Ultrathin two-dimensional (2D) inorganic materials, characterized by their exceptionally high specific surface areas and tunable electronic structures, have emerged as ideal platforms for advancing small-molecule photocatalysis. Yet, the intrinsic limitations of pristine 2D materials - such as rigid band structures and catalytically inert surface sites - often restrict their practical photocatalytic efficiency. Strategic atomic-level modification therefore becomes imperative to unlock their full potential. This review systematically classifies modification strategies for ultrathin 2D materials into two complementary paradigms: surface atomic engineering, which focuses on charge localization and active-site optimization, and interface atomic engineering, which facilitates directional charge transfer and heterojunction-enabled carrier separation. Through a synergistic combination of theoretical simulations and advanced experimental techniques, we first elucidate how these approaches precisely tailor the electronic landscapes of ultrathin 2D materials. We then critically examine their impact on the activity and selectivity of four key photocatalytic small-molecule transformations: H2O splitting, CO2 reduction, alkane conversion, and N2 fixation, offering strategic insights for overcoming reaction-specific bottlenecks. Furthermore, we highlight state-of-the-art in situ characterization methods that establish structure-activity correlations under working conditions. Looking forward, we outline transformative prospects in the field, including big-data-assisted design of surface/interface architectures, next-generation multiscale simulation frameworks, and scalable flow-reactor technologies. These pathways aim not only to deepen the fundamental understanding of photocatalytic mechanisms, but also to accelerate the transition of ultrathin 2D photocatalysts from laboratory innovation toward real-world solar-fuel and chemical production.
In solid state and nanoscale materials, a phase refers to a distinct entity that can be identified in X-ray diffraction (XRD) analysis. For bulk metals, common phases include the face-centered cubic (fcc), hexagonal close-packed (hcp), and body-centered cubic (bcc) structures. When the dimensions of a metal are shrunk to the nanometer scale (1-100 nm), new phases that are not observed in bulk metals can be attained due to the increasing importance of surface energy as a new handle for phase control. In this review, we summarize the recent progress in phase engineering of atomically precise nanoclusters (APNCs, typically 1-3 nm) of coinage metals such as gold and silver. Five primary types of phases are categorized, including the fcc, hcp, bcc, multi-twinned phases (both the icosahedron and decahedron), as well as mixed phases in APNCs. The attainment of non-fcc phases at the nanoscale is intriguing because bulk gold and silver are exclusively fcc; this underscores the importance of tailoring surface energy by protecting ligands as well as the kinetic control and other synthetic strategies as discussed in this review. Following the structural discussion, we illustrate the distinct physical and chemical properties from phase engineering of APNCs, including the optical absorption and related photothermal application, photoluminescence in the near-infrared (NIR) and related bioimaging application, as well as the catalytic reactivity. Finally, we put forth some future perspectives.
The structural dynamics of DNA underpin essential biological processes, yet conventional structural biology methods often obscure conformational heterogeneity through ensemble averaging. Atomic force microscopy (AFM) provides single-molecule topographical maps capable of capturing both local and global variation, but extracting quantitative insight from these images remains challenging. Here, we introduce an automated framework that reduces AFM data to spline representations of the DNA backbone and applies cyclic Procrustes analysis to quantify shape similarity across ensembles. Using purified topoisomers of 339 bp DNA minicircles ranging from relaxed to highly negatively supercoiled, we resolved and measured the relative abundance of conformational states across the different topoisomers, capturing gradual transitions among open circles, compact conformations, and self-crossing structures that are invisible to techniques such as gel electrophoresis or cryoelectron microscopy (cryo-EM). We show that beyond quantification, Procrustes distances provide supervisory signals for neural network training, enabling feature extraction tuned to conformational geometry and supporting robust conformation classification of AFM images. Extending the same spline representation to molecular dynamics simulations allows experimental and computational ensembles to be directly compared, establishing a common shape-based framework for probing conformational variability. Together, these advances transform AFM from a descriptive imaging tool into a quantitative platform for mapping conformational continua, with broad applicability to DNA and other dynamic biomolecular systems.
A trimetallic Au3Ag3Cu6 nanocluster, obtained by Ag doping of Au3Cu9, exhibits enhanced NO3RR-to-NH3 performance, delivering 90.66% faradaic efficiency at -0.6 V vs. RHE. DFT calculations reveal that Ag incorporation strengthens Au-Ag-Cu electronic synergy, optimizes key intermediate adsorption/conversion, and promotes reaction kinetics.
Fluorine-containing volatile organic compound (FVOCs) elimination is urgently desired for protecting the ozone layer and alleviating global warming, but high temperature and catalyst deactivation hinder the development of greener catalytic oxidation. Accurate design of atomic cluster catalysts with special functional sites has been a longstanding challenge. Here, the supported WMo atomic cluster exhibits extraordinarily activity and selectivity for CCl2F2 and CF3CH2F hydrolysis-oxidation. A series of experiments and AIMD simulation results of H2O dissociation over the DFT-optimized models of Mo/TiO2 and WMo/TiO2 strongly support that the bimetallic atomic catalysts possess a unique "W-OH-Mo" bridging hydroxyl group (OHB). With the assistance of a considerable number of OHB at the W-OH-Mo sites, the reaction rate for CCl2F2 was 0.174 μmol g-1 s-1 at 280 °C, much higher than that over reported catalysts, and CF3CH2F elimination reached 85% conversion at a recorded low temperature (480 °C). The significant promotional effect arises from surface OHB, which promotes the cleavage of the αC-F bond and dehydrofluorination in R134a to form alkene intermediates, thus enabling higher FVOCs conversion and enhanced production of CO2 and HF. The consumed OHB was regenerated through the coactivation of O2 and H2O at high-concentration oxygen vacancies between neighboring W and Mo atoms. This work opens a promising pathway for developing robust atomic cluster catalysts for environmental catalysis.
Ultrafine high-entropy intermetallics (HEIs), which synergistically combine high-entropy effect, size effect, and ordered intermetallic structure, have attracted considerable attention as promising candidates for highly active and durable catalysts. However, the sluggish diffusion inherent to the complex multielement environment of high-entropy systems necessitates high temperatures to drive atomic ordering, which typically leads to severe nanoparticle sintering in the traditional thermal process. Here, we address this challenge via a multithermal-pulse control strategy that effectively suppresses long-range atomic diffusion-mediated interparticle sintering and intraparticle phase separation during short-range atomic ordering, enabling the synthesis of uniform sub-3 nm HEI nanoparticles. By varying the number of thermal pulses to precisely control the chemical ordering degree of PtFeCuCoNi alloy nanoparticles, we discover a clear positive correlation between the ordering degree and oxygen reduction performance. The HO-PtFeCuCoNi nanoparticles with the highest ordering degree exhibit the best catalytic activity (MA of 0.95 A mgPt-1 and SA of 1.18 mA cmPt-2) and stability (MA retention of 70.8%, SA retention of 70.4%, and ΔE1/2 of 6 mV) compared to their less ordered counterparts and commercial Pt/C. This multipulse strategy provides a feasible route to utilize pulse dynamics for capturing ordered structures and regulating the sintering-ordering trade-off and offers useful guidance for the structural design of multicomponent alloys.
Atomically precise gold nanoclusters have garnered significant attention for their diverse applications, ranging from biological labeling to optoelectronics. Their potential in optical quantum computing, which calls for ideal single-photon sources, has recently become a key area of interest. In the current work, we use photon antibunching experiments to explore the single-photon emission efficiency of atomically precise Au24 nanoclusters protected by 4-tert-butylbenzyl mercaptan ligands (Au24(TBBM)20). This cluster exhibits quantum emission with good photostability and without any observable blinking or spectral drift at room temperature under an inert gas atmosphere, with antibunching dips (g2(0)) as low as 0.07 in the solid state or, equivalently, a single-photon purity of 93% under time-gated conditions. Transient absorption and time-gated antibunching studies reveal that the short emission lifetime of this cluster and its high photoluminescence quantum yield in the solid state play critical roles in enhancing the emitted single-photon purity. This research advances the understanding of single-emitter behavior in atomically precise gold nanoclusters, contributing to the development of stable quantum emitters that are essential for quantum computing and cryptography.
Lithium-sulfur (Li-S) batteries have been considered among the most promising next-generation battery systems owing to their exceptionally high theoretical energy density, low cost, and environmental friendliness. However, their development continues to be hindered by the dissolution and sluggish conversion kinetics of the intermediate polysulfides. Efficient catalysts have shown significant potential in anchoring and catalytically converting polysulfides. Among them, dual-atom catalysts (DACs) are gaining increasing attention due to their high atomic utilization efficiency and structurally well-defined active sites. Through the synergistic interaction between neighboring metal atoms, DACs demonstrate enhanced capabilities for the coordinated adsorption and catalytic conversion of polysulfides, which helps minimize the shuttle effect and improve reaction kinetics. This review offers a comprehensive overview of recent advances in DACs for Li-S batteries, including the controlled synthesis, atomic-scale structural characterization, and detailed mechanistic insights into both homonuclear and heteronuclear DACs. It also highlights the promising role of combined first-principles calculations and machine learning in guiding the rational design and rapid screening of high-performance DACs. Finally, key challenges and future research directions are outlined, emphasizing the pathway toward computationally-guided design and practical implementation of DACs in advanced energy storage systems.
The atomically precise engineering of impurities in graphene and the understanding of their structural and carrier-dependent electronic properties at the nanoscale are crucial for advancing graphene-based nanoelectronics, catalysis, and energy technologies. Here, we demonstrate controllable incorporation of the elusive 3-fold-coordinated O substitutions into graphene using low-energy O+ ion implantation under ultrahigh-vacuum conditions. By combining high-resolution scanning tunneling microscopy and spectroscopy (STM/S), bond-resolved noncontact atomic force microscopy techniques, and density functional theory (DFT) calculations, we resolve both the structural and electronic properties of the O-related defects. The STM/S measurements, corroborated by DFT calculations, uncover a characteristic impurity state that is energetically pinned to the Dirac point across different charge-carrier doping regimes. Molecular dynamics simulations further reveal the distribution of implantation-induced configurations and identify the formation of 3-fold-coordinated O dopants. This work provides a viable route to incorporate 3-fold-coordinated O dopants and opens new opportunities for controlled defect engineering in graphene.
Photothermal catalysis has emerged as a powerful strategy to complement photocatalysis by harnessing full-spectrum sunlight for the production of solar fuels and chemical upgrading. Despite its promise, practical implementation remains limited by inefficient management of light-to-electron/heat conversion and unclear catalytic mechanisms. Superlattice materials, featuring periodic structural order across atomic to macroscopic scales, offer tuneable sized crystals, controllable electronic structures, and unique catalytic functionalities. These attributes enable enhanced light capture, energy conversion, and highly efficient catalytic kinetics for photothermal catalysis. This review provides an overview of superlattice architectures in photothermal catalysis, with a systematic review of the mechanism at both atomic and macroscopic scales. Emphasis is placed on the mechanism of superlattice engineering in regulating key photothermal processes, including photo-electron-phonon coupling and multi-energy-carrier dynamics. Advanced microscopic and operando characterization techniques are highlighted to elucidate reaction pathways and disentangle photothermal and photoelectrochemical contributions. Representative photothermal reactions are discussed to demonstrate how superlattice nanostructures enhance reactivity, selectivity, and product upgrading. Finally, challenges and future opportunities are outlined. This work aims to advance photothermal catalysis by introducing ultra-fast, directional photon-electron-phonon transport channels and tailoring highly ordered surfaces and interfaces to steer solar-driven catalysis toward more efficient, selective, and value-oriented chemical transformations.
Ultracold atomic gases with uniform density can be created by flat-bottom optical traps. These gases provide an ideal platform to study many-body physics in a system that allows for simple connections with theoretical models and emulation of numerous effects from a wide range of fields of physics. In Earth-bound laboratories the trap sizes, number of species and states, as well as the range of physical effects are largely restricted by the adopted levitation technique. Homogeneous ultracold gases in microgravity simulators and space however offer an interesting perspective which is actively being pursued. To effectively make use of a gravity-compensated laboratory, realizing box potentials with large spatial extent enables access to previously inaccessible length scales and reduce finite-size and boundary effects. We present an approach based on two identical orthogonally aligned acousto-optic deflector setups to generate large time-averaged optical potentials with trapping volumes up to three orders of magnitude larger than conventional setups. These potentials follow power-law scalings with exponents of up to 152. We characterize the system and validate its performance through simulations of the mean-field ground state of a quantum gas, including dynamical excitations arising from the realistic time-dependent painting potentials. The implementation of this setup may open new directions at the interface with condensed matter, few-body Efimov physics or the exploration of critical, non-equilibrium phenomena.
Coal-fired power plants widely employ low-NOx combustion and selective catalytic reduction (SCR) techniques to mitigate NOx emissions. However, the former compromises combustion stability, while the latter consumes substantial NH3. In this study, the electrochemical conversion from nitric oxide (NO), the main component of NOx, to NH3 offers a sustainable strategy to simultaneously reduce pollutant emissions and recover valuable nitrogen resources. Herein, density functional theory (DFT) calculations are employed to systematically investigate the electrocatalytic NO reduction reaction (NORR) on two-dimensional BC2N monolayers doped with single transition-metal atoms (Ti, V, Cr, Mn, Fe, Co, Ni, Zr, Nb, Mo, Ru, Rh, and Pd) at B vacancy sites. All doped configurations are thermodynamically stable, with Co@BC2N further confirmed by ab initio molecular dynamics (AIMD) simulations to maintain structural integrity at 300 K. The calculated free energies indicate that NO adsorption dominates over the competing hydrogen evolution reaction (HER), ensuring high NO selectivity. Among all candidates, Co@BC2N exhibits the lowest rate-determining free energy barrier (0.10 eV) and favorable charge transfer between Co-3d and NO-2p orbitals, facilitating efficient NO activation and hydrogenation. This study highlights Co@BC2N as an outstanding active and stable NORR catalyst, offering atomic-level insight for the rational design of next-generation NO-to-NH3 electrocatalysts.
Building crystalline heterostructures with arbitrary material combinations, which has been often referred to as "anything-on-anything" integration, has remained a central challenge in materials science and device platforms. Membrane-based technologies provide a viable pathway toward this goal by decoupling thin-film growth from the resulting heterostructures. By isolating high-quality single-crystal layers from their host substrates, lift-off techniques bypass the intrinsic constraints imposed by substrate properties and enable the production of freestanding films and nanomembranes as previously inaccessible material building blocks. Among various lift-off strategies, mechanical lift-off is particularly attractive due to its wide applicability to virtually any material systems. However, the difficulty in manipulating cracks during mechanical lift-off, which determines the properties of exfoliated membranes, has limited the widespread adoption of the technology. Here, we provide key insights into the fundamental mechanics governing mechanical lift-off and discuss how recent breakthroughs in interface design, epitaxy techniques, and crack-guiding principles have enabled highly controlled spalling with atomic precision, scalability, and throughput. We then highlight how such innovations in mechanical lift-off technology, along with other emerging lift-off methods, have advanced membrane-based technologies and have opened new application spaces. Finally, we discuss remaining challenges not only in the lift-off processes themselves but also across the full process flow, outlining pathways toward the broader adoption of lift-off technologies for both fundamental scientific studies and advanced device platforms.
Fabrication of remote-controlled ion and molecule delivery devices with sub-nanomolar accuracy possesses enormous potential for application in futuristic areas like bio-sensing, molecular machines, energy storage/harvesting, separation, and purification. Here, we combined the excellent photothermal characteristics of oxidised multiwalled carbon nanotubes (o-CNT) with atomically thin 2D sheets of V2O5 (VO) to fabricate fluidic devices capable of discharging ions and molecules with sub-nanomolar accuracy. Remarkably, the delivery of ionic/molecular pulses can be remotely modulated through irradiation with IR light. The heterostructured fluidic membrane (HFM), fabricated as a bilayer membrane with distinctive regions of o-CNT and VO, can continuously deliver K+ ions at the rate of ∼0.58 nmol min-1. The ionic flow rate can be further tuned by modulating the thickness and composition of the HFM. Moreover, triangular HFM can discharge ions in nanomolar pluses (∼1.85 to 6.92 nmol per pulse) when remotely triggered through light of different intensities. Likewise, the intermixed fluidic membrane (IFM), fabricated through vacuum filtration of the homogeneous mixtures of the VO and o-CNT, can discharge biologically relevant molecules like tryptophan (Trp) and aspirin at a rate as low as 0.1 nmol min-1, which can be further modulated with light and heat.
This study developed a multiplexed aptasensor for the simultaneous detection of an endocrine-disrupting chemical (EDC) cluster-specifically, a mixture of bisphenol A (BPA), phthalic acid esters (PAEs), and nonylphenol (NP)-utilizing a three-dimensional DNA nanostructure framework. A tetrahedral 3D-DNA nanostructure was constructed, and its formation and morphology were verified via fluorescence-based assessments, gel electrophoresis, atomic force microscopy, and dynamic light scattering analysis. The reaction complex was fabricated by anchoring multiple aptamers to 3D-DNA nanostructures immobilized onto magnetic beads. This design enabled the quantitative detection of NP, PAEs, and BPA by measuring the signal decrease from DNA probes conjugated to specific quantum dots (QD565, QD605, and QD655). To validate the multiplexing capability, signal crosstalk and the dissociation of the QD-probe DNA upon target binding were examined. The quantitative performance of the multiplexed EDC detection was assessed in terms of selectivity against potential interferences, dynamic range, and limit of quantification (LOQ). The assay demonstrated dynamic ranges of 1-100 ng/mL for NP, 0.001-100 ng/mL for di-n-butyl phthalate (DBP), and 0.0005-100 ng/mL for BPA, with corresponding LOQs of 1 ng/mL (ppb), 1 pg/mL (ppt), and 0.5 pg/mL (ppt), respectively. Finally, the assay was applied to EDC mixtures spiked into real-world sample matrices (tap water and beverages), and recovery rates were evaluated to demonstrate its practical applicability.
Structural interpretation of accurate experimental data becomes increasingly challenging when nonlocal effects become important, particularly in delocalized open-shell systems and flexible hydrogen-bonded molecules. In these regimes, density functional methods, even when augmented by bond corrections, become insufficient because the transferable unit of the structural error is defined at an inappropriate scale. Here, we show that transferable corrections can instead be formulated by shifting from individual bonds to larger interaction-driven building blocks. These can be identified automatically and assigned, in a black-box fashion, to the appropriate rung of a general accuracy ladder, thereby establishing a direct correspondence between the scale at which errors arise and the level of electronic-structure theory required to describe them, while retaining an affordable computational cost. Across radicals, nonplanar aromatic systems, and flexible hydrogen-bonded networks, quantitative agreement between theory and experiment is recovered only when the physically relevant transferable unit of the error is matched to a commensurate level of theory and vibrational averaging is treated consistently. Accurate structures are therefore obtained not by uniformly increasing the level of theory but by matching the scale of error transferability with the appropriate level of description. Within this framework, multilevel strategies naturally emerge for large systems in which different regions are treated at different rungs of the ladder according to their structural complexity. This enables available experimental rotational constants to be reproduced within ∼0.1-0.2% at affordable cost, corresponding to root-mean-square deviations of atomic positions on the order of 1-2 × 10-3 Å.
The oxygen evolution reaction (OER) is central to water electrolysis for green hydrogen production, where the development of efficient, stable, and cost-effective catalysts remains a key challenge. Ruthenium (Ru)-based catalysts are highly promising for OER, yet their practical deployment is limited by poor stability and low atomic utilization. Here, we report a fluorination-assisted loading strategy to fabricate an amorphous@crystalline composite catalyst, in which amorphous RuOx active species are anchored onto a fluorine-modified Ruddlesden-Popper (R-P) perovskite support (La1.2Sr0.8Ni0.6Fe0.4O4+δFy) to form the catalyst denoted as RuOx@LSNF-F. Fluorination induces robust metal-support interactions via electronegativity differences, stabilizing Ru4+ to balance the activity and stability while constructing an ordered hydrogen-bond network and optimizing the surface wettability. Electrochemical tests in 1.0 M KOH confirm the exceptional OER performance of RuOx@LSNF-F, which achieves an overpotential of 287 mV at 10 mA cm-2, a Tafel slope of 75 mV dec-1, and a double-layer capacitance of 3.7 mF cm-2 while maintaining stable operation for over 100 h, outperforming its non-fluorinated counterpart, the pristine support, and other benchmark Ru-based catalysts. Mechanistic studies reveal that the exceptional performance arises from the synergistic coupling of the adsorbate evolution mechanism (AEM) and lattice oxygen mechanism (LOM), enabled by the fluorination-engineered interface, while a hydrogen bond network is enhanced on the amorphous RuOx layer to accelerate deprotonation and improve the surface wettability. This work presents a strategy for designing high-performance perovskite-based OER catalysts and advances the understanding of their structure-activity relationships.
Malignant peripheral nerve sheath tumors (MPNST) exhibit pronounced alterations in lipid organization that contribute to tumor aggressiveness and resistance to radiotherapy. In this work, we combine atomic force microscopy-infrared spectroscopy (AFM-IR) and fluorescence imaging to investigate nanoscale lipid remodeling in Schwann and MPNST cells exposed to cannabidiol (CBD) and ionizing radiation, while introducing a new semiquantitative strategy for AFM-IR image analysis. Conventional band-based AFM-IR spectroscopy was first employed to identify characteristic biochemical signatures in the perinuclear region, revealing CBD- and irradiation-dependent modifications of phospholipids (1260 cm-1-1240 cm-1) and cholesteryl esters, monitored via the ester carbonyl band at 1740 cm-1. These spectral changes provided a biochemical basis for further nanoscale analysis, but were restricted to intensity-based interpretation. To overcome this limitation, we introduce, for the first time, a pixel-based AFM-IR semi-quantification framework that converts nanospectroscopic maps into statistically robust biochemical metrics. High-resolution AFM-IR images were processed to extract pixel-resolved ester-specific signals, enabling semi-quantitative determination of both the average cholesteryl ester signal intensity and the nanoscale surface area occupied by ester-rich domains. Statistical evaluation using ANOVA with Tukey's post-hoc test allowed direct comparison of lipid redistribution across experimental conditions. Application of this framework revealed distinct nanoscale patterns of cholesteryl ester remodeling in Schwann versus MPNST cells under CBD and irradiation, including pronounced spatial reorganization that was not evident from spectral intensities alone. Importantly, the AFM-IR-derived spatial metrics were independently validated by fluorescence lipid droplet staining, demonstrating similar trends between nanoscale infrared measurements and cellular lipid abundance. In parallel, AFM-IR analysis of the Amide I and II regions uncovered CBD-dependent modulation of protein secondary structure, highlighting differential responses between normal and malignant cells. Overall, this study establishes a transferable, pixel-based AFM-IR analysis strategy for nanoscale biochemical semi-quantification and demonstrates its utility in resolving lipid organization and remodeling in complex biological systems.
The elucidation of molecular structures and properties typically relies on the integration of multiple spectroscopic techniques, which provide rich structural information from complementary perspectives. Although the synergistic analysis of multi-modal spectra can significantly enhance the accuracy of structural identification, its practical application is severely constrained by the scarcity of high-quality experimental spectral data. In particular, comprehensive datasets encompassing multiple spectroscopic modalities for the same molecule are exceedingly rare. To overcome this data bottleneck, we developed MolSpectra, a universal deep learning framework capable of predicting multi-modal spectra directly from molecular SMILES representations. MolSpectra provides a unified molecular representation architecture, enabling the synergistic input of experimental metadata, such as physical phase states, alongside molecular structural information into an end-to-end model. This framework eliminates the need to modify the core model architecture; it can predict various types of spectra simply by adjusting configuration files. We systematically evaluated MolSpectra on six datasets covering four spectroscopic techniques. These include experimental infrared (IR) spectral datasets from the NIST and Chemotion databases, a simulated IR spectral benchmark dataset based on the USPTO reaction corpus, an ultraviolet-visible (UV-Vis) spectral dataset from the NIST database, an electron ionization mass spectrometry (EI-MS) dataset from the commercial NIST 23 database, and nuclear magnetic resonance (NMR) datasets from nmrshiftdb2 and HMDB (Human Metabolome Database) for chemical shift and full-spectrum predictions, respectively. Experimental results demonstrate that MolSpectra comprehensively outperforms existing baseline models: under strict InChIKey-based splitting, it achieves maximum cosine similarities of 0.916 for IR (NIST experimental dataset) and 0.861 for UV-Vis (NIST-UV dataset) predictions; for EI-MS prediction, it reaches a maximum cosine similarity of 0.880 (NIST 23 dataset) and a Top-1 accuracy of 0.530 in spectral library matching. Regarding NMR, the framework achieves mean absolute errors (MAEs) as low as 0.153 ppm and 1.176 ppm for 1 H/ 13 C chemical shift predictions on the nmrshiftdb2 subset, respectively. MolSpectra takes SMILES strings as input and constructs molecular graphs using RDKit. The model employs a Message Passing Graph Neural Network (MPGNN) equipped with Sequential Signal Mixing Aggregation (SSMA) to model local and medium-range atomic interactions. To capture long-range dependencies and local structural details, the model integrates a Hierarchical Distance Structural Encoding (HDSE) and a structurally biased attention mechanism within an enhanced Transformer architecture. Node features are updated iteratively, and flexible prediction heads are employed: a node-level prediction head is used for NMR chemical shifts, while graph-level prediction heads with permutation-invariant readout functions are utilized for molecular-level spectra such as IR, UV-Vis, and EI-MS. Furthermore, the framework supports the encoding of metadata, such as physical phase states, to accelerate model convergence and improve generalization capabilities. In MolSpectra, each spectroscopic modality is trained using an independent model instance that shares the same backbone architecture, with a modality-specific prediction head adapted to the corresponding spectral dimensionality. The source code is publicly available at https://github.com/lzjforyou/MolSpectra .
Voltage dependent anion channels (VDACs 1, 2 and 3) in the outer mitochondrial membrane control the flux of anions and oxidizable substrates that sustain mitochondrial metabolism. NADH closes VDAC by binding to a pocket, conserved in all isoforms, located in the inner wall of the channel. Previously, we identified the small molecule SC18 that targets the NADH-binding pocket of VDAC1 employing computational analysis. Here, we explored the interaction between SC18 and VDAC1 using High-resolution Nuclear Magnetic Resonance spectroscopy and Molecular Dynamics simulations. Atomically resolved data precisely confirmed the computational results, showing that SC18 binds to a site on VDAC1 that partially overlaps with the NADH binding pocket. SC18, in the presence of NADH blocked the conductance of VDAC1 reconstituted in lipid bilayers. To determine the metabolic effect of SC18, we combined readouts of mitochondrial metabolism and glycolysis with functional metabolomics and proteomics. Short-term treatment with SC18 inhibited mitochondrial metabolism and ATP production. Treatment over 24 h and 48 h further reduced mitochondrial uptake of pyruvate and glutamine, utilization of tricarboxylic acid cycle intermediates, as well as lipid, DNA and amino acid synthesis. Concomitant with the inhibition of mitochondrial metabolism, cellular uptake of glucose and glutamine increased in parallel with augmented lactate release. These results indicate that compensatory enhanced glycolysis sustains ATP production after impaired mitochondrial function induced by SC18 blockage of VDAC1. Our work set a mechanistic foundation for VDAC1 inhibition as a novel strategy to target and reprogram cancer metabolism through modulation of the biosynthetic ability of mitochondria.