Fast Screening of Minimum Energy Crossing Points with Semiempirical Tight-Binding Methods

The investigation of photochemical processes is a highly active field in computational chemistry. One research direction is the automated exploration and identification of minimum energy conical intersection (MECI) geometries. However, due to the immense technical effort required to calculate nonadiabatic potential energy landscapes, the routine application of such computational protocols is severely limited. In this study, we will discuss the prospect of combining adiabatic potential energy surfaces from semiempirical quantum mechanical calculations with specialized confinement potential and metadynamics simulations to identify S₀/T₁ minimum energy crossing point (MECP) geometries. It is shown that MECPs calculated at the GFN2-xTB level can provide suitable approximations to high-level S₀/S₁ab initio conical intersection geometries at a fraction of the computational cost. Reference MECIs of benzene are studied to illustrate the basic concept. An example application of the presented protocol is demonstrated for a set of photoswitch molecules.

Exploiting Ligand Additivity for Transferable Machine Learning of Multireference Character across Known Transition Metal Complex Ligands

Accurate virtual high-throughput screening (VHTS) of transition metal complexes (TMCs) remains challenging due to the possibility of high multireference (MR) character that complicates property evaluation. We compute MR diagnostics for over 5,000 ligands present in previously synthesized octahedral mononuclear transition metal complexes in the Cambridge Structural Database (CSD). To accomplish this task, we introduce an iterative approach for consistent ligand charge assignment for ligands in the CSD. Across this set, we observe that the MR character correlates linearly with the inverse value of the averaged bond order over all bonds in the molecule. We then demonstrate that ligand additivity of the MR character holds in TMCs, which suggests that the TMC MR character can be inferred from the sum of the MR character of the ligands. Encouraged by this observation, we leverage ligand additivity and develop a ligand-derived machine learning representation to train neural networks to predict the MR character of TMCs from properties of the constituent ligands. This approach yields models with excellent performance and superior transferability to unseen ligand chemistry and compositions.

Steric and Electronic Origins of Fluorescence in GFP and GFP-like Proteins

Fluorescent proteins have become routine tools for biological imaging. However, their nanosecond lifetimes on the excited state present computational hurdles to a full understanding of these photoactive proteins. In this work, we simulate approximately 0.5 nanoseconds of ab initio molecular dynamics to elucidate steric and electronic features responsible for fluorescent protein behavior. Using green fluorescent protein (GFP) and Dronpa2─widely used fluorescent proteins with contrasting functionality─as case studies, we leverage previous findings in the gas phase and solution to explore the deactivation mechanisms available to these proteins. Starting with ground-state analyses, we identify steric (the distribution of empty pockets near the chromophore) and electronic (electric fields exerted on chromophore moieties) factors that offer potential avenues for rational design. Picosecond timescale simulations on the excited state reveal that the chromophore can access twisted structures in Dronpa2, while the chromophore is largely confined to planarity in GFP. We couple ab initio multiple spawning (AIMS) and enhanced sampling simulations to discover and characterize conical intersection seams that facilitate internal conversion, which is a rare event in both systems. Our AIMS simulations correctly capture the relative fluorescence profiles of GFP and Dronpa2 within the first few picoseconds, and we attribute the diminished fluorescence intensity of Dronpa2, relative to GFP, to flexible chromophore intermediates on the excited state. Furthermore, we predict that twisted chromophore intermediates produce red-shifted intensities in the Dronpa2 fluorescence spectrum. If confirmed experimentally, this spectroscopic signature would provide valuable insights when screening and developing novel fluorescent proteins.

Multiscale simulation unravels the light-regulated reversible inhibition of dihydrofolate reductase by phototrexate

Molecular photoswitches are widely used in photopharmacology, where the biomolecular functions are photo-controlled reversibly with high spatiotemporal precision. Despite the success of this field, it remains elusive how the protein environment modulates the photochemical properties of photoswitches. Understanding this fundamental question is critical for designing more effective light-regulated drugs with mitigated side effects. In our recent work, we employed first-principles non-adiabatic dynamics simulations to probe the effects of protein on the trans to cis photoisomerization of phototrexate (PTX), a photochromic analog of the anticancer therapeutic methotrexate that inhibits the target enzyme dihydrofolate reductase (DHFR). Building upon this study, in this work, we employ multiscale simulations to unravel the full photocycle underlying the light-regulated reversible inhibition of DHFR by PTX, which remains elusive until now. First-principles non-adiabatic dynamics simulations reveal that the cis to trans photoisomerization quantum yield is hindered in the protein due to backward isomerization on the ground-state following non-adiabatic transition, which arises from the favorable binding of the cis isomer with the protein. However, free energy simulations indicate that cis to trans photoisomerization significantly decreases the binding affinity of the PTX. Thus, the cis to trans photoisomerization most likely precedes the ligand unbinding from the protein. We propose the most probable photocycle of the PTX-DHFR system. Our comprehensive simulations highlight the trade-offs among the binding affinity, photoisomerization quantum yield, and the thermal stability of the ligand's different isomeric forms. As such, our work reveals new design principles of light-regulated drugs in photopharmacology.

Nonadiabatic Molecular Dynamics with Extended Density Functional Tight-Binding: Application to Nanocrystals and Periodic Solids

In this work, we report a new methodology for nonadiabatic molecular dynamics calculations within the extended tight-binding (xTB) framework. We demonstrate the applicability of the developed approach to finite and periodic systems with thousands of atoms by modeling "hot" electron relaxation dynamics in silicon nanocrystals and electron-hole recombination in both a graphitic carbon nitride monolayer and a titanium-based metal-organic framework (MOF). This work reports the nonadiabatic dynamic simulations in the largest Si nanocrystals studied so far by the xTB framework, with diameters up to 3.5 nm. For silicon nanocrystals, we find a non-monotonic dependence of "hot" electron relaxation rates on the nanocrystal size, in agreement with available experimental reports. We rationalize this relationship by a combination of decreasing nonadiabatic couplings related to system size and the increase of available coherent transfer pathways in systems with higher densities of states. We emphasize the importance of proper treatment of coherences for obtaining such non-monotonic dependences. We characterize the electron-hole recombination dynamics in the graphitic carbon nitride monolayer and the Ti-containing MOF. We demonstrate the importance of spin-adaptation and proper sampling of surface hopping trajectories in modeling such processes. We also assess several trajectory surface hopping schemes and highlight their distinct qualitative behavior in modeling the excited-state dynamics in superexchange-like models depending on how they handle coherences between nearly parallel states.

Machine Learning Models Predict Calculation Outcomes with the Transferability Necessary for Computational Catalysis

Virtual high-throughput screening (VHTS) and machine learning (ML) have greatly accelerated the design of single-site transition-metal catalysts. VHTS of catalysts, however, is often accompanied with a high calculation failure rate and wasted computational resources due to the difficulty of simultaneously converging all mechanistically relevant reactive intermediates to expected geometries and electronic states. We demonstrate a dynamic classifier approach, i.e., a convolutional neural network that monitors geometry optimizations on the fly, and exploit its good performance and transferability in identifying geometry optimization failures for catalyst design. We show that the dynamic classifier performs well on all reactive intermediates in the representative catalytic cycle of the radical rebound mechanism for the conversion of methane to methanol despite being trained on only one reactive intermediate. The dynamic classifier also generalizes to chemically distinct intermediates and metal centers absent from the training data without loss of accuracy or model confidence. We rationalize this superior model transferability as arising from the use of electronic structure and geometric information generated on-the-fly from density functional theory calculations and the convolutional layer in the dynamic classifier. When used in combination with uncertainty quantification, the dynamic classifier saves more than half of the computational resources that would have been wasted on unsuccessful calculations for all reactive intermediates being considered.

Photochemical Deracemization of Chiral Alkenes via Triplet Energy Transfer

A visible-light-mediated, enantioselective approach to axially chiral alkenes is described. Starting from a racemic mixture, a major alkene enantiomer is formed due to selective triplet energy transfer from a catalytically active chiral sensitizer. A catalyst loading of 2 mol % was sufficient to guarantee consistently high enantioselectivities and yields (16 examples, 51%-quant., 81-96% ee). NMR studies and DFT computations revealed that triplet energy transfer is more rapid within the substrate-catalyst complex of the minor alkene enantiomer. Since this enantiomer is continuously racemized, the major enantiomer is enriched in the photostationary state.

InteraChem: Exploring Excited States in Virtual Reality with Ab Initio Interactive Molecular Dynamics

InteraChem is an ab initio interactive molecular dynamics (AI-IMD) visualizer that leverages recent advances in virtual reality hardware and software, as well as the graphical processing unit (GPU)-accelerated TeraChem electronic structure package, in order to render quantum chemistry in real time. We introduce the exploration of electronically excited states via AI-IMD using the floating occupation molecular orbital-complete active space configuration interaction method. The optimization tools in InteraChem enable identification of excited state minima as well as minimum energy conical intersections for further characterization of excited state chemistry in small- to medium-sized systems. We demonstrate that finite-temperature Hartree-Fock theory is an efficient method to perform ground state AI-IMD. InteraChem allows users to track electronic properties such as molecular orbitals and bond order in real time, resulting in an interactive visualization tool that aids in the interpretation of excited state chemistry data and makes quantum chemistry more accessible for both research and educational purposes.

Effects of Enzyme-Ligand Interactions on the Photoisomerization of a Light-Regulated Chemotherapeutic Drug

Molecular photoswitches permit using light to control protein activity with high spatiotemporal resolutions, thereby alleviating the side effects of conventional chemotherapy. However, due to the challenges in probing ultrafast photoisomerization reactions in biological environments, it remains elusive how the protein influences the photochemistry of the photoswitches, which hampers the rational design of light-regulated therapeutics. To overcome this challenge, we employed first-principles nonadiabatic dynamics simulations to characterize the photodynamics of the phototrexate (PTX), a recently developed photoswitchable anticancer chemotherapeutic that reversibly inhibits its target enzyme dihydrofolate reductase (DHFR). Our simulations show that the protein environment impedes the trans to cis photoisomerization of the PTX. The confinement in the ligand-binding cavity slows down the isomerization kinetics and quantum yield of the photoswitch by reshaping its conical intersection, increasing its excited-state free-energy barrier and quenching its local density fluctuations. Also, the protein environment results in a suboptimal binding mode of the photoproduct that needs to undergo large structural rearrangement to effectively inhibit the enzyme. Therefore, we predict that the PTX's trans → cis photoisomerization in solution precedes its binding with the protein, despite the favorable binding energy of the trans isomer. Our findings highlight the importance of the protein environment on the photochemical reactions of the molecular photoswitches. As such, our work represents an important step toward the rational design of light-regulated drugs in photopharmacology.

Drug Repurposing of the Unithiol: Inhibition of Metallo-β-Lactamases for the Treatment of Carbapenem-Resistant Gram-Negative Bacterial Infections

The increasing antibiotic resistance is a clinical problem worldwide. Numerous Gram-negative bacteria have already become resistant to the most widely used class of antibacterial drugs, β-lactams. One of the main mechanisms is inactivation of β-lactam antibiotics by bacterial β-lactamases. Appearance and spread of these enzymes represent a continuous challenge for the clinical treatment of infections and for the design of new antibiotics and inhibitors. Drug repurposing is a prospective approach for finding new targets for drugs already approved for use. We describe here the inhibitory potency of known detoxifying antidote 2,3-dimercaptopropane-1-sulfonate (unithiol) against metallo-β-lactamases. Unithiol acts as a competitive inhibitor of meropenem hydrolysis by recombinant metallo-β-lactamase NDM-1 with the K_I of 16.7 µM. It is an order of magnitude lower than the K_I for l-captopril, the inhibitor of angiotensin-converting enzyme approved as a drug for the treatment of hypertension. Phenotypic methods demonstrate that the unithiol inhibits natural metallo-β-lactamases NDM-1 and VIM-2 produced by carbapenem-resistant K. pneumoniae and P. aeruginosa bacterial strains. The 3D full atom structures of unithiol complexes with NDM-1 and VIM-2 are obtained using QM/MM modeling. The thiol group is located between zinc cations of the active site occupying the same place as the catalytic hydroxide anion in the enzyme–substrate complex. The sulfate group forms both a coordination bond with a zinc cation and hydrogen bonds with the positively charged residue, lysine or arginine, responsible for proper orientation of antibiotics upon binding to the active site prior to hydrolysis. Thus, we demonstrate both experimentally and theoretically that the unithiol is a prospective competitive inhibitor of metallo-β-lactamases and it can be utilized in complex therapy together with the known β-lactam antibiotics.

Calculating Photoabsorption Cross-Sections for Atmospheric Volatile Organic Compounds

Characterizing the photochemical reactivity of transient volatile organic compounds (VOCs) in our atmosphere begins with a proper understanding of their abilities to absorb sunlight. Unfortunately, the photoabsorption cross-sections for a large number of transient VOCs remain unavailable experimentally due to their short lifetime or high reactivity. While structure-activity relationships (SARs) have been successfully employed to estimate the unknown photoabsorption cross-sections of VOCs, computational photochemistry offers another promising strategy to predict not only the vertical electronic transitions of a given molecule but also the width and shape of the bands forming its absorption spectrum. In this work, we focus on the use of the nuclear ensemble approach (NEA) to determine the photoabsorption cross-section of four exemplary VOCs, namely, acrolein, methylhydroperoxide, 2-hydroperoxy-propanal, and (microsolvated) pyruvic acid. More specifically, we analyze the influence that different strategies for sampling the ground-state nuclear density-Wigner sampling and ab initio molecular dynamics with a quantum thermostat-can have on the simulated absorption spectra. We highlight the potential shortcomings of using uncoupled harmonic modes within Wigner sampling of nuclear density to describe flexible or microsolvated VOCs and some limitations of SARs for multichromophoric VOCs. Our results suggest that the NEA could constitute a powerful tool for the atmospheric community to predict the photoabsorption cross-section for transient VOCs.

Fermi.jl: A Modern Design for Quantum Chemistry

Approximating molecular wave functions involves heavy numerical effort; therefore, codes for such tasks are written completely or partially in efficient languages such as C, C++, and Fortran. While these tools are dominant throughout quantum chemistry packages, the efficient development of new methods is often hindered by the complexity associated with code development. In order to ameliorate this scenario, some software packages take a dual approach where a simpler, higher-level language, such as Python, substitutes the traditional ones wherever performance is not critical. Julia is a novel, dynamically typed, programming language that aims to solve this two-language problem. It gained attention because of its modern and intuitive design, while still being highly optimized to compete with "low-level" languages. Recently, some chemistry-related projects have emerged exploring the capabilities of Julia. Herein, we introduce the quantum chemistry package Fermi.jl, which contains the first implementations of post-Hartree-Fock methods written in Julia. Its design makes use of many Julia core features, including multiple dispatch, metaprogramming, and interactive usage. Fermi.jl is a modular package, where new methods and implementations can be easily added to the existing code. Furthermore, it is designed to maximize code reusability by relying on general functions with specialized methods for particular cases. The feasibility of the project is explored through evaluating the performance of popular ab initio methods. It is our hope that this project motivates the usage of Julia within the community and brings new contributions into Fermi.jl.

The role of cysteine residues in the allosteric modulation of the chromophore phototransformations of biphotochromic fluorescent protein SAASoti

Biphotochromic fluorescent protein SAASoti contains five cysteine residues in its sequence and a V127T point mutation transforms it to the monomeric form, mSAASoti. These cysteine residues are located far from the chromophore and might control its properties only allosterically. The influence of individual, double and triple cysteine substitutions of mSAASoti on fluorescent parameters and phototransformation reactions (irreversible green-to-red photoconversion and reversible photoswitching) is studied. A set of mSAASoti mutant forms (C21N, C117S, C71V, C105V, C175A, C21N/C71V, C21N/C175A, C21N/C71G/C175A) is obtained by site-directed mutagenesis. We demonstrate that the C21N variant exists in a monomeric form up to high concentrations, the C71V substitution accelerates photoconversion to the red form and the C105V variant has the maximum photoswitching rate. All C175A-containing variants demonstrate different photoswitching kinetics and decreased photostability during subsequent switching cycles compared with other considered systems. Classical molecular dynamic simulations reveal that the F177 side chain located in the vicinity of the chromophore is considerably more flexible in the mSAASoti compared with its C175A variant. This might be the explanation of the experimentally observed slowdown the thermal relaxation rate, i.e., trans-cis isomerization of the chromophore in mSAASoti upon C175A substitution.

Inq, a Modern GPU-Accelerated Computational Framework for (Time-Dependent) Density Functional Theory

We present inq, a new implementation of density functional theory (DFT) and time-dependent DFT (TDDFT) written from scratch to work on graphic processing units (GPUs). Besides GPU support, inq makes use of modern code design features and takes advantage of newly available hardware. By designing the code around algorithms, rather than against specific implementations and numerical libraries, we aim to provide a concise and modular code. The result is a fairly complete DFT/TDDFT implementation in roughly 12 000 lines of open-source C++ code representing a modular platform for community-driven application development on emerging high-performance computing architectures.

LSSmScarlet, dCyRFP2s, dCyOFP2s and CRISPRed2s, Genetically Encoded Red Fluorescent Proteins with a Large Stokes Shift

Genetically encoded red fluorescent proteins with a large Stokes shift (LSSRFPs) can be efficiently co-excited with common green FPs both under single- and two-photon microscopy, thus enabling dual-color imaging using a single laser. Recent progress in protein development resulted in a great variety of novel LSSRFPs; however, the selection of the right LSSRFP for a given application is hampered by the lack of a side-by-side comparison of the LSSRFPs' performance. In this study, we employed rational design and random mutagenesis to convert conventional bright RFP mScarlet into LSSRFP, called LSSmScarlet, characterized by excitation/emission maxima at 470/598 nm. In addition, we utilized the previously reported LSSRFPs mCyRFP1, CyOFP1, and mCRISPRed as templates for directed molecular evolution to develop their optimized versions, called dCyRFP2s, dCyOFP2s and CRISPRed2s. We performed a quantitative assessment of the developed LSSRFPs and their precursors in vitro on purified proteins and compared their brightness at 488 nm excitation in the mammalian cells. The monomeric LSSmScarlet protein was successfully utilized for the confocal imaging of the structural proteins in live mammalian cells and multicolor confocal imaging in conjugation with other FPs. LSSmScarlet was successfully applied for dual-color two-photon imaging in live mammalian cells. We also solved the X-ray structure of the LSSmScarlet protein at the resolution of 1.4 Å that revealed a hydrogen bond network supporting excited-state proton transfer (ESPT). Quantum mechanics/molecular mechanics molecular dynamic simulations confirmed the ESPT mechanism of a large Stokes shift. Structure-guided mutagenesis revealed the role of R198 residue in ESPT that allowed us to generate a variant with improved pH stability. Finally, we showed that LSSmScarlet protein is not appropriate for STED microscopy as a consequence of LSSRed-to-Red photoconversion with high-power 775 nm depletion light.

The Influence of Electronic Polarization on Nonlinear Optical Spectroscopy

The environment surrounding a chromophore can dramatically affect the energy absorption and relaxation process, as manifested in optical spectra. Simulations of nonlinear optical spectroscopy, such as two-dimensional electronic spectroscopy (2DES) and transient absorption (TA), will be influenced by the computational model of the environment. We here compare a fixed point charge molecular mechanics model and a quantum mechanical (QM) model of the environment in computed 2DES and TA spectra of Nile red in water and the chromophore of photoactive yellow protein (PYP) in water and protein environments. In addition to simulating these nonlinear optical spectra, we directly juxtapose the computed excitation energy correlation function to the dynamic Stokes shift function often used to analyze environment dynamics. Overall, we find that for the three systems studied here the mutual electronic polarization provided by the QM environment manifests in broader 2DES signals, as well as a larger reorganization energy and a larger static Stokes shift due to stronger coupling between the chromophore and the environment.

Two Sides of Quantum-Based Modeling of Enzyme-Catalyzed Reactions: Mechanistic and Electronic Structure Aspects of the Hydrolysis by Glutamate Carboxypeptidase

We report the results of a computational study of the hydrolysis reaction mechanism of N-acetyl-l-aspartyl-l-glutamate (NAAG) catalyzed by glutamate carboxypeptidase II. Analysis of both mechanistic and electronic structure aspects of this multistep reaction is in the focus of this work. In these simulations, model systems are constructed using the relevant crystal structure of the mutated inactive enzyme. After selection of reaction coordinates, the Gibbs energy profiles of elementary steps of the reaction are computed using molecular dynamics simulations with ab initio type QM/MM potentials (QM/MM MD). Energies and forces in the large QM subsystem are estimated in the DFT(PBE0-D3/6-31G**) approximation. The established mechanism includes four elementary steps with the activation energy barriers not exceeding 7 kcal/mol. The models explain the role of point mutations in the enzyme observed in the experimental kinetic studies; namely, the Tyr552Ile substitution disturbs the "oxyanion hole", and the Glu424Gln replacement increases the distance of the nucleophilic attack. Both issues diminish the substrate activation in the enzyme active site. To quantify the substrate activation, we apply the QTAIM-based approaches and the NBO analysis of dynamic features of the corresponding enzyme-substrate complexes. Analysis of the 2D Laplacian of electron density maps allows one to define structures with the electron density deconcentration on the substrate carbon atom, i.e., at the electrophilic site of reactants. The similar electronic structure element in the NBO approach is a lone vacancy on the carbonyl carbon atom in the reactive species. The electronic structure patterns revealed in the NBO and QTAIM-based analyses consistently clarify the reactivity issues in this system.

Quantum Chemical Study of a Radical Relay Mechanism for the HydG-Catalyzed Synthesis of a Fe(II)(CO)2(CN)cysteine Precursor to the H-Cluster of [FeFe] Hydrogenase

The [FeFe] hydrogenase catalyzes the redox interconversion of protons and H₂ with a Fe-S "H-cluster" employing CO, CN, and azadithiolate ligands to two Fe centers. The biosynthesis of the H-cluster is a highly interesting reaction carried out by a set of Fe-S maturase enzymes called HydE, HydF, and HydG. HydG, a member of the radical S-adenosylmethionine (rSAM) family, converts tyrosine, cysteine, and Fe(II) into an organometallic Fe(II)(CO)₂(CN)cysteine "synthon", which serves as the substrate for HydE. Although key aspects of the HydG mechanism have been experimentally determined via isotope-sensitive spectroscopic methods, other important mechanistic questions have eluded experimental determination. Here, we use computational quantum chemistry to refine the mechanism of the HydG catalytic reaction. We utilize quantum mechanics/molecular mechanics simulations to investigate the reactions at the canonical Fe-S cluster, where a radical cleavage of the tyrosine substrate takes place and proceeds through a relay of radical intermediates to form HCN and a COO^•- radical anion. We then carry out a broken-symmetry density functional theory study of the reactions at the unusual five-iron auxiliary Fe-S cluster, where two equivalents of CN^- and COOH^• coordinate to the fifth "dangler iron" in a series of substitution and redox reactions that yield the synthon as the final product for further processing by HydE.

Combining Machine Learning and Computational Chemistry for Predictive Insights Into Chemical Systems

Machine learning models are poised to make a transformative impact on chemical sciences by dramatically accelerating computational algorithms and amplifying insights available from computational chemistry methods. However, achieving this requires a confluence and coaction of expertise in computer science and physical sciences. This Review is written for new and experienced researchers working at the intersection of both fields. We first provide concise tutorials of computational chemistry and machine learning methods, showing how insights involving both can be achieved. We follow with a critical review of noteworthy applications that demonstrate how computational chemistry and machine learning can be used together to provide insightful (and useful) predictions in molecular and materials modeling, retrosyntheses, catalysis, and drug design.

Computational Discovery of Transition-metal Complexes: From High-throughput Screening to Machine Learning

Transition-metal complexes are attractive targets for the design of catalysts and functional materials. The behavior of the metal-organic bond, while very tunable for achieving target properties, is challenging to predict and necessitates searching a wide and complex space to identify needles in haystacks for target applications. This review will focus on the techniques that make high-throughput search of transition-metal chemical space feasible for the discovery of complexes with desirable properties. The review will cover the development, promise, and limitations of "traditional" computational chemistry (i.e., force field, semiempirical, and density functional theory methods) as it pertains to data generation for inorganic molecular discovery. The review will also discuss the opportunities and limitations in leveraging experimental data sources. We will focus on how advances in statistical modeling, artificial intelligence, multiobjective optimization, and automation accelerate discovery of lead compounds and design rules. The overall objective of this review is to showcase how bringing together advances from diverse areas of computational chemistry and computer science have enabled the rapid uncovering of structure-property relationships in transition-metal chemistry. We aim to highlight how unique considerations in motifs of metal-organic bonding (e.g., variable spin and oxidation state, and bonding strength/nature) set them and their discovery apart from more commonly considered organic molecules. We will also highlight how uncertainty and relative data scarcity in transition-metal chemistry motivate specific developments in machine learning representations, model training, and in computational chemistry. Finally, we will conclude with an outlook of areas of opportunity for the accelerated discovery of transition-metal complexes.

Electronic Structure Methods for the Description of Nonadiabatic Effects and Conical Intersections

Nonadiabatic effects are ubiquitous in photophysics and photochemistry, and therefore, many theoretical developments have been made to properly describe them. Conical intersections are central in nonadiabatic processes, as they promote efficient and ultrafast nonadiabatic transitions between electronic states. A proper theoretical description requires developments in electronic structure and specifically in methods that describe conical intersections between states and nonadiabatic coupling terms. This review focuses on the electronic structure aspects of nonadiabatic processes. We discuss the requirements of electronic structure methods to describe conical intersections and nonadiabatic couplings, how the most common excited state methods perform in describing these effects, and what the recent developments are in expanding the methodology and implementing nonadiabatic couplings.

Photochemical Deracemization of Primary Allene Amides by Triplet Energy Transfer: A Combined Synthetic and Theoretical Study

The photochemical deracemization of 2,4-disubstituted 2,3-butadienamides (allene amides) was investigated both experimentally and theoretically. The reaction was catalyzed by a thioxanthone which is covalently linked to a chiral 1,5,7-trimethyl-3-azabicyclo[3.3.1]nonan-2-one skeleton providing a U-shaped arrangement of the sensitizing unit relative to a potential hydrogen-bonding site. Upon irradiation at λ = 420 nm in the presence of the sensitizer (2.5 mol %), the amides reached at -10 °C a photostationary state in which one enantiomer prevailed. The enantioenriched allene amides (70-93% ee) were isolated in 74% to quantitative yield (19 examples). Based on luminescence data and DFT calculations, energy transfer from the thioxanthone to the allene amides is thermodynamically feasible, and the achiral triplet allene intermediate was structurally characterized. Hydrogen bonding of the amide enantiomers to the sensitizer was monitored by NMR titration. The experimental association constants (K_a) were similar (59.8 vs 25.7 L·mol^-1). DFT calculations, however, revealed a significant difference in the binding properties of the two enantiomers. The major product enantiomer exhibits a noncovalent dispersion interaction of its arylmethyl group to the external benzene ring of the thioxanthone, thus moving away the allene from the carbonyl chromophore. The minor enantiomer displays a CH-π interaction of the hydrogen atom at the terminal allene carbon atom to the same benzene ring, thus forcing the allene into close proximity to the chromophore. The binding behavior explains the observed enantioselectivity which, as corroborated by additional calculations, is due to a rapid triplet energy transfer within the substrate-catalyst complex of the minor enantiomer.

Flyby reaction trajectories: Chemical dynamics under extrinsic force

Dynamic effects are an important determinant of chemical reactivity and selectivity, but the deliberate manipulation of atomic motions during a chemical transformation is not straightforward. Here, we demonstrate that extrinsic force exerted upon cyclobutanes by stretching pendant polymer chains influences product selectivity through force-imparted nonstatistical dynamic effects on the stepwise ring-opening reaction. The high product stereoselectivity is quantified by carbon-13 labeling and shown to depend on external force, reactant stereochemistry, and intermediate stability. Computational modeling and simulations show that, besides altering energy barriers, the mechanical force activates reactive intramolecular motions nonstatistically, setting up "flyby trajectories" that advance directly to product without isomerization excursions. A mechanistic model incorporating nonstatistical dynamic effects accounts for isomer-dependent mechanochemical stereoselectivity.

Harnessing the Power of Multi-GPU Acceleration into the Quantum Interaction Computational Kernel Program

We report a new multi-GPU capable ab initio Hartree-Fock/density functional theory implementation integrated into the open source QUantum Interaction Computational Kernel (QUICK) program. Details on the load balancing algorithms for electron repulsion integrals and exchange correlation quadrature across multiple GPUs are described. Benchmarking studies carried out on up to four GPU nodes, each containing four NVIDIA V100-SXM2 type GPUs demonstrate that our implementation is capable of achieving excellent load balancing and high parallel efficiency. For representative medium to large size protein/organic molecular systems, the observed parallel efficiencies remained above 82% for the Kohn-Sham matrix formation and above 90% for nuclear gradient calculations. The accelerations on NVIDIA A100, P100, and K80 platforms also have realized parallel efficiencies higher than 68% in all tested cases, paving the way for large-scale ab initio electronic structure calculations with QUICK.

Mechanism of Guanosine Triphosphate Hydrolysis by the Visual Proteins Arl3-RP2: Free Energy Reaction Profiles Computed with Ab Initio Type QM/MM Potentials

We report the results of calculations of the Gibbs energy profiles of the guanosine triphosphate (GTP) hydrolysis by the Arl3-RP2 protein complex using molecular dynamics (MD) simulations with ab initio type QM/MM potentials. The chemical reaction of GTP hydrolysis to guanosine diphosphate (GDP) and inorganic phosphate (Pi) is catalyzed by GTPases, the enzymes, which are responsible for signal transduction in live cells. A small GTPase Arl3, catalyzing the GTP → GDP reaction in complex with the activating protein RP2, constitute an essential part of the human vision cycle. To simulate the reaction mechanism, a model system is constructed by motifs of the crystal structure of the Arl3-RP2 complexed with a substrate analog. After selection of reaction coordinates, energy profiles for elementary steps along the reaction pathway GTP + H2O → GDP + Pi are computed using the umbrella sampling and umbrella integration procedures. QM/MM MD calculations are carried out, interfacing the molecular dynamics program NAMD and the quantum chemistry program TeraChem. Ab initio type QM(DFT)/MM potentials are computed with atom-centered basis sets 6-31G** and two hybrid functionals (PBE0-D3 and ωB97x-D3) of the density functional theory, describing a large QM subsystem. Results of these simulations of the reaction mechanism are compared to those obtained with QM/MM calculations on the potential energy surface using a similar description of the QM part. We find that both approaches, QM/MM and QM/MM MD, support the mechanism of GTP hydrolysis by GTPases, according to which the catalytic glutamine side chain (Gln71, in this system) actively participates in the reaction. Both approaches distinguish two parts of the reaction: the cleavage of the phosphorus-oxygen bond in GTP coupled with the formation of Pi, and the enzyme regeneration. Newly performed QM/MM MD simulations confirmed the profile predicted in the QM/MM minimum energy calculations, called here the pathway-I, and corrected its relief at the first elementary step from the enzyme-substrate complex. The QM/MM MD simulations also revealed another mechanism at the part of enzyme regeneration leading to pathway-II. Pathway-II is more consistent with the experimental kinetic data of the wild-type complex Arl3-RP2, whereas pathway-I explains the role of the mutation Glu138Gly in RP2 slowing down the hydrolysis rate.

Quantum chemistry for molecules at extreme pressure on graphical processing units: Implementation of extreme-pressure polarizable continuum model

Pressure plays essential roles in chemistry by altering structures and controlling chemical reactions. The extreme-pressure polarizable continuum model (XP-PCM) is an emerging method with an efficient quantum mechanical description of small- and medium-sized molecules at high pressure (on the order of GPa). However, its application to large molecular systems was previously hampered by a CPU computation bottleneck: the Pauli repulsion potential unique to XP-PCM requires the evaluation of a large number of electric field integrals, resulting in significant computational overhead compared to the gas-phase or standard-pressure polarizable continuum model calculations. Here, we exploit advances in graphical processing units (GPUs) to accelerate the XP-PCM-integral evaluations. This enables high-pressure quantum chemistry simulation of proteins that used to be computationally intractable. We benchmarked the performance using 18 small proteins in aqueous solutions. Using a single GPU, our method evaluates the XP-PCM free energy of a protein with over 500 atoms and 4000 basis functions within half an hour. The time taken by the XP-PCM-integral evaluation is typically 1% of the time taken for a gas-phase density functional theory (DFT) on the same system. The overall XP-PCM calculations require less computational effort than that for their gas-phase counterpart due to the improved convergence of self-consistent field iterations. Therefore, the description of the high-pressure effects with our GPU-accelerated XP-PCM is feasible for any molecule tractable for gas-phase DFT calculation. We have also validated the accuracy of our method on small molecules whose properties under high pressure are known from experiments or previous theoretical studies.

Putting Density Functional Theory to the Test in Machine-Learning-Accelerated Materials Discovery

Accelerated discovery with machine learning (ML) has begun to provide the advances in efficiency needed to overcome the combinatorial challenge of computational materials design. Nevertheless, ML-accelerated discovery both inherits the biases of training data derived from density functional theory (DFT) and leads to many attempted calculations that are doomed to fail. Many compelling functional materials and catalytic processes involve strained chemical bonds, open-shell radicals and diradicals, or metal-organic bonds to open-shell transition-metal centers. Although promising targets, these materials present unique challenges for electronic structure methods and combinatorial challenges for their discovery. In this Perspective, we describe the advances needed in accuracy, efficiency, and approach beyond what is typical in conventional DFT-based ML workflows. These challenges have begun to be addressed through ML models trained to predict the results of multiple methods or the differences between them, enabling quantitative sensitivity analysis. For DFT to be trusted for a given data point in a high-throughput screen, it must pass a series of tests. ML models that predict the likelihood of calculation success and detect the presence of strong correlation will enable rapid diagnoses and adaptation strategies. These "decision engines" represent the first steps toward autonomous workflows that avoid the need for expert determination of the robustness of DFT-based materials discoveries.

Exploring the Minimum-Energy Pathways and Free-Energy Profiles of Enzymatic Reactions with QM/MM Calculations

Understanding molecular mechanisms of enzymatic reactions is of vital importance in biochemistry and biophysics. Here, we introduce new functions of hybrid quantum mechanical/molecular mechanical (QM/MM) calculations in the GENESIS program to compute the minimum-energy pathways (MEPs) and free-energy profiles of enzymatic reactions. For this purpose, an interface in GENESIS is developed to utilize a highly parallel electronic structure program, QSimulate-QM (https://qsimulate.com), calling it as a shared library from GENESIS. Second, algorithms to search the MEP are implemented, combining the string method (E et al. J. Chem. Phys. 2007, 126, 164103) with the energy minimization of the buffer MM region. The method implemented in GENESIS is applied to an enzyme, triosephosphate isomerase, which converts dihyroxyacetone phosphate to glyceraldehyde 3-phosphate in four proton-transfer processes. QM/MM-molecular dynamics simulations show performances of greater than 1 ns/day with the density functional tight binding (DFTB), and 10-30 ps/day with the hybrid density functional theory, B3LYP-D3. These performances allow us to compute not only MEP but also the potential of mean force (PMF) of the enzymatic reactions using the QM/MM calculations. The barrier height obtained as 13 kcal mol-1 with B3LYP-D3 in the QM/MM calculation is in agreement with the experimental results. The impact of conformational sampling in PMF calculations and the level of electronic structure calculations (DFTB vs B3LYP-D3) suggests reliable computational protocols for enzymatic reactions without high computational costs.

First-Principles Nonadiabatic Dynamics Simulation of Azobenzene Photodynamics in Solutions

The photoisomerization of azobenzene is a prototypical reaction of various light-activated processes in material and biomedical sciences. However, its reaction mechanism has been under debate for decades, partly due to the challenges in computational simulations to accurately describe the molecule's photodynamics. A recent study (J. Am. Chem. Soc. 2020, 142 (49), 20,680-20,690) addressed the challenges by combining the hole-hole Tamm-Dancoff Approximated (hh-TDA) density functional theory (DFT) method with the ab initio multiple spawning (AIMS) algorithm. The hh-TDA-DFT/AIMS method was applied to first-principles nonadiabatic dynamics simulation of azobenzene's photodynamics in the vacuum. However, it remains necessary to benchmark this new method in realistic molecular environments against experimental data. In the current work, the hh-TDA-DFT/AIMS method was employed in a quantum mechanics/molecular mechanics setting to characterize the trans azobenzene's photodynamics in explicit methanol and n-hexane solvents, following both the S₁ (nπ*) and S₂ (ππ*) excitations. The simulated absorption and fluorescence spectra following the S₂ excitation quantitatively agree with the experiments. However, the hh-TDA-DFT method overestimates the torsional barrier on the S₁ state, leading to an overestimation of the S₁ state lifetime. The excited-state population decays to the ground state through two competing channels. The reactive channel partially yields the cis azobenzene photoproduct, and the unreactive channel exclusively leads to the reactant. The S₂ excitation increases the decay through the unreactive channel and thus decreases the isomerization quantum yield compared to the S₁ excitation. The solvent slows down the azobenzene's torsional dynamics on the S₁ state, but its polarity minimally affects the reaction kinetics and quantum yields. Interestingly, the dynamics of the central torsion and angles of azobenzene play a critical role in determining the final isomer of the azobenzene. This benchmark study validates the hh-TDA-DFT/AIMS method's accuracy for simulating the azobenzene's photodynamics in realistic molecular environments.

The integrative biology of pigment organelles, a quantum chemical approach

Coloration is a complex phenotypic trait involving both physical and chemical processes at a multiscale level, from molecules to tissues. Pigments, whose main property is to absorb specific wavelengths of visible light, are usually deposited in specialized organelles or complex matrices comprising proteins, metals, ions and redox compounds, among others. By modulating electronic properties and stability, interactions between pigments and these molecular actors can lead to color tuning. Furthermore, pigments are not only important for visual effects but also provide other critical functions, such as detoxification and antiradical activity. Hence, integrative studies of pigment organelles are required to understand how pigments interact with their cellular environment. In this review, we show how quantum chemistry, a computational method that models the molecular and optical properties of pigments, has provided key insights into the mechanisms by which pigment properties, from color to reactivity, are modulated by their organellar environment. These results allow to rationalize and to predict the way pigments behave in supramolecular complexes, up to the complete modelling of pigment organelles. We also discuss the main limitations of quantum chemistry, emphasizing the need for carrying experimental work with identical vigor. We finally suggest that taking into account the ecology of pigments (i.e. how they interact with these various other cellular components and at higher organizational levels) will lead to a greater understanding of how and why animals are vividly and variably colored, two fundamental questions in organismal biology.

Electrostatic Control of Photoisomerization in Channelrhodopsin 2

Channelrhodopsin 2 (ChR2) is the most commonly used tool in optogenetics. Because of its faster photocycle compared to wild-type (WT) ChR2, the E123T mutant of ChR2 is a useful optogenetic tool when fast neuronal stimulation is needed. Interestingly, in spite of its faster photocycle, the initial step of the photocycle in E123T (photoisomerization of retinal protonated Schiff base or RPSB) was found experimentally to be much slower than that of WT ChR2. The E123T mutant replaces the negatively charged E123 residue with a neutral T123 residue, perturbing the electric field around the RPSB. Understanding the RPSB photoisomerization mechanism in ChR2 mutants will provide molecular-level insights into how ChR2 photochemical reactivity can be controlled, which will lay the foundation for improving the design of optogenetic tools. In this work, we combine ab initio nonadiabatic dynamics simulation, excited state free energy calculation, and reaction path search to comprehensively characterize the RPSB photoisomerization mechanism in the E123T mutant of ChR2. Our simulation agrees with previous experiments in predicting a red-shifted absorption spectrum and significant slowdown of photoisomerization in the E123T mutant. Interestingly, our simulations predict similar photoisomerization quantum yields for the mutant and WT despite the differences in excited-state lifetime and absorption maximum. Upon mutation, the neutralization of the negative charge on the E123 residue increases the isomerization barrier, alters the reaction pathway, and changes the relative stability of two fluorescent states. Our findings provide new insight into the intricate role of the electrostatic environment on the RPSB photoisomerization mechanism in microbial rhodopsins.

Boronic Acids as Prospective Inhibitors of Metallo-β-Lactamases: Efficient Chemical Reaction in the Enzymatic Active Site Revealed by Molecular Modeling

Boronic acids are prospective compounds in inhibition of metallo-β-lactamases as they form covalent adducts with the catalytic hydroxide anion in the enzymatic active site upon binding. We compare this chemical reaction in the active site of the New Delhi metallo-β-lactamase (NDM-1) with the hydrolysis of the antibacterial drug imipenem. The nucleophilic attack occurs with the energy barrier of 14 kcal/mol for imipenem and simultaneously upon binding a boronic acid inhibitor. A boron atom of an inhibitor exhibits stronger electrophilic properties than the carbonyl carbon atom of imipenem in a solution that is quantified by atomic Fukui indices. Upon forming the prereaction complex between NDM-1 and inhibitor, the lone electron pair of the nucleophile interacts with the vacant p-orbital of boron that facilitates the chemical reaction. We analyze a set of boronic acid compounds with the benzo[b]thiophene core complexed with the NDM-1 and propose quantitative structure-sroperty relationship (QSPR) equations that can predict IC50 values from the calculated descriptors of electron density. These relations are applied to classify other boronic acids with the same core found in the database of chemical compounds, PubChem, and proposed ourselves. We demonstrate that the IC50 values for all considered benzo[b]thiophene-containing boronic acid inhibitors are 30-70 μM.

Light-Induced Change of Arginine Conformation Modulates the Rate of Adenosine Triphosphate to Cyclic Adenosine Monophosphate Conversion in the Optogenetic System Containing Photoactivated Adenylyl Cyclase

We report the first computational characterization of an optogenetic system composed of two photosensing BLUF (blue light sensor using flavin adenine dinucleotide) domains and two catalytic adenylyl cyclase (AC) domains. Conversion of adenosine triphosphate (ATP) to the reaction products, cyclic adenosine monophosphate (cAMP) and pyrophosphate (PPi), catalyzed by ACs initiated by excitation in photosensing domains has emerged in the focus of modern optogenetic applications because of the request in photoregulated enzymes that modulate cellular concentrations of signaling messengers. The photoactivated AC from the soil bacterium Beggiatoa sp. (bPAC) is an important model showing a considerable increase in the ATP to cAMP conversion rate in the catalytic domain after the illumination of the BLUF domain. The 1 μs classical molecular dynamics simulations reveal that the activation of the BLUF domain leading to tautomerization of Gln49 in the chromophore-binding pocket results in switching of the position of the side chain of Arg278 in the active site of AC. Allosteric signal transmission pathways between Gln49 from BLUF and Arg278 from AC were revealed by the dynamical network analysis. The Gibbs energy profiles of the ATP → cAMP + PPi reaction computed using QM(DFT(ωB97X-D3/6-31G**))/MM(CHARMM) molecular dynamics simulations for both Arg278 conformations in AC clarify the reaction mechanism. In the light-activated system, the corresponding arginine conformation stabilizes the pentacoordinated phosphorus of the α-phosphate group in the transition state, thus lowering the activation energy. Simulations of the bPAC system with the Tyr7Phe replacement in the BLUF demonstrate occurrence of both arginine conformations in an equal ratio, explaining the experimentally observed intermediate catalytic activity of the bPAC-Y7F variant as compared with the dark and light states of the wild-type bPAC.

Explicit environmental and vibronic effects in simulations of linear and nonlinear optical spectroscopy

Accurately simulating the linear and nonlinear electronic spectra of condensed phase systems and accounting for all physical phenomena contributing to spectral line shapes presents a significant challenge. Vibronic transitions can be captured through a harmonic model generated from the normal modes of a chromophore, but it is challenging to also include the effects of specific chromophore-environment interactions within such a model. We work to overcome this limitation by combining approaches to account for both explicit environment interactions and vibronic couplings for simulating both linear and nonlinear optical spectra. We present and show results for three approaches of varying computational cost for combining ensemble sampling of chromophore-environment configurations with Franck-Condon line shapes for simulating linear spectra. We present two analogous approaches for nonlinear spectra. Simulated absorption spectra and two-dimensional electronic spectra (2DES) are presented for the Nile red chromophore in different solvent environments. Employing an average Franck-Condon or 2DES line shape appears to be a promising method for simulating linear and nonlinear spectroscopy for a chromophore in the condensed phase.

Voice-controlled quantum chemistry

Over the past decade, artificial intelligence has been propelled forward by advances in machine learning algorithms and computational hardware, opening up myriads of new avenues for scientific research. Nevertheless, virtual assistants and voice control have yet to be widely used in the natural sciences. Here, we present ChemVox, an interactive Amazon Alexa skill that uses speech recognition to perform quantum chemistry calculations. This new application interfaces Alexa with cloud computing and returns the results through a capable device. ChemVox paves the way to making computational chemistry routinely accessible to the wider community.

An ab initio exciton model for singlet fission

We present an ab initio exciton model that extends the Frenkel exciton model and includes valence, charge-transfer, and multiexcitonic excited states. It serves as a general, parameter-free, yet computationally efficient and scalable approach for simulation of singlet fission processes in multichromophoric systems. A comparison with multiconfigurational methods confirms that our exciton model predicts consistent energies and couplings for the pentacene dimer and captures the correct physics. Calculations of larger pentacene clusters demonstrate the computational scalability of the exciton model and suggest that the mixing between local and charge-transfer excitations narrows the gap between singlet and multiexcitonic states. Local vibrations of pentacene molecules are found to facilitate singlet-multiexcitonic state-crossing and hence are important for understanding singlet fission. The exciton model developed in this work also sets the stage for further implementation of the nuclear gradients and nonadiabatic couplings needed for first principles nonadiabatic quantum molecular dynamics simulations of singlet fission.