Photoswitches
Rational Design and Exploration of Imine-based Photoswitches
Molecular photoswitches, by definition, undergo changes in their geometric structure and chemical properties when excited by ultraviolet and visible light. A well-known biological example of a photoswitch, Rhodopsin, is a critical photosensor in vertebrate vision. Rhodopsin is a 11-cis-retinal and opsin protein complex found in the rods cell of the retina. When excited by particular colors of light, 11-cis-retinal undegoes cis (Z) to trans (E) photoisomerization, which induces a conformational change in the opsin protein and initiates a signaling cascade that ultimately leads to light perception in the brain. Beyond their role in vision, molecular photoswitches have found many applications in fields such as fluorescence bioimaging, solar energy storage, and optical data storage. Among them, azobenzenes (R-N=N-R, azos) are the most studied because of their vibrant colors, synthetic versatility, and well-understood photoisomerization and thermal reversion mechanisms. Azos are used in a wide range of applications, from traditional food and clothing dyes to more modern technologies such as photopharmaceuticals (e.g. vision restoration, cancer treatment, antibacterial resistance) and molecular solar thermal fuels.
Beyond azos, atypical photoswitches including heteroaryl imines (Figure 1c) and iminothioindoxyls (Figure 1d) are severly understudied but have the potential to expand the capacity of photopharmacology, in particular, by adding new routes for ‘x-extension’ or ‘x-logization’, where ‘x’ is a switch of interest. Imine-based photoswitches are synthetically accessible because they can be synthesized through a straightforward condensation reaction between a primary amine and an aldehyde or ketone. Iminothioindoxyls can be synthesized in a similar fashion through a condensation reaction with thioindoxyl and a nitrosobenzene. The amine and carbonyl precursors can be largely modified to selectively tune the properties of the final imine products. An interesting properties of these systems is that both the E and Z isomers adopt a non-planar (more globular) geometric structure, at least for simple derivatives. This differs from azobenzenes whose E-isomer is usually planar and whose Z-isomer is globular. Through the addition of intramolecular hydrogen bonding moieties, it should be possible to switch between globular and planar geometries through computationally guided synthetic tuning. The Knepp Lab will use electronic structure calculations to design and rationalize the ground- and excited-state properties of atypical imine- and iminothioindoxyl-based photoswitches. Molecules of interest will be synthesized and spectroscopically charaterized.
Computer-aided Photoswitchable Drug Design
Photopharmacology introduces a powerful mechanism to achieve spatial and temporal control over traditional pharmaceuticals through light-induced drug activation or deactivation through the addition of a photoswitchable moiety, typically azobenzenes. Common strategies to accomplish this include ‘azo-extension’, where the photoswitch is appended to existing drugs, and ‘azo-logization’, where fractions of drug molecules with electronic and geometric structure similar to azobenzene are replaced with a whole or a fraction of a azobenzene (or heteroaryl azobenzene) molecule. Even though it is technically possible to design and optimize photoswitchable ligands from scratch, this approach poses a significant challenge and requires substantial financial investment. In fact, pharmaceutical companies already allocate ridiculous amounts of financial resources to identify hits and leads with various computational approaches. As a result, most successful applications of photopharmacology focus on using azo-extension and azo-logization rather than designing from scratch. In general, both methods provide a practical way to introduce an on/off switch to existing drugs or drug candidates.
Since this field emerged relatively recently (2004), most research has been conducted in vitro or in vivo with minimal to no computational guidance. Although rare, some studies have utilized molecular docking and molecular dynamics simulations to guide and optimize photopharmaceuticals but most do not take advantage of the full potential of in silico design and optimization. For example, I am confident that free energy perturbation (FEP) methods would elevate rational design in photopharmacology to a new level. Although standard molecular dynamic simulations can be a powerful tool for observing ligand-protein interactions over time, FEP provides a strategy for hit-to-lead optimization in photopharmacology. FEP predicts changes in binding affinity between different ligands, making it an ideal tool for fine-tuning ligand affinity (Figure 2). Beyond optimizing the binding affinities of existing photopharmaceuticals in the literature, the Knepp Lab will use FEP to maximize the binding affinity difference between the E and Z isomers. This aspect of the project is particularly exciting because it has the potential to optimize on/off switches so that one isomer is bioactive and the other is bioinactive. If one focuses solely on enhancing the binding affinity of the bioactive isomer without considering the inactive one, there is a risk that the switch could end up being largely bioactive in both isomeric states, effectively leading to an on/on photoswitch rather than an on/off photoswitch. In practice, we will perform relative binding free energy (RBFE) calculations using a range of classical force fields and by using the CHARMM-GUI interface to ensure accurate, reproducible, and efficient workflows. CHARMM-GUI greatly simplifies the setup and execution of these jobs, making them more accessible to undergraduate students. $\Delta\Delta G$ values will be estimated via alchemical transformations, capturing the free energy changes associated with ligand substitution both in solution (∆GA) and within the protein environment (∆GB ) (Figure 2). Should our results prove unreliable or insufficient using classical force-field potentials, we will switch to QM/MM methods to improve the accuracy of the ∆∆G calculations.
Triplet State Influence on Photoisomerization and Thermal Reversion
The photoisomerization \((E \to Z)\) and thermal reversion \((E \to Z)\) mechanisms in molecular photoswitches have been studied for nearly two centuries. It is widely accepted that photoisomerization of azobenzene-based photoswitches involves a N=N dihedral rotation in the excited \(S_1\) singlet state, which, after a few picoseconds, results in the formation of metastable Z-isomer. However, the mechanism of thermal reversion has been and is currently under debate in the literature. Depending on the molecular substituents, azobenzene-based photoswitches can thermally revert to the E-isomer through an in-plane phenyl inversion, a N=N dihedral rotation, or some combination of the rotation and inversion (e.g., rotation-assisted inversion).
Recently, with the support of strong theoretical and experimental evidence, it has been suggested that the lowest-lying triplet excited state (\(T_1\)) is actively involved in \(Z\to E\) thermal reversion of azobenzene and its derivatives. The importance and consideration of the transient triplet rotational mechanism is often overshadowed by a simplified, monoexponential, kinetic model, at least in theoretical calculations. This often occurs because calculating proper rate constants involving the triplet manifold requires non-adiabatic transition state theory
\[k_{\mathrm{NA-TST}}=\frac{k_BT}{h}\gamma \exp{\left(-\Delta G^\ddagger/RT\right)}.\]
to properly describe the minimum energy crossing points (MECPs) and the spin-orbital coupling (SOC) between the singlet and triplet surfaces. Calculating the transmission coefficient (\(\gamma\)) is, to say the least, complicated, as it involves several terms that require additional and expensive calculations. Using the Wentzel-Kramers-Brillouin (WKB) approximation of NA-TST, \(\gamma\) approximately describes the crossing between the singlet and triplet manifolds through SOC where
\[\gamma = \frac{\pi^{3/2}\beta}{2\sqrt{\epsilon_0k_BT}}\left[ 1+\frac{1}{2}\exp{\left( \frac{1}{12\beta^2}\frac{1}{(\epsilon_0k_BT)^3}\right)}\right],\]
\[\beta = \frac{4V_{SO}^{3/2}}{\hbar}\left( \frac{\mu}{\bar{F}\Delta F}\right)^{1/2},\ \ \epsilon_0 = \frac{\Delta F}{2\bar{F}V_{SO}},\ \ \Delta F = | \mathbf{F}_1-\mathbf{F}_2|,\ \ \mathrm{and}\ \bar{F} = (\mathbf{F}_1\mathbf{F}_2)^{1/2}.\]
Here, \(V_{SO}\) is the SOC between the singlet and triplet states at the MECPs, \(\mu\) is the reduced mass orthogonal to the crossing seam, and \(\mathbf{F_i}\) is the gradient on a particular surface \(\mathbf{i}\) at the MECPs. In the limit where \(\gamma=1\), then \(k_{\mathrm{NA-TST}}=k_{\mathrm{TST}}\). Unlike the standard TST, this mechanism involves interconnected elementary steps, so the rate constant is calculated as \(k=\left( \sum_i1/k_i \right)^{-1}\). Due to the complexity of these equations, the requirement of additional calculations, and the need for high-level wavefunction and multireference methods to adequately calculate such quantities, researchers often shy away from NA-TST in favor of TST combined with TD-DFT and SF-TD-DFT.
Beyond azobenzenes, imine- and iminothioindoxyl-based photoswitches could transiently populate the \(T_1\) triplet state by thermal reversion through a rotation around the imine double bond. Using molecular model systems that involve a double bond rotation (Figure X), I have found, with high-level electronic structure theory, that the \(T_1\) triplet state has a lower energy than the \(S_0\) singlet state at the rotational transition state between the \(E\) and \(Z\) isomers for systems that have a N=N and C=C double bond. Interestingly, for the C=N model system, the \(T_1\) triplet state is higher energy than the \(S_0\) singlet state at the rotational transition state.
Using TD-DFT, CCSD(T), and CASPT2 calculations, the Knepp Lab will explore the effect of the \(T_1\) triplet manifold on the photoisomerization and thermal reversion dynamics of the atypical imine- and iminothioindoxyl-based photoswitches. If the triplet state is involved in the thermal reversion process, NA-TST in the WKB approximation will be used to predict the rate constants. These will be compared to experimental rate constants obtained in house or through collaborations. If the triplet state is not involved in the thermal reversion, we will try to rationalize why with molecular orbital theory and energetic arguments.