Conformational Folding Activates Photoinduced Electron Transfer

Open AccessCCS ChemistryRESEARCH ARTICLES6 Aug 2024Conformational Folding Activates Photoinduced Electron Transfer Shiqing Huang†, Syed Ali Abbas Abedi†, Zhifeng Li†, Rongrong Huang, Xiaoyu Yan, Mohammad Izadyar, Qinglong Qiao, Yu Fang, Zhaochao Xu and Xiaogang Liu Shiqing Huang† Fluorescence Research Group, Singapore University of Technology and Design, Singapore 487372 School of Chemistry and Life Resources, Renmin University of China, Beijing 100872 , Syed Ali Abbas Abedi† Fluorescence Research Group, Singapore University of Technology and Design, Singapore 487372 , Zhifeng Li† CAS Key Laboratory of Separation Science for Analytical Chemistry, Dalian Institute of Chemical Physics, Chinese Academy of Sciences, Dalian 116023 , Rongrong Huang Fluorescence Research Group, Singapore University of Technology and Design, Singapore 487372 , Xiaoyu Yan School of Chemistry and Life Resources, Renmin University of China, Beijing 100872 , Mohammad Izadyar Fluorescence Research Group, Singapore University of Technology and Design, Singapore 487372 Research Center for Modeling and Computational Sciences, Faculty of Science, Ferdowsi University of Mashhad, Mashhad 9177948974 , Qinglong Qiao *Corresponding authors: E-mail Address: [email protected] E-mail Address: [email protected] E-mail Address: [email protected] E-mail Address: [email protected] CAS Key Laboratory of Separation Science for Analytical Chemistry, Dalian Institute of Chemical Physics, Chinese Academy of Sciences, Dalian 116023 , Yu Fang *Corresponding authors: E-mail Address: [email protected] E-mail Address: [email protected] E-mail Address: [email protected] E-mail Address: [email protected] Key Laboratory of Applied Surface and Colloid Chemistry (Ministry of Education), School of Chemistry and Chemical Engineering, Shaanxi Normal University, Xi'an, Shaanxi 710119 , Zhaochao Xu *Corresponding authors: E-mail Address: [email protected] E-mail Address: [email protected] E-mail Address: [email protected] E-mail Address: [email protected] CAS Key Laboratory of Separation Science for Analytical Chemistry, Dalian Institute of Chemical Physics, Chinese Academy of Sciences, Dalian 116023 and Xiaogang Liu *Corresponding authors: E-mail Address: [email protected] E-mail Address: [email protected] E-mail Address: [email protected] E-mail Address: [email protected] Fluorescence Research Group, Singapore University of Technology and Design, Singapore 487372 Cite this: CCS Chemistry. 2024;0:1–10https://doi.org/10.31635/ccschem.024.202404541 SectionsSupplemental MaterialAboutAbstractPDF ToolsAdd to favoritesDownload CitationsTrack Citations ShareFacebookTwitterLinked InEmail Photoinduced electron transfer (PET) is a critical process in many functional materials, underpinning various technological applications (i.e., fluorescent probes and photocatalysts). Despite its significance, the detailed structural dynamics of PET, particularly during the excited state, remain poorly understood. This study investigates the mechanisms of conformational folding and their implications for activating PET in molecular systems characterized by a fluorophore-spacer-receptor configuration. We demonstrate that traditional computational models, primarily based on frontier molecular orbitals, often fall short in capturing these conformational dynamics, leading to inadequate explanations of PET phenomena. With the incorporation of conformational folding, our computational model has achieved excellent agreement with experimental data, thereby resolving several long-standing debates on PET mechanisms. This mechanistic advancement not only deepens our understanding of PET but also opens new avenues for designing advanced functional materials. We have thus successfully demonstrated the imaging of lysosomes in live cells using a PET probe. Download figure Download PowerPoint Introduction The photoinduced electron transfer (PET) mechanism is commonly utilized in the creation of diverse functional materials, including photocatalysts,1 photodynamic therapy agents,2,3 and fluorescent probes.4–9 A comprehensive understanding and precise modeling of the PET process is vital for the rational design of these materials with enhanced performance. However, knowledge of the excited-state dynamics in PET-based functional materials remains limited, and current computational methods encounter numerous difficulties in accurately depicting the PET process, resulting in ongoing debates. Numerous PET-based functional molecules, particularly fluorescent probes, employ the fluorophore-spacer-receptor configuration. In this arrangement, the fluorophore either accepts (a-PET) or donates (d-PET) an electron to the receptor upon photoexcitation, activating the PET process. The resulting electron-transfer (ET) state is "dark" (with an almost zero oscillator strength), quenching fluorescence and activating various other processes, such as triplet state formation.10–12 In fluorescent probes, the receptor often serves as a recognition moiety to selectively detect different analytes, either activating or deactivating the PET process. The resulting changes in fluorescence intensity or lifetime enable precise analyte detection with high spatial and temporal resolution. The spacer in these molecules usually consists of a flexible chain, like saturated hydrocarbon chains, although it can also be eliminated by directly connecting the receptor to the fluorophore.13 Previous studies have also demonstrated that rigidifying the conformation of PET-based sensors can significantly enhance their sensitivity and efficiency.14,15 These probes serve as valuable models for understanding the photophysical processes involved in PET. Theoretical calculations serve as a pivotal technique for chemists to directly "visualize" short-lived ET states. To this end, numerous theoretical frameworks have been developed to explain or predict PET occurrence. The Gibbs free energy (ΔG), calculated using the Rehm–Weller equation, can predict PET, with a negative ΔG indicating significant PET upon photoexcitation.16 The rate of the PET processes can also be estimated using the Marcus equation. 17 However, many parameters are unknown in these equations, such as electronic coupling matrix elements and electron donor–acceptor distance connected by a flexible chain, limiting their predictive power. More recently, the advancement of density functional theory (DFT) and time-dependent density functional theory (TD-DFT) calculations enable computational chemists to directly calculate and compare the energy levels of various excited states. The relative stability between the bright locally excited (LE)/intramolecular charge transfer and dark ET states allows chemists to assess PET feasibility.18 Nonetheless, excited-state calculations require significant computational resources and sophisticated solvent effect corrections. Modeling excited states with considerable charge transfer, like the ET state, remains a bottleneck in TD-DFT calculations, affecting the accuracy of the computational results.19 The most widely used PET descriptor is based on the energy levels of frontier molecular orbitals (FMOs), as proposed by de Silva et al. (Figure 1a).20 According to this model, PET occurs if the receptor introduces a filled (for a-PET) or empty (for d-PET) molecular orbital with an energy level between the fluorophore's highest occupied molecular orbital (HOMO) and lowest unoccupied molecular orbital (LUMO). This FMO descriptor only needs simple ground-state calculations and has been successfully applied to PET-based probe design and development. However, this "intuitive" threshold failed in many PET-based probes and configurations.5 Given the limitations of these computational models, controversies surrounding PET mechanisms persist.21,22 Addressing these controversies and uncovering their molecular origins holds immense importance, as it can provide invaluable insights and rational guidelines for engineering functional materials based on PET principles. Figure 1 | (a) Shows the FMO model for the PET process from the previous work. (b) Details the conformational changes that activate the PET, highlighting the dynamic nature of this process. (c) Presents a schematic of the PET mechanism, influenced by conformational changes, as visualized on the PES. Download figure Download PowerPoint In this study, we have showcased the pivotal role of conformational folding in the excited state, which reduces the spatial distance between the electron donor and acceptor and facilitates more effective electron transfer in fluorophore-spacer-receptor probes (Figure 1b,c). By incorporating the significance of potential conformational folding—a factor often overlooked in prior computational models—we have effectively resolved several longstanding debates concerning the PET mechanism in a range of fluorophores (i.e., naphthalimide, BODIPYs, rhodamines, etc.). Inspired by these results, we have also successfully demonstrated the bioimaging utilities of a PET probe in imaging lysosomes in live cells, further highlighting the practical implications of our research findings. Experimental Methods DFT and TD-DFT were employed to rationalize the luminescent mechanism behind the investigated fluorophores. Geometry optimizations in the ground and excited states were carried out with the ωB97-XD23 functional combined with the def2SVP basis set unless stated otherwise. For the ET states in ethanol, we first optimized the structures with B3LYP24 followed by a single-point energy calculation using ωB97-XD. Frequency analysis was performed to confirm that we have obtained stable structures on the potential energy surfaces (PESs). When the solvent effect (in both ethanol and toluene) was applicable, it was accounted for by using the solvation model based on the solute electron density (SMD) model.25 In these calculations, the electronic energies in the excited states were calculated based on the corrected linear response26 solvent formalism. All DFT/TD-DFT calculations were carried out with Gaussian 16A.27 The ωB97-XD functional was chosen for geometry optimizations due to its reliable performance in capturing long-range electron correlation effects, which are crucial for accurately describing PET processes. The def2SVP basis set was selected to balance computational cost and accuracy, ensuring reliable results while maintaining feasibility for large systems. Additionally, B3LYP was used for initial geometry optimizations of ET states in ethanol, followed by single-point energy calculations using ωB97-XD to refine the electronic energies. The distance of charge transfer is calculated with Multiwfn 3.8.28 xTB29 was used to do the conformation search using the molecular dynamic method, where the temperature was set to 400 K and the modeling time was set to 100 ps. B3LYP-D3(BJ)/6-31G* level optimization was employed to the xTB preoptimized conformations with energies within 3 kcal/mol of the global minimum. The treatment yields various representative conformations of the dyes. Frequency calculations were performed at the same level to confirm the stability of the conformations and to obtain thermodynamic correction data. Further experimental details are available in the Supporting Information. Results and Discussion Experimental validation of the PET mechanism of BN-1 Our investigation commenced with compound BN-1 (Figure 2), a classic PET-based probe.30,31 Our in-house synthesis and measurements revealed that BN-1 had a low quantum yield in polar solvents (2.9% in ethanol) but was highly emissive in nonpolar solvents (69.4% in toluene; Table 1, Figure 2a,b). Furthermore, its fluorescence intensity was pH-sensitive, emitting strongly in low pH (acidic) environments while displaying an extremely low quantum yield in high pH (basic) conditions (Figure 2c,d and Supporting Information Figure S2). In contrast, reference compound BN-2, without the tertiary amino receptor, was highly emissive in solvents with varying polarities ( Supporting Information Figure S1a,d and Table S1). Figure 2 | (a) Photophysical properties of BN-1 in various solvents, including peak UV–vis absorption wavelengths (λabs), peak emission wavelengths (λem), maximum molar absorption coefficients (ε), Stokes shifts (Δλ), and fluorescence quantum yields (φ). UV–vis absorption and fluorescence emission spectra of BN-1. (a, b) in various solvents. (c, d) at different pH (ranging from pH 2 to pH 12). (e) UV–vis absorption and (f) fluorescence emission spectra of BN-1 in the binary mixture of glycerol and MeOH, with different volume ratios of glycerol. (g) Quantum yield of BN-1 at different viscosity. (h) FMOs of BN-1 in ethanol. (i) Schematic illustration of the photophysical mechanisms and the calculated excitation/de-excitation energy (as well as oscillator strength f) of representative states of BN-1 in ethanol (the inset shows the chemical structure of BN-1. Note that energy values are not drawn to scale for clarity). (j) Optimized molecular structures of BN-1 in representative states, as well as the corresponding electron and hole distributions in ethanol (VES and AES denote vertically excited state and adiabatic excited state, respectively). The inset highlights the conformational folding during the transition to the ET state. Download figure Download PowerPoint Table 1 | Photophysical Data of BN-1 in Various Solventsa Solvent λabs (nm) λem (nm) ε (M−1cm−1) Δλ (nm) Φ Toluene 420 492 15,645 72 0.694 Diethyl ether 420 494 16,313 74 0.765 DCM 428 504 16,908 76 0.715 EA 426 506 16,439 80 0.396 CHCl3 430 500 17,186 70 0.762 Dioxane 422 495 15,909 73 0.758 Acetone 430 509 17,018 79 0.026 ACN 430 520 17,595 90 0.018 DMF 437 521 16,435 84 0.014 EtOH 437 519 17,575 82 0.029 MeOH 437 516 17,715 79 0.064 DMSO 444 531 17,135 87 0.013 aλabs, peak UV–vis absorption wavelength; λem, peak fluorescence wavelength; ξ, molar extinction coefficient at λabs; Δλ, the Stokes shift; and φ, fluorescence quantum yields. Abbreviations: DCM, dichloromethane; EA, ethyl acetate; ACN, acetonitrile; DMF, dimethylformamide; DMSO, dimethyl sulfoxide. These experimental results support the PET mechanism: the tertiary amino receptor, connected to the naphthalimide fluorophore via a hydrocarbon chain, injects an electron into the fluorophore upon photoexcitation in polar solvents, activating PET, quenching fluorescence, and reducing quantum yield. The resulting ET state possesses a large dipole moment and is significantly stabilized in polar solvents due to strong dipole–dipole and induced dipole interactions. However, this stabilizing force diminishes in nonpolar solvents, weakening or even preventing the formation of the ET state.5 Additionally, protonation of the tertiary amino group in acidic environments substantially reduces its electron-donating power, inhibiting PET and restoring fluorescence intensities. This polarity and pH dependence are characteristic of many PET probes. Failure of the classical computational protocols in predicting PET Unfortunately, the classical FMO model was unable to predict PET occurrence in BN-1. We performed a conformation search using available crystal structures. Our findings showed that the most stable conformation, as determined by the conformational search, closely matched BN-1's crystal structure,32 except for a mirror symmetry at the alkyl moiety ( Supporting Information Figure S3). Surprisingly, our results did not reveal any FMOs from the receptor situated between the HOMO and LUMO of the naphthalene scaffold (Figure 2h). Instead, we found that the receptor's FMO (from the dimethylamino group's lone pair) was positioned below the naphthalene moiety's HOMO, with an energy difference of 0.81 eV in ethanol. These FMO energy levels fail to predict PET and contradict experimental data. Directly modeling the excited states based on the most stable ground-state conformation failed to capture the PET process of BN-1 in ethanol as well. During the vertical excitation to the Franck-Condon states, we observed that the S1 state corresponds to an LE state, which exhibits a high oscillator strength of 0.49 (Figure 2i,j). Additionally, we were unable to identify the ET state which involves electron transfer from the amino receptor to the naphthalimide fluorophore (within the first five excited states). This absence signifies the inhibition of the PET process in this specific molecular conformation, henceforth referred to as the unfolded (UF) conformation due to its extended structure. Towards conformational folding: An activator of PET To reconcile computational and experimental findings, we focused on potential conformational changes in the excited states. A molecule's PESs may possess many degrees of freedom, leading to multiple local minima. Comprehensively modeling all these local minima, rather than focusing on just one, could provide a complete understanding of the excited-state dynamics and the behavior of BN-1 ( Supporting Information Figure S4). Among the multiple representative conformations under consideration, we hypothesized that the most stable S1 conformation might exhibit a fully folded (FF) structure, aimed at minimizing the distance between the receptor and the fluorophore. This proximity was anticipated to greatly amplify the electrostatic interactions between the positively charged amino receptor and the negatively charged naphthalimide fluorophores, thereby stabilizing the ET state and activating the PET process. Contrary to our expectations, the ET state based on a semifolded (SF) conformation emerged as the most stable on the S1 PES (Figure 2i,j). We evaluated UF, SF, and FF conformations for BN-1 ( Supporting Information Figure S13). Notably, the energy level of the ET state in the FF conformation was found to be higher by 0.49 eV compared to the SF conformation. This significant difference underscores a complex interplay among various forces that govern the molecule's folding dynamics: fully folding minimizes the distance between the electron donor and acceptor, amplifies electrostatic attractions, and increases steric repulsion. Due to this trade-off, the SF conformation achieves greater stability compared to the FF conformation. The complete PES, illustrating UF, SF, and FF conformations, is depicted in Supporting Information Figure S13. In good agreement with experimental observation, the energy level of the optimized ET based on the SF conformation was significantly lower than that of LE, implying state-crossing from the LE state to the ET state upon photoexcitation. Furthermore, the oscillator strength of ET was nearly zero, signifying that this state was nonemissive (Figure 2i). These findings effectively rationalized that PET induces fluorescence quenching of BN-1 in ethanol, resulting in excellent agreement with experimental data. Notably, the geometry of the LE state of BN-1 features an extended structure (UF), while it starts to fold after relaxation to the ET state (SF). This geometrical change enables strong electrostatic interactions between the receptor and the fluorophore, stabilizes the ET state, and facilitates a fast PET process, resulting in fluorescence quenching of BN-1 in ethanol. It is thus clear that conformation folding could activate PET in fluorophore-spacer-receptor systems. Similar calculations accounting for conformational folding also yielded good agreements with experimental data in toluene ( Supporting Information Figure S5). In toluene, the energy levels of ET are higher than that of LE, suggesting the absence of state-crossing to the dark ET states (or lack of PET). Consequently, BN-1 is highly emissive due to intense fluorescence from the LE state. The solvent environment significantly impacts the electron transfer dynamics, primarily through its influence on the relative stabilization of charged intermediates. In polar solvents, such as ethanol, the stabilization of charged species enhances PET efficiency by lowering the activation energy barrier. Conversely, nonpolar solvents like toluene provide less stabilization, reducing PET efficiency. These results in both ethanol and toluene collectively demonstrate that structural folding in BN-1 is a crucial step to activate PET in polar solvents. To verify our computational prediction regarding conformational folding in polar solvents, we measured the viscosity response of BN-1 (Figure 2e,f,g). As the viscosity of the solvent (the binary mixture of methanol and glycerol) increases, the UV–vis absorption spectra of BN-1 remain nearly unchanged. However, we observed a significant fluorescence enhancement of BN-1 by 11.6 times (Figure 2g and Supporting Information Table S2). This observation is consistent with the conformational folding during the PET process: increasing viscosity inhibits conformational changes (i.e., folding), and thus inhibits PET, leading to an increase in fluorescence intensity. Conformational folding during the PET processes of other fluorophores We also extended the success of our multiconformation calculations to other fluorophores (Figure 3a and Supporting Information Figures S6–S24). In these compounds with the fluorophore-spacer-receptor configuration, the classical FMO model based on the most stable ground-state conformations failed to predict the occurrence of PET (Figure 3b,e and Supporting Information Figures S14 and S15). However, by calculating the most stable ET states with folded geometries, we successfully justified the PET mechanism in these compounds ( Supporting Information Figures S16–S24). Figure 3 | (a) Chemical structures of the compounds investigated in this work. FMOs of (b) Acri-1 and (e) Rhod-1 in ethanol. The PES and the calculated excitation/de-excitation energy of (c) Acri-1 and (f) Rhod-1 in ethanol (Note that energy values are not drawn to scale for clarity and all values are in eV) and optimized molecular structures of (d) Acri-1 and (g) Rhod-1 in the ground and excited states, as well as the corresponding electron and hole distributions in ethanol (VES and AES denote vertically excited state and adiabatic excited state, respectively). The inset highlights the conformational folding during the transition to the ET state. Download figure Download PowerPoint For example, our calculations revealed that for Acri-1, the ET state based on the SF structure is more stable than the LE state based on the UF structure, with a difference of 1.42 eV in ethanol and 0.79 eV in toluene, respectively (Figure 3b–d and Supporting Information Figures S16 and S17). Similarly, in considering the conformational folding in Rhod-1, although the classic FMO model failed to predict PET, our computational modeling demonstrates that the ET state with an energy of 2.09 eV is more stable than the LE state with an energy of 2.97 eV (Figure 3e–g and Supporting Information Figures S18 and S19). This suggests the feasibility of PET, aligning well with experimental observations. Interestingly, our calculations showed that the ET state of NBD-1 is more stable than the LE state in both ethanol and toluene ( Supporting Information Figures S20 and S21). In other words, PET could substantially occur in both polar and nonpolar solvents for NBD-1, owing to its strong PET tendency. To validate our theoretical predictions, we synthesized this compound as well as the reference compound without the receptor (NBD-2) and characterized their photophysical properties in various solvents ( Supporting Information Figure S1b,c,e,f and Table S3). In comparison to the bright emissions of NBD-2 in aprotic solvents, NBD-1 exhibits extremely low quantum yields in all solvents, corroborating the accuracy of our theoretical predictions of the PET mechanism. It is worth highlighting that our calculations resolved a nearly decade-long controversy over the working mechanism of BDP-1 ( Supporting Information Figure S24).21,22 BDP-1 is a pH-controlled activatable viscosity probe. While Xiao et al. proposed that disabling PET acts as the activation mechanism, other teams challenged this view using FMO calculations.21,22 However, the conclusion drawn from the classical FMO calculations may not be reliable, as this model does not fully consider excited state conformational changes (along with many other factors, i.e., solvent effects and exciton binding energy). By taking these factors into account, our calculations demonstrated that PET could efficiently take place and quench the fluorescence of BDP-1, corroborating Xiao's proposition. Notably, we discovered significant conformational changes during the transition from the LE to the ET state. Without accounting for such conformational changes, it would be impossible to identify a stable ET state ( Supporting Information Figure S24). A revised workflow for predicting PET The classical FMO models do not account for the structural folding in the excited state, which is crucial for PET probes designed with a fluorophore-spacer-receptor configuration. To address this shortcoming, we propose a modified computational protocol that incorporates structural changes, thereby providing a more comprehensive understanding of the PES. This approach enables a deeper insight into the behavior of these compounds in the excited state. We recommend that chemists adopt the following steps when modeling such compounds: • Optimize multiple conformations in the ground state, including both extended and folded conformations. • Conduct excited-state calculations using multiple representative conformations (i.e., UF, SF, and FF) to evaluate PET tendencies. Notably, capturing multiple conformations (including those meta-stable ones) in the ground state could aid molecular modelers in identifying potential excited-state conformations and optimizing various ET states across representative conformations. By adopting this protocol, researchers can more accurately predict and understand PET processes in compounds of interest, thereby enhancing the study of photophysical mechanisms. While our study provides significant insights into the PET process, several limitations must be acknowledged. The computational models used, although robust, may not capture all the nuances of real-world molecular dynamics. Additionally, experimental conditions such as temperature and pH can impact PET efficiency, suggesting the need for further studies under diverse conditions. Addressing these limitations in future research will enhance the generalizability and applicability of our findings. PET-induced dramatic geometry changes hold potential for various applications, such as viscosity sensing.33 If folded structures become stable or available even in the ground state for specific molecules, the resulting low-lying dark states (due to PET) could give rise to intriguing photophysical and photochemical phenomena.34–36 Engineering the linker fragment to facilitate structural folding may further enhance the performance of PET-based materials.7 Bioimaging utilities of BN-1 Utilizing BN-1, we developed fluorogenic probes targeting specific proteins by synthesizing Halo-BN-1, which incorporates a ligand structure reactive with one popular self-labeling tag, the HaloTag.37 This synthesis enabled the successful imaging of HaloTag proteins in the cell nucleus using Halo-BN-1. In a polar medium (i.e., cytoplasm), the conformational folding in Halo-BN-1 could activate PET, quenching fluorescence. When this probe binds to a nonpolar protein, the ET state becomes unstable; no conformational folding associated with PET occurs, thus recovering bright fluorescence. Consequently, this environmental sensitivity endows Halo-BN-1 with an excellent signal-to-noise ratio, approximately 12 times higher than the background ( Supporting Information Figures S25–S27). This high ratio facilitated dynamic imaging of the HaloTag protein within the cell nucleus. We also induced intracellular acidification and alkalinization to observe their effects on nuclear fluorescence. Acidification, simulated by adding 50 mM CH3COONa to living cell