., 2014, 50, 11550–3. X. Zhang, Y. Tian, Z. Li, X. Tian, H. Sun, H. Liu, A. Moore and C. Ran, J. Am. Chem. Soc., 2013, 135, 16397–16409. R. Chongzhao, X. Xiaoyin, S. B. Raymond, B. J. Ferrara, K. Neal, B. J. Bacskai, Z. Medarova and A. Moore, J. Am. Chem. Soc., 2009, 131, 15257–15261. E. E. Nesterov, J. Skoch, B. T. Hyman, W. E. Klunk, B. J. Bacskai and T. M. Swager, Angew. Chemie Int. Ed., 2005, 44, 5452–5456.

26

S. T. Meek, E. E. Nesterov and T. M. Swager, Org. Lett., 2008, 10, 2991–3.

27

S. B. Raymond, J. Skoch, I. D. Hills, E. E. Nesterov, T. M. Swager and B. J. Bacskai, Eur. J. Nucl. Med. Mol. Imaging, 2008, 35, 93–98.

28

M. Cui, M. Ono, H. Watanabe, H. Kimura, B. Liu and H. Saji, J Am Chem Soc,

30



2014, 136, 3388–3394.

View Article Online

DOI: 10.1039/C5CP07274C

29

A. Dreuw and M. Head‐gordon, Chem. Rev., 2005, 105, 4009–4037.

30

D. Jacquemin, E. A. Perpète, I. Ciofini and C. Adamo, Acc. Chem. Res., 2009, 42, 326–334.

31

D. Jacquemin, B. Mennucci and C. Adamo, Phys. Chem. Chem. Phys., 2011, 13, 16987.


32

D. Jacquemin, Y. Zhao, R. Valero, C. Adamo, I. Ciofini and D. G. Truhlar, J. Chem. Theory Comput., 2012, 8, 1255–1259.

33

F. Peccati, J. Hernando, L. Blancafort, X. Solans‐Monfort and M. Sodupe, Phys. Chem. Chem. Phys., 2015, 17, 19718–19725.

34

R. Sabate, L. Rodriguez‐Santiago, M. Sodupe, S. J. Saupe and S. Ventura, Chem. Commun., 2013, 49, 5745–7.

35

O. Valsson and C. Filippi, J. Chem. Theory Comput., 2010, 6, 1275–1292.

36

M. Filatov, J. Chem. Theory Comput., 2013, 9, 4526–4541.

37

M. Huix‐Rotllant, M. Filatov, S. Gozem, I. Schapiro, M. Olivucci and N. Ferré, J. Chem. Theory Comput., 2013, 9, 3917–3932.

38

F. Molnar, M. Ben‐Nun, T. J. Martínez and K. Schulten, J. Mol. Struct. THEOCHEM, 2000, 506, 169–178.

39

J. Lu, W. Qiang, W. Yau, C. D. Schwieters, S. C. Meredith and R. Tycko, Cell, 2013, 154, 1257–1268.

40

R. Tycko, Annu. Rev. Phys. Chem., 2011, 62, 279–299.

41

Härd and Torleif, Phys. Chem. Lett., 2014, 5, 607–614.

42

T. Le Bahers, C. Adamo and I. Ciofini, J. Chem. Theory Comput., 2011, 7, 2498–2506.

31


Page 31 of 33


43

Page 32 of 33

View Article Online I. Ciofini, T. Le Bahers, C. Adamo, F. Odobel and D. Jacquemin, J. Phys. Chem. DOI: 10.1039/C5CP07274C C, 2012, 116, 11946–11955.

44

J. Kruszewski and T. M. Krygowski, Tetrahedron Lett., 1972, 13, 3839.

45

J. C. Dobrowolski and S. Ostrowski, RSC Adv., 2015, 5, 9467–9471.

46

N. A. Murugan, R. Zaleśny, J. Kongsted, A. Nordberg and H. Agren, Chem. Commun. (Camb)., 2014, 12–14.


47

C. A. Guido, P. Cortona, B. Mennucci and C. Adamo, J. Chem. Theory Comput., 2013, 9, 3118–3126.

48

T. Yanai, D. P. Tew and N. C. Handy, Chem. Phys. Lett., 2008, 393, 51–57.

49

M. J. Frisch, G. W. Trucks and H. B. Schlegel et al., Gaussian 09, Revis. A.02, 2009.

50

R. Ditchfield, W. J. Hehre and J. A. Pople, J. Chem. Phys., 1971, 54, 724.

51

A. V. Marenich, C. J. Cramer and D. G. Truhlar, J. Phys. Chem. B, 2009, 113, 6378–6396.

52

R. Cammi, B. Mennucci and J. Tomasi, J. Phys. Chem. A, 2000, 104, 5631– 5637.

53

J. Tomasi, B. Mennucci and R. Cammi, Chem. Rev., 2005, 105, 2999–3093.

54

J. VandeVondele, M. Krack, F. Mohamed, M. Parrinello, T. Chassaing and J. Hutter, Comput. Phys. Commun., 2005, 167, 103–128.

55

J. Perdew, K. Burke and M. Ernzerhof, Phys. Rev. Lett., 1996, 77, 3865– 3868.

56

S. Grimme, J. Comput. Chem., 2006, 27, 1787–1799.

57

S. Grimme, J. Comput. Chem., 2004, 25, 1463–1473.

58

C. Hartwigsen, S. Goedecker and J. Hutter, Phys. Rev. B, 1998, 58, 3641–

32



View Article Online

DOI: 10.1039/C5CP07274C

59

S. Goedecker, M. Teter and J. Hutter, Phys. Rev. B, 1996, 54, 1703–1710.

60

J. VandeVondele and J. Hutter, J. Chem. Phys., 2007, 127, 114105.

61

K. W. Borrelli, A. Vitalis, R. Alcantara and V. Guallar, J. Chem. Theory Comput., 2005, 1, 1304–1311.

62 Published on 20 January 2016. Downloaded on 25/01/2016 14:55:59.

3662.

63 64 65 66

K. W. Borrelli, B. Cossins and V. Guallar, J. Comput. Chem., 2010, 31, 1224– 35. W. L. Jorgensen, D. S. Maxwell and J. Tirado‐Rives, J. Am. Chem. Soc., 1996, 118, 11225–11236. D. Qiu, P. S. Shenki, F. Hollinger and W. C. Still, J. Phys. Chem. A, 1997, 101, 3005. T. M. Krygowski, M. K. Cyran ski, Z. Czarnocki, G. Hafelingerb and A. R. Katritzkyc, J. Chem. Inf. Comput. Sci., 1993, 33, 70–78. N. De Mitri, S. Monti, G. Prampolini and V. Barone, J. Chem. Theory Comput., 2013, 9, 4507–4516.

67

D. E. Clark and S. D. Pickett, Drug Discov. Today, 2000, 5, 49–58.

68

P. Ertl, R. Bernhard and P. Selzer, J. Med. Chem., 2000, 43, 3714–3717.

69

R. Send and D. Sundholm, J. Phys. Chem. A, 2007, 111, 27–33.

70

A. T. Petkova, W. Yau and R. Tycko, Biochemistry, 2006, 45, 498–512.

71

T. Lührs, C. Ritter, M. Adrian, D. Riek‐Loher, B. Bohrmann, H. Döbeli, D. Schubert and R. Riek, Proc. Natl. Acad. Sci. U. S. A., 2005, 102, 17342– 17347.

72

Y. Miller, B. Ma and R. Nussinov, J. Am. Chem. Soc., 2011, 133, 2742–8.

33


Page 33 of 33

Role of coherence and delocalization in photo-induced electron transfer at organic interfaces.

Exciton delocalization, charge transfer, and electronic coupling for singlet excitation energy transfer between stacked nucleobases in DNA: an MS-CASPT2 study.

Protonation induced shifting of electron-accepting centers in intramolecular charge transfer chromophores: a theoretical study.

localization in multibranched push-pull chromophores.

Optical Fluorescence Microscopy for Spatially Characterizing Electron Transfer across a Solid-Liquid Interface on Heterogeneous Electrodes.

An electron energy-loss study of picene and chrysene based charge transfer salts.

Mechanism of back electron transfer in an intermolecular photoinduced electron transfer reaction: solvent as a charge mediator.

Charge separation and charge delocalization identified in long-living states of photoexcited DNA.

Localization-delocalization matrices and electron density-weighted adjacency matrices: new electronic fingerprinting tools for medicinal computational chemistry.

Cyclic π electron delocalization in fluoroborazines.

Real-time simulations of photoinduced coherent charge transfer and proton-coupled electron transfer.

Different mechanisms for hole and electron transfer along identical molecular bridges: the importance of the initial state delocalization.

Photoinduced charge accumulation by metal ion-coupled electron transfer.

Structural factors influencing the intramolecular charge transfer and photoinduced electron transfer in tetrapyrazinoporphyrazines.

Charge transfer-assisted self-limited decyanation reaction of TCNQ-type electron acceptors on Cu(100).

Intramolecular charge transfer photoemission of a silicon-based copolymer containing carbazole and divinylbenzene chromophores. Electron transfer across silicon bridges.

Molecular dynamics and charge transport in organic semiconductors: a classical approach to modeling electron transfer.

Computational methods for optical mapping.

Photoinduced electron transfer in a charge-transfer complex formed between corannulene and Li+@C60 by concave-convex π-π interactions.

Electron transfer dynamics and excited state branching in a charge-transfer platinum(II) donor-bridge-acceptor assembly.

Extreme electron polaron spatial delocalization in π-conjugated materials.

Branching Ratio between Proton Transfer and Electron Transfer Channels of a Bidirectional Proton-Coupled Electron Transfer.

Reaction dynamics and proton coupled electron transfer: studies of tyrosine-based charge transfer in natural and biomimetic systems.

Full-electron ligand-to-ligand charge transfer in a compact Re(I) complex.