Origin of DNA-Induced Circular Dichroism in a ... - ACS Publications

Oct 2, 2017 - Patrick Norman,. †. Mathieu Surin,*,‡ and Mathieu Linares*,†,§. †. School of Biotechnology, Division of Theoretical Chemistry &...
0 downloads 0 Views 2MB Size
Article Cite This: J. Am. Chem. Soc. 2017, 139, 14947-14953

pubs.acs.org/JACS

Origin of DNA-Induced Circular Dichroism in a Minor-Groove Binder Nanna Holmgaard List,† Jérémie Knoops,‡ Jenifer Rubio-Magnieto,‡ Julien Idé,‡ David Beljonne,‡ Patrick Norman,† Mathieu Surin,*,‡ and Mathieu Linares*,†,§ †

School of Biotechnology, Division of Theoretical Chemistry & Biology, KTH Royal Institute of Technology, Roslagstullsbacken 15, 114 21 Stockholm, Sweden ‡ Laboratory for Chemistry of Novel Materials, Center for Innovation and Research in Materials and Polymers, University of Mons−UMONS, Place du Parc, 20, 7000 Mons, Belgium § Swedish e-Science Research Centre, KTH Royal Institute of Technology, 104 50 Stockholm, Sweden S Supporting Information *

ABSTRACT: Induced circular dichroism (ICD) of DNAbinding ligands is well known to be strongly influenced by the specific mode of binding, but the relative importance of the possible mechanisms has remained undetermined. With a combination of molecular dynamics simulations, CD response calculations, and experiments on an AT-sequence, we show that the ICD of minor-groove-bound 4′,6-diamidino-2-phenylindole (DAPI) originates from an intricate interplay between the chiral imprint of DNA, off-resonant excitonic coupling to nucleobases, charge-transfer, and resonant excitonic coupling between DAPIs. The significant contributions from charge-transfer and the chiral imprint to the ICD demonstrate the inadequacy of a standard Frenkel exciton theory of the DAPI−DNA interactions.



INTRODUCTION Since its first synthesis in 1971 by Otto Dann and co-workers,1 4′,6-diamidino-2-phenylindole (DAPI, see Scheme 1) has been

previously reported by different groups, this interaction mechanism produces drastic changes in the (chir)optical properties of DAPI.10,17,18 Besides the enhancement of fluorescence of DAPI upon binding, another important consequence is the appearance of an induced circular dichroism (ICD), i.e., the achiral DAPI acquires a circular dichroism signal when bound to the chiral DNA. This ICD signal is observed at the level of a single DAPI bound in the minor groove at a central AATT sequence of short (dodecamer) dsDNA, and it varies with the length and the sequence of the (AT) tract on longer dsDNA.18 To further exploit the full potential of DAPI in DNA staining, it is important to fully understand the spectroscopic signatures upon binding to DNA. However, the origin of these ICD signals remains undetermined so far. In this Article, we identify the microscopic origins of such ICD signals and provide an estimate of their relative importance, so as to shed light on chiroptical investigations of biological processes involving DNA. To that aim, we investigate in depth the binding of DAPI to a double-stranded DNA with a model (AT) sequence, through a joint theoretical and experimental approach. This sequence has been selected since it is wellestablished that DAPI binds to the minor groove on (AT) sequences, whereas an intercalation binding mode could occur on (GC) sequences, which is still controversial.9−16,19 Besides, a previous study on the interaction between DAPI and single-

Scheme 1. Chemical Structure of DAPI

extensively used for biological applications.2,3 Today, it remains a popular fluorescent stain for DNA visualization and quantitation,4 and is extensively used for studying the cellular nuclei under different environments and stresses.5,6 A recent report reveals that DAPI can be effectively used for constructing DNA density maps in the imaging of nuclei and chromosomes using single-molecule localization microscopy, which is promising to further understand DNA repair and DNA replication at the molecular scale.7 More recently, DAPI has also been exploited to probe the condensation of chromatin in human chromosomes.8 DAPI preferentially binds in the minor groove of doublestranded DNA (dsDNA) with a high AT content. This DNA binding mode has been evidenced by crystallography, NMR data, spectroscopy, and molecular modeling, showing a stable minor-groove binding on (AT) sites of dsDNA.9−16 As © 2017 American Chemical Society

Received: June 9, 2017 Published: October 2, 2017 14947

DOI: 10.1021/jacs.7b05994 J. Am. Chem. Soc. 2017, 139, 14947−14953

Article

Journal of the American Chemical Society

therefore added at the extremities of the (AT) sequence to avoid an unpairing close to the binding sites. Either one DAPI (1D-DNA simulation) or four DAPIs (4D-DNA simulation) were docked in the DNA minor groove, and MD simulations were performed for those assemblies as well as for pure DAPI in aqueous solution. Different starting conformations for the 1D-DNA simulation and arrangements for the 4D simulation (head-to-tail, head-to-head, and tail-to-tail) have been considered. The MD simulations were prepared and performed with Amber 16 and Ambertools 16,22 using the parmbsc1 for DNA23 and a reparameterized version of the gaff 2.1 force field for DAPI.24 Time-dependent density functionnal theory (TDDFT) calculations employing the CAM-B3LYP functional25 and varying basis sets were performed on different models, referred to as XD-Y@ ZD-W, where X denotes the number of DAPI molecules and Y the treatment of the environment in the QM calculations, while Z and W denote the same but refer to the MD simulation from which the structures originate. Further details of the calculations are given in the Supporting Information (SI). All molecular graphics were created with the UCSF Chimera package.26 DNA−DAPI Interactions. In order to describe the direct electronic effects of DNA on the induced CD of DAPI, a model of a single DAPI bound to the minor groove of DNA, comprising DAPI and 6 base pairs, sodium counterions were constructed from three randomly selected snapshots from the 1D-DNA simulation. For this purpose, the entire model system was described at the CAM-B3LYP level of theory while simulating the effects of the solvent by means of the polarizable continuum model27 (see SI for further details). These calculations were performed using Gaussian 09.28 Transition properties of the three snapshots of the model system and the corresponding isolated DAPIs are summarized in Table S3. To examine the relative importance of the various contributions to the computed CD response, the charge-transfer number analysis by Plasser et al.29 was employed to decompose the excitations into fragment contributions. The results of this decomposition into localized excitation, chargetransfer, and exciton character for the two lowest transitions of the three model systems are provided in Figure S10 together with an analogous decomposition of the associated ICD. Details on this analysis is provided in SI section S2.3.4. Dielectric Screening by the Environment. The polarizable embedding (PE) model30,31 was employed to examine the effect of the environment on the transition properties and electronic couplings between DAPIs. The latter was studied within the framework of an exciton coupling model (ECM) including singly excited exciton states.32 In this framework, all the components used to construct the exciton Hamiltonian and rotational strengths, namely site energies, electronic couplings, and electric and magnetic transition dipoles, are computed from PE−TDDFT calculations on the monomers. The electronic couplings in the PE−ECM model were approximated as the Coulomb interaction between transition density-fitted charges of the interacting monomers. Further details on the PE−ECM model are given in SI section S2.3.1. The embedding potential representing the nucleic acids and water molecules consisted of conformationally averaged partial atomic charges and distributed isotropic dipole− dipole polarizabilities taken from refs 33 and 34, respectively. Conformation-specific embedding parameters for DAPI were derived according to the same procedure, but excluding the averaging step. The PE−TDDFT calculations were performed using a locally modified version of the DALTON program,35,36 linked to the PE library37 (environment contributions) and the QFITLIB library38 (transition density-fitted charges). Transition properties for DAPIs obtained in the PE−TDDFT calculations and associated excitonic couplings are given in Tables S4−S6. The convergence of the excitonic CD spectrum with respect to number of DAPI transitions is given in Figure S5. A comparison to the corresponding vacuum calculations (ECM with DAPIs ground-state polarized in the field of other DAPIs and full TDDFT) are summarized in Table S7.

strand oligonucleotides showed that DAPI interact with sequence containing (AAT) as trinucleotide repeating unit, by favoring the folding of the single strand into a hairpin structure with minor groove binding.20 In our spectroscopy experiments, we exploited a sequence of 23 base pairs (bp), for which multiple DAPIs can bind in the DNA minor groove. The results are deciphered through a theoretical approach, which allows us to dissect the effects of DAPI conformation, DNA− DAPI interactions, and DAPI−DAPI interactions on the ICD signals.



METHODS

Experimental Details. Preparation of the Samples. The d(A-T)· d(T-A) oligonucleotide duplex (ODN) was purchased from Eurogentec (Belgium) with the highest purity grade (UltraPureGold, RPHPLC >95% pure in sequence) in the dried state. The sequence of this ODN is 5′-ATA TAT ATA TAT ATA TAT ATA TA-3′ (and reverse complementary sequence). The composition of the DNA was checked by MALDI-ToF. The buffer was prepared by using tris(hydroxymethyl)aminomethane (Tris), EDTA, and Milli-Q water (hereafter named TE buffer). The oligonucleotide was dissolved in a volume of TE buffer (pH 7.4, 10 mM Tris buffer and 1 mM EDTA) at a concentration of 100 μM (experimental estimation error for the stock concentration is around 7%), and it was annealed above the melting temperature and slowly cooled to allow for the duplex formation. The solution of ODN was centrifuged during 2 min at 2000 rpm. A volume of 5 μL of this solution was used in order to prepare a diluted aliquot, to which was added a solution of TE buffer, and the final solution was mixed using a vortex. The concentration of DNA in the buffer solution was determined by UV−vis absorption at 25 °C using the specific extinction coefficient at 260 nm (ε260) of DNA (363844 L·mol−1·cm−1). 4′,6-Diamidino-2-phenylindole dihydrochloride (DAPI) (purity grade 98%) was purchased from Merck and was used without further purification. DAPI was also dissolved in TE buffer using heating and sonication, and the molar ratio between DAPI and DNA was adjusted to the DNA concentration (around 5 μM). DAPI solution stock was added to the DNA solution, and the mixed solution was stirred using the vortex at vigorous speed during 2 min and allowed to equilibrate for at least 10 min. The DAPI solution stock (1 mg/mL) was stored in the dark at 4 °C and was stable for 2−3 weeks. Spectroscopy. The UV−vis absorption, electronic circular dichroism (CD), and fluorescence measurements were recorded using a Chirascan Plus CD spectrometer from Applied Photophysics. The UV−vis and CD measurements were carried out using 2 mm Suprasil quartz cells from Hellma Analytics. The spectra were recorded between 210 and 450 nm, with a bandwidth of 1 nm, time per point of 1 s, and two repetitions. The RMS noise of CD signal is around 0.03 mdeg at 250 nm. The spectra of TE-buffered aqueous solutions were used as baselines and were automatically subtracted from the ECD spectra of the samples. The variable temperature spectroscopic experiments were performed using a TC125 temperature controller from Quantum Northwestern running on the Chirascan Plus spectrophotometer. The temperatures were varied from 20 to 90 °C at rate of 1 °C/min. The temperature within the quartz cells was determined using a temperature probe. The emission spectra were recorded at 20 °C by using a 10 mm quartz cells (1 mL) from Lightpath Optical. The excitation wavelength was set at 364 nm, and the spectra were recorded between 380 and 650 nm, with a bandwidth of 1 nm and time per point of 1 s. Computational Details. The DNA sequence chosen for the simulations was ds(CGCG[AT]10CGCG), thus containing 28 bp in a symmetric sequence. The long AT tract of 20 bp can fit up to four DAPIs in the minor groove, corresponding to a 0.2 DAPI/bp ratio. Four CG base pairs were added at each end to avoid any unpairing of AT base pairs during the molecular dynamics (MD) simulations. Indeed, it has been demonstrated by Zgarbova et al. that base pairs opening (fraying) at the ends of DNA double strands is overestimated in molecular dynamics simulations.21 The four CG base pairs were 14948

DOI: 10.1021/jacs.7b05994 J. Am. Chem. Soc. 2017, 139, 14947−14953

Article

Journal of the American Chemical Society

Figure 1. Spectroscopic properties of DAPI bound to DNA. (a) UV−vis and (b) CD spectra of a double-stranded DNA with alternating d(A-T) sequence upon titration with DAPI. Measurements were done in TE buffer, pH 7.4, and at 20 °C. [DNA] = 5 μM. (c) UV−vis spectra in extinction coefficient. (d) evolution of the ICD of DAPI in difference of extinction coefficient. (e) Fluorescence titration for DNA with DAPI, as monitored at 467 nm (excitation at 364 nm), and each titration performed at 5.0 μM DNA. (f) CD titration for DNA with DAPI. The CD intensity was monitored at 365 nm.



DAPI Binding to DNA. MD simulations provide insights into the changes in DAPI conformation (see next section on structurally induced CD) and in DNA structure upon the binding process. A snapshot from the simulation of 4 DAPI bound to the minor groove of the same DNA (4D-DNA) is shown in Figure 2b. When DAPI binds to DNA, a significant reduction of the minor groove width takes place (Figure S6a). To quantify this structural modification, we estimated the distance between H5′ and H4′ hydrogen atoms as defined by Neidle.45 A strong decrease of the standard deviations upon binding is observed (see Table S1). This characterizes an increase of the stiffness of the DNA double strand, which was reported earlier.18 The H5′−H4′ distance illustrates the reduction of the minor groove width. This reduction does not significantly depend on the number of DAPI bound to the DNA (see Table S1). It is related to a maximization of van der Waals interactions between DAPI and DNA: the H5′−H4′ distance upon binding (5.9 Å) is close to the sum of the thickness of an aromatic ring and twice the radius of a hydrogen atom (5.7 Å), while the native width is of 6.6 Å. This is also shown by the complementarity of van der Waals surfaces of DAPI and DNA upon binding compared to the native form (Bhelix) of DNA (see Figure S6). In this binding mode obtained by the MD simulation (see Figure S7, hydrogen bonds depicted in green), the amine of the phenylindole moiety forms a hydrogen bond with the N3 nitrogen of an adenine, while each amidinium group forms a hydrogen bond with the O2 oxygen of a thymine. These hydrogen bonds (DAPI−adenine and one DAPI−thymine) are highly conserved (99.6% and 98% over 500 snapshots, respectively). While, the hydrogen bond between amidinium and thymine (represented in orange Figure S7) is only observed for ca. 47% of the snapshots (using default criteria: 3.5 Å between donor−acceptor atoms and donor− hydrogen−acceptor angle >120°). Changes in DAPI Conformation upon Binding and Structurally Induced CD. The conformational freedom of DAPI is mainly dictated by the three dihedral angles α, β, and γ (Figure 2b), and the chiroptical properties of the lowest two transitions in DAPI, responsible for the broad long-wavelength absorption band in solution,46 are largely dictated by the central

RESULTS AND DISCUSSION The binding of DAPI to the alternating AT double-stranded DNA sequence with 23bp (d(A-T)·d(T-A)) has been studied by chiroptical spectroscopy, see Figure 1. First, a significant redshift of the maximum of the absorption band of DAPI is observed upon DNA binding (Figure 1a). The largest shift is obtained for 0.15 and 0.2 DAPI/bp, with a maximum at 359 nm (compared to 344 nm for DAPI without DNA). This interaction also leads to a strong ICD around 365 nm (Figure 1b) that may originate from different phenomena: conformational bias, DAPI−DNA interactions, and couplings between several DAPI molecules bound to the same dsDNA.39−42 As seen in Figure 1d, the highest CD per DAPI occurs at 0.2 DAPI/bp with a Δε around 33 M−1 cm−1this intensity is of the same order of magnitude as the ones observed for small chiral molecules.43 The increase of the CD per DAPI up to this ratio suggests that couplings between DAPIs contribute to the ICD and that the ratio of 0.2 DAPI/bp corresponds to a filling of the DNA minor groove. Note that, in the low concentration range used for the experiments (μM), DAPI molecules very likely interact with DNA as single species (no DAPI aggregation), as supported by UV−vis/CD measurements with a heating/cooling cycle (Figure S1). The fluorescence of DAPI increases upon DNA binding (Figures 1c and S2). The highest fluorescence intensity is obtained at the same ratio as the highest CD, namely 0.2 DAPI/bp. According to ref 44, this enhancement can be explained by a reduced geometrical flexibility and a reduced polarization of DAPI when bound to the minor groove. Recently, these authors investigated the absorption and fluorescence properties of DAPI either bound to the DNA minor groove or intercalated between two base pairs, using TDDFT combined with classical models. To our knowledge, however, theoretical modeling has not been used to understand the ICD of DAPI upon DNA binding. In our study, all-atom classical molecular dynamics (MD) simulations combined with quantum-mechanical (QM) response calculations on various model systems were used to dissect the contributions from the possible mechanisms to the ICD. 14949

DOI: 10.1021/jacs.7b05994 J. Am. Chem. Soc. 2017, 139, 14947−14953

Article

Journal of the American Chemical Society

Figure 2. Explanation of sign of the structurally ICD. (a) Four DAPI molecules bound to the minor groove of the ds(CGCG[AT]10CGCG) sequence. (b) Distributions of the dihedral angles α, β, and γ of DAPI bound in the minor groove (colored, 4D-DNA MD run) and in aqueous solution (black line, 1D-water). (c) Isodensity surfaces of the orbitals mainly involved in the lowest transition in DAPI (isovalue: 0.035 au). (d) Electric (blue) and magnetic (green) transition dipole moments for the indole (I) and phenyl (P) fragments in DAPI with angles (α,β,γ) = (0°,25°,0°). (e) Schematic representation of the ϕPi ϕPf distribution (i: HOMO, f: LUMO) as viewed along the x-direction from the indole (thick gray dashed line) to the phenyl (thick black line). The contribution from C4 is behind C2 and of opposite sign. For atom labeling, see panel d. The magnetic transition dipole acquires an x-component antiparallel to the x-direction due to the circular current induced upon excitation by the positive β angle. Note that only relative signs of the transition dipole moments arer meaningful.

dihedral angle β (Table S2 and Figure S8). As shown in Figure 2b, all three dihedral angles are symmetrically distributed around zero in aqueous solution, but upon minor-groove binding, they are strongly shifted toward positive angles, giving an average β angle of 26°. For α and β, the maximum occurrences are obtained in the range from 20° to 30°, while for γ, the range from 30° to 40° has the highest frequency. This is consistent with the values reported by Biancardi et al. for a QM optimized structure (23°, 25°, and 35.5° for α, β, and γ, respectively).44 Those conformational changes break the strict orthogonal arrangement of electric and magnetic transition dipoles in an idealized planar structure, hence giving rise to ICD. The large degree of correlation between the sign of the structurally ICD and β can be explained by the nature of the first (second) electronic excitation, which predominantly involves a transition from the highest occupied molecular orbital (HOMO) (HOMO−1 for the second transition) to the lowest unoccupied molecular orbital (LUMO) (Figures 2c and S9). These excitations can be described as ππ* transitions over the two ring systems involving intramolecular charge transfer from the indole to the phenyl moiety, referred to as fragments I and P, respectively. Rotation around β induces a circular current upon excitation perpendicular to the rotation axis such that the magnetic transition dipole acquires an x-component whose direction, and in turn the sign of the ICD, are determined by the rotation direction of β. Figure 2d shows the I and P contributions to the electric and the magnetic transition dipole moment (blue and green arrows, respectively), dominating the ICD. Further details are given in SI section S3.2. As schematically illustrated in Figure 2e, an analysis of the orbital transition density of fragment P (ϕPi ϕPf ) shows that a positive β leads to an antiparallel alignment of the x-

components of the electric and magnetic transition dipoles. Consequently, the structural component results in a negative ICD (green line, Figure 4b, below), and thus it cannot explain the observed positive bands in the 320−400 nm region reported by Eriksson et al. for a single DAPI bound in the minor groove of a short dsDNA (a dodecamer containing a central AATT sequence, see Figure 3 in ref 18). DAPI−DNA Interactions. As suggested by Frenkel exciton theory, off-resonant excitonic coupling between the nucleobases and minor-groove-bound ligands with electric transition dipoles oriented along the groove, like DAPI, is expected to be the source of the strong, positive ICD indicative for this binding mode.47,48 However, the previous studies did not consider the contribution from the chiral electronic imprint of DNA to the observed ICD nor the possibility of charge-transfer interactions between DAPI and DNA due to their close proximity. The former type of effect describes that the presence of the chiral electronic structure of DNA imprints a chiral electronic response in DAPI beyond that induced by the structurally changes discussed in the previous section. To assess the importance of excitonic delocalization, chiral imprint, and charge transfer in an integrated way, we performed full TDDFT calculations on three snapshots of a truncated model system (denoted QM/PCM) of DAPI bound to a 6 bp d(A-T)·d(T-A) sequence combined with a continuum solvent description. A similar model system was used in the work by Biancardi et al.44 on absorption and fluorescence properties of DAPI. The transitions were then characterized by an atomic-orbital-based decomposition of the one-electron transition density of the two lowest excitations into fragment contributions (DAPI, ions, nucleobases, and sugar phosphate backbone) according to the charge-transfer number analysis in ref 29 (eq 14). This allows 14950

DOI: 10.1021/jacs.7b05994 J. Am. Chem. Soc. 2017, 139, 14947−14953

Article

Journal of the American Chemical Society

and cannot be ignored, affecting the magnitude of the ICD significantly by partially counteracting the exciton ICD. In addition, our results indicate that the electronic imprint of DNA provides a contribution to the ICD for the lowest transition that is of opposite sign of that associated with the biased conformational distribution of DAPI in the minor groove. DAPI−DAPI Interactions. Let us return to the experimental CD results presented in Figure 1d. The increase of the normalized CD signal up to 0.2 DAPI/bp with DAPI concentration indicates that DAPI−DAPI interactions play a role. At the same time, the insensitivity of the absorption spectrum with respect to DAPI concentration points to intrinsically small couplings. This is corroborated by the large distance between neighboring DAPIs (average center-of-mass distance is 8.4 ± 1.9 Å) and the poor alignment of their electric transition dipoles (average angle between long axis of neighboring DAPIs is 97 ± 11°). The coupling strength is further hampered by the substantial energy difference between the lowest transition in different DAPIs caused by structural variations and inequivalent binding sites. The former gives an average energy dispersion of 0.12 eV. As a consequence, the wave functions will be strongly localized to individual chromophores. The interesting question to address computationally is whether a strong cooperative enhancement of the CD signal is possible despite these weak couplings. The most straightforward way to address the ICD due to intermolecular DAPI interactions is to perform full TDDFT calculations on tetramer structures extracted from the MD simulation while neglecting the electronic effects of DNA. The resulting absorption and CD spectra normalized with respect to DAPIs are shown in Figure 4a and c, respectively (blue lines). The negative structurally (monomer) ICD is overwhelmed by that of the excitonic tetramer, leading to a positive band around 375 nm that is about 3−8 times as intense as the monomer signal (green line, Figure 4b). To better understand the behavior of the CD signal, we considered an ECM on a single conformation. The exciton coupling is approximated by the Coulomb interaction between atomic transition charges derived from the transition density of each DAPI.32 This approximation is fully adequate in the present case of weakly interacting monomers, as demonstrated by the agreement between the CD spectra obtained from the ECM and full TDDFT (red and green lines, Figure 4c). The electronic couplings between minor-groove-bound DAPIs are in general weak and only significant between neighboring molecules (see Table S6). For the lowest-lying transition in neighboring DAPIs the couplings are in the order of 5−10 meV, i.e., ∼10−25 times smaller than the average difference in transition energies. This can be compared to a coupling of 134 meV obtained for two π-stacked planar DAPIs separated by 3.5 Å. It is thus clear that the excitonic coupling strengths between minor-groove-bound DAPIs are weak but assisted by the coincidental neardegeneracy of monomer transition energies. In fact, as shown in Figure S13, there is an exponential reduction of the excitonic rotatory strength with increasing transition energy mismatch (ΔE) and hence decreasing coupling strength. However, the excitonic CD signal, which is the convoluted sum of rotatory strengths, is resilient over values of ΔE up to 0.15 eV (see SI section S4 for explanation). The positive long-wavelength band in the couplet (Figure 4c) follows from the excitonic states formed by a weak coupling between the lowest (long-axis polarized) transitions in neighboring DAPIs, as predicted by

one to distinguish localized, charge-transfer, and excitonic contributions. The role of the electronic imprint was determined as the difference between the structurally induced CD and the DAPI contribution to the CD in the model systems. As can be seen in Figure 4d, explicit inclusion of the electronic effects of DNA indeed leads to a positive ICD for the two lowest transitions in all three snapshots, in agreement with experiment. From the structure of the charge-transfer number matrix (Figures 3a and S10, note the logarithmic color scale), it

Figure 3. Decomposition of the ICD signal of a single minor-groovebound DAPI. (a) Electron−hole correlation plots of the chargetransfer number matrix for the lowest transition in the three snapshots of the QM/PCM truncated model system. Note the logarithmic color scale. (b) Atomic-orbital-based fragment decomposition (DAPI and DNA) of the ICD for DAPI bound to the minor groove of the QM/ PCM truncated model systems, showing the relative magnitude of the underlying mechanisms.

follows that the excitations are strongly localized to DAPI. The off-diagonal elements represent the extent of charge-transfer character (both with and without net transfer of charge in the excitation process), whereas the small diagonal elements on DNA fragments, corresponding to exciton delocalization, indicate off-resonant excitonic coupling between DAPI and DNA. While they play only a minor role in the transition density matrix, their respective contributions to the ICD are significant, as seen from the analogous fragment decomposition of the ICD in Figure 3b (see details in SI section S2.3.4). The off-resonant excitonic coupling between DAPI and DNA provides a large positive ICD in line with previous work,48 which is responsible for the overall positive Cotton effect. However, the present full QM treatment shows that chargetransfer contributions are significant for the lowest transition 14951

DOI: 10.1021/jacs.7b05994 J. Am. Chem. Soc. 2017, 139, 14947−14953

Article

Journal of the American Chemical Society

that leads to a reduction of the couplings by a factor of ∼2 (Table S6). Despite this screening and a consequent decrease in CD intensity (black line, Figure 4c), the exciton mechanism between minor-groove-bound DAPIs still persists and leads to a 3-fold enhancement of the ICD for the considered snapshot. What is more important is that the DAPI−DAPI excitonic mechanism is of the same order of magnitude as the contribution from the direct DAPI−DNA interactions. Thus, our results suggest that the ICD of minor-groove-bound DAPI originates from an intricate interplay between several mechanisms, where the bisignate ICD signature from DAPI excitonic couplings is balanced by the positive ICD from DAPI−DNA interactions, as proposed by Kubista et al.47 The ICD keeps increasing beyond a ratio of 0.2 DAPI/bp (Figure 1b). Our computational model cannot explain this observation since it assumes that DAPIs are bound in the minor groove without room for additional molecules. In reality, additional DAPIs can possibly bind to much less favorable binding sites, not explored in this work, such as in the major groove or at the bp extremities of the dsDNA, by π-stacking with the terminal base pairs. Those two binding modes were proposed with dsDNA containing CG bp.19,51,52



CONCLUSIONS In conclusion, we have demonstrated that excitonic coupling between minor-groove-bound DAPIs gives a strong bisignate ICD that dominates over the negative structurally ICD of the bound species. Our analysis also demonstrates the defining role of the electronic coupling between DAPI and DNA, giving rise to (i) electronic imprint, (ii) charge transfer, and (iii) offresonant DAPI−nucleobase excitonic coupling. Taken together, these mechanisms give a strong, positive ICD of the bound monomer DAPI that counter the negative component of the bisignate exciton band when several DAPIs are bound. This explains the overall asymmetric, positive ICD observed in the experiment. The analytical approach developed here is timely and can further help in elucidating the geometries and binding modes of the many DNA binders and intercalators, such as chemotherapeutic drugs and photosensitizers, via chiroptical techniques.

Figure 4. Simulated spectra of minor-groove-bound DAPI(s). Simulated (a) UV−vis and (b,c) CD spectra normalized with respect to the number of DAPI molecules, for a single DAPI in water, and one or four DAPIs bound to the DNA minor groove (1D and 4D, respectively), with and without inclusion of the environment (PE or QM/PCM). The lines indicated by “avg” have been obtained as ensemble averages over 500 MD configurations, with pale colored areas representing standard deviations. The gray-shaded area marks the region overlapping with transitions in dsDNA. All stick spectra have been convoluted with a Gaussian line shape function (fwhm = 0.35 eV).

the exciton chirality rule for the right-handed helical arrangement of DAPIs. The negative band encompasses not only the contribution from the paired higher-energy excitonic states, but also those from exciton couplings involving the second monomer transition. To clarify if the DAPI excitonic mechanism is also possible when the dielectric screening of DNA and water is taken into account, we performed CD calculations on the same conformation but using an ECM formulation within the PE model. In addition to modifying the excitonic couplings, this model includes the static and dynamic reaction field as well as the so-called effective external field effect.49,50 The latter describes the effect of the polarization induced in the surroundings by the probing field. The classical representation of the environment, however, excludes the direct electronic mechanisms (charge transfer and exchange−repulsion) between DAPI and DNA, considered in the previous section. With this in mind, it should be noted that the PE model reproduces the ∼0.2 eV environment-induced blue-shift for the two lowest transitions of DAPI obtained with the QM/PCM models (see Tables S3−S5), whereas it does not capture the extent and direction of the chiral electronic imprint on the ICD. This points to the significance of wave function confinement for the electronic imprint. For the electronic couplings between DAPIs, the environmental effects are dominated by the indirect screening term



ASSOCIATED CONTENT

* Supporting Information S

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/jacs.7b05994. Computational details, additional experimental and computational results, and coordinates for the QM/ PCM model systems, including sections S1−S5, Figures S1−S13, and Tables S1−S7 (PDF)



AUTHOR INFORMATION

Corresponding Authors

*[email protected] *[email protected] ORCID

Patrick Norman: 0000-0002-1191-4954 Mathieu Linares: 0000-0002-9720-5429 Notes

The authors declare no competing financial interest. 14952

DOI: 10.1021/jacs.7b05994 J. Am. Chem. Soc. 2017, 139, 14947−14953

Article

Journal of the American Chemical Society



(31) List, N. H.; Olsen, J. M. H.; Kongsted, J. Phys. Chem. Chem. Phys. 2016, 18, 20234−20250. (32) Steinmann, C.; Kongsted, J. J. Chem. Theory Comput. 2015, 11, 4283−4293. (33) Nørby, M. S.; Steinmann, C.; Olsen, J. M. H.; Li, H.; Kongsted, J. J. Chem. Theory Comput. 2016, 12, 5050−5057. (34) Beerepoot, M. T. P.; Steindal, A. H.; List, N. H.; Kongsted, J.; Olsen, J. M. H. J. Chem. Theory Comput. 2016, 12, 1684−1695. (35) Aidas, K.; et al. WIREs Comput. Mol. Sci. 2014, 4, 269−284. (36) DALTON2016.0, A Molecular Electronic Structure Program; http://daltonprogram.org/, 2015. (37) The PElib Developers. The Polarizable Embedding Library (Development Version); 2016. (38) Steinmann, C. qfitlib-1.0.4; https://zenodo.org/record/14949#. WR759lOGPOY, 2015. (39) Allenmark, S. Chirality 2003, 15, 409−422. (40) Garbett, N. C.; Ragazzon, P. A.; Chaires, J. B. Nat. Protoc. 2007, 2, 3166−3172. (41) Biver, T. Appl. Spectrosc. Rev. 2012, 47, 272−325. (42) Jackson, K.; Mason, S. F. Trans. Faraday Soc. 1971, 67, 966− 989. (43) Harada, N.; Kuwahara, S. In Comprehensive Chiroptical Spectroscopy, Applications in Stereochemical Analysis of Synthetic Compounds, Natural Products, and Biomolecules; Berova, N., Polavarapu, P. L., Nakanishi, K., Woody, R. W., Eds.; Wiley: Hoboken, NJ, 2012; pp 167−215. (44) Biancardi, A.; Biver, T.; Secco, F.; Mennucci, B. Phys. Chem. Chem. Phys. 2013, 15, 4596−4603. (45) Neidle, S. FEBS Lett. 1992, 298, 97−99. (46) Kubista, M.; Aakerman, B.; Albinsson, B. J. Am. Chem. Soc. 1989, 111, 7031−7035. (47) Kubista, M.; Aakerman, B.; Norden, B. Biochemistry 1987, 26, 4545−4553. (48) Kubista, M.; Aakerman, B.; Norden, B. J. Phys. Chem. 1988, 92, 2352−2356. (49) List, N. H.; Jensen, H. J. Aa.; Kongsted, J. Phys. Chem. Chem. Phys. 2016, 18, 10070−10080. (50) List, N. H.; Norman, P.; Kongsted, J.; Jensen, H. J. Aa. J. Chem. Phys. 2017, 146, 234101. (51) Kim, S. K.; Eriksson, S.; Kubista, M.; Nordén, B. J. Am. Chem. Soc. 1993, 115, 3441−3447. (52) Trotta, E.; D’Ambrosio, E.; Ravagnan, G.; Paci, M. Nucleic Acids Res. 1995, 23, 1333−1340.

ACKNOWLEDGMENTS Funding from the Carlsberg Foundation (Grant No. CF150792), the F.R.S.-FNRS (DRaPo, CHIRNATES and MISSHERPA projects), SeRC (Swedish e-Science Research Center), and the Swedish Research Council (Grant No. 6212014-4646) as well as resources provided by the CECI and the Swedish National Infrastructure for Computing (SNIC) at National Supercomputer Centre (NSC) are acknowledged.



REFERENCES

(1) Dann, O.; Bergen, G.; Demant, E.; Volz, G. Justus Liebigs Ann. Chem. 1971, 749, 68−89. (2) Kapuscinski, J. Biotech. Histochem. 1995, 70, 220−233. (3) Hannon, M. J. Chem. Soc. Rev. 2007, 36, 280−295. (4) Chazotte, B. Cold Spring Harbor Protoc. 2011, 2011, 80−82. (5) Versaevel, M.; Grevesse, T.; Gabriele, S. Nat. Commun. 2012, 3, 671. (6) Takata, H.; Hanafusa, T.; Mori, T.; Shimura, M.; Iida, Y.; Ishikawa, K.; Yoshikawa, K.; Yoshikawa, Y.; Maeshima, K. PLoS One 2013, 8, e75622. (7) Szczurek, A. T.; Prakash, K.; Lee, H.-K.; Zurek Biesiada, D. J.; Best, G.; Hagmann, M.; Dobrucki, J. W.; Cremer, C.; Birk, U. Nucleus 2014, 5, 331−340. (8) Estandarte, A. K.; Botchway, S.; Lynch, C.; Yusuf, M.; Robinson, I. Sci. Rep. 2016, 6, 31417. (9) Larsen, T. A.; Goodsell, D. S.; Cascio, D.; Grzeskowiak, K.; Dickerson, R. E. J. Biomol. Struct. Dyn. 1989, 7, 477−491. (10) Wilson, W. D.; Tanious, F. A.; Barton, H. J.; Jones, R. L.; Fox, K.; Wydra, R. L.; Strekowski, L. Biochemistry 1990, 29, 8452−8461. (11) Tanious, F. A.; Veal, J. M.; Buczak, H.; Ratmeyer, L. S.; Wilson, W. D. Biochemistry 1992, 31, 3103−3112. (12) Trotta, E.; D’Ambrosio, E.; Del Grosso, N.; Ravagnan, G.; Cirilli, M.; Paci, M. J. Biol. Chem. 1993, 268, 3944−51. (13) Trotta, E.; Paci, M. Nucleic Acids Res. 1998, 26, 4706. (14) Vlieghe, D.; Sponer, J.; Meervelt, L. V. Biochemistry 1999, 38, 16443−16451. (15) Verma, S. D.; Pal, N.; Singh, M. K.; Sen, S. J. Phys. Chem. B 2015, 119, 11019−11029. (16) Sbirkova-Dimitrova, H. I.; Shivachev, B. Acta Crystallogr., Sect. F: Struct. Biol. Commun. 2017, 73, 500−504. (17) Jansen, K.; Norden, B.; Kubista, M. J. Am. Chem. Soc. 1993, 115, 10527−10530. (18) Eriksson, S.; Kim, S. K.; Kubista, M.; Norden, B. Biochemistry 1993, 32, 2987−2998. (19) Banerjee, D.; Pal, S. K. J. Phys. Chem. B 2008, 112, 1016−1021. (20) Trotta, E.; Del Grosso, N.; Erba, M.; Melino, S.; Cicero, D.; Paci, M. Eur. J. Biochem. 2003, 270, 4755−4761. (21) Zgarbová, M.; Otyepka, M.; Šponer, J.; Lankaš, F.; Jurečka, P. J. Chem. Theory Comput. 2014, 10, 3177−3189. (22) Case, D. A.; et al. AMBER 2016; http://www.ambermd.org, 2016. (23) Ivani, I.; et al. Nat. Methods 2016, 13, 55−58. (24) Wang, J.; Wolf, R. M.; Caldwell, J. W.; Kollman, P. A.; Case, D. A. J. Comput. Chem. 2004, 25, 1157−1174. (25) Yanai, T.; Tew, D. P.; Handy, N. C. Chem. Phys. Lett. 2004, 393, 51−57. (26) Pettersen, E. F.; Goddard, T. D.; Huang, C. C.; Couch, G. S.; Greenblatt, D. M.; Meng, E. C.; Ferrin, T. E. J. Comput. Chem. 2004, 25, 1605−1612. (27) Cancès, E.; Mennucci, B.; Tomasi, J. J. Chem. Phys. 1997, 107, 3032−3041. (28) Frisch, M. J.; et al. Gaussian 09, Revision E.01; Gaussian Inc.: Wallingford, CT, 2013. (29) Plasser, F.; Lischka, H. J. Chem. Theory Comput. 2012, 8, 2777− 2789. (30) Olsen, J. M. H.; Aidas, K.; Kongsted, J. J. Chem. Theory Comput. 2010, 6, 3721−3734. 14953

DOI: 10.1021/jacs.7b05994 J. Am. Chem. Soc. 2017, 139, 14947−14953