Atsuhiro Tomita
Atsuhiro Tomita has a total of 45 co-authors. Here are the top co-authors by number of shared publications:
- Keitaro Yamashita
Structural Biology Division, Research Center for Advanced Science and Technology, The University of Tokyo, Tokyo, Japan
5 shared publications - Tsukasa Kusakizako
Department of Biological Sciences, Graduate School of Science, The University of Tokyo, Tokyo, Japan
5 shared publications - Osamu Nureki
Department of Biological Sciences, Graduate School of Science, The University of Tokyo, Tokyo, Japan
4 shared publications Inoue A
3 shared publications- Kazuhiro Kobayashi
Research Center for Advanced Science and Technology, The University of Tokyo, Meguro, Tokyo, Japan
3 shared publications Nureki O
3 shared publications
Latest Publications
- Open Access
- 19 December 2025
- Published by Springer Nature in Communications Chemistry
- Vol. 9 (1)
Generalizable compound protein interaction prediction with a model incorporating protein structure aware and compound property aware language model representations
Compound–protein interaction (CPI) prediction plays a crucial role in drug discovery by aiding the identification of binding and affinities between small molecules and proteins. Current deep learning models rely heavily on sequence-based representations and suffer from a lack of labeled data, which restricts their accuracy and generalizability. To overcome these challenges, we propose GenSPARC (a model with Gen eralized S tructure- and P roperty- A ware R epresentations of protein and chemical language models for C PI prediction), a deep learning model that leverages structure-aware protein representations derived from AlphaFold2 predictions and FoldSeek’s three-dimensional interaction alphabet. Compound features were extracted using graph convolutional networks and a pretrained chemical language model, thereby ensuring comprehensive multimodal representation. An attention mechanism further enhanced interaction modeling by capturing intricate binding patterns. GenSPARC was validated successfully with multiple CPI benchmark datasets, demonstrating strong generalizability across challenging data splits and competitive results in virtual screening tasks. Therefore, GenSPARC will substantially advance artificial intelligence-driven drug discovery.
- 19 December 2025
- Published by Springer Nature in Communications Chemistry
- Vol. 9 (1)
Open AccessGeneralizable compound protein interaction prediction with a model incorporating protein structure aware and compound property aware language model representations
Compound–protein interaction (CPI) prediction plays a crucial role in drug discovery by aiding the identification of binding and affinities between small molecules and proteins. Current deep learning models rely heavily on sequence-based representations and suffer from a lack of labeled data, which restricts their accuracy and generalizability. To overcome these challenges, we propose GenSPARC (a model with Gen eralized S tructure- and P roperty- A ware R epresentations of protein and chemical language models for C PI prediction), a deep learning model that leverages structure-aware protein representations derived from AlphaFold2 predictions and FoldSeek’s three-dimensional interaction alphabet. Compound features were extracted using graph convolutional networks and a pretrained chemical language model, thereby ensuring comprehensive multimodal representation. An attention mechanism further enhanced interaction modeling by capturing intricate binding patterns. GenSPARC was validated successfully with multiple CPI benchmark datasets, demonstrating strong generalizability across challenging data splits and competitive results in virtual screening tasks. Therefore, GenSPARC will substantially advance artificial intelligence-driven drug discovery.
- Open Access
- 7 November 2025
- Published by Springer Nature in Research Square
Structural Insights into the Broad Substrate Recognition Mechanism of Human ABCD3
Human peroxisomal ABC subfamily D transporters, ABCD1, ABCD2, and ABCD3, transport acyl-CoA molecules and bile acid precursors, and thereby participate in fatty acid metabolism and cholesterol homeostasis. ABCD3 can recognize a wide range of substrates, including polyunsaturated fatty acyl-CoAs, dicarboxylic acyl-CoAs, and ATP molecules. However, the structural mechanism of this broad substrate recognition by ABCD3 remains largely unknown. Here we report five cryo-electron microscopy structures of ABCD3: the apo state, the eicosapentaenoyl coenzyme A (C20:5-CoA)-bound state, the cholesteryl hemisuccinate (CHS)-bound state, the ATP-bound inward-facing state, and the ATP-bound outward-facing state. In the C20:5-CoA-bound state, the CoA moiety is recognized by ABCD3 in a similar manner to ABCD1, but the long, unsaturated bent acyl chain moiety is accommodated in a large cavity between protomers, distinct from the saturated straight acyl chain moiety-binding site observed in ABCD1. Furthermore, we found that the structurally and chemically different substrates, ATP and CHS, bind to distinct hydrophilic and hydrophobic sites, respectively, in the transmembrane domain of ABCD3. Combined with molecular dynamics simulations, these structures, encompassing both inward- and outward-facing conformations, reveal the molecular mechanism by which ABCD3 recognizes and transports a broad range of substrates.Main text
- 7 November 2025
- Published by Springer Nature in Research Square
Open AccessStructural Insights into the Broad Substrate Recognition Mechanism of Human ABCD3
Human peroxisomal ABC subfamily D transporters, ABCD1, ABCD2, and ABCD3, transport acyl-CoA molecules and bile acid precursors, and thereby participate in fatty acid metabolism and cholesterol homeostasis. ABCD3 can recognize a wide range of substrates, including polyunsaturated fatty acyl-CoAs, dicarboxylic acyl-CoAs, and ATP molecules. However, the structural mechanism of this broad substrate recognition by ABCD3 remains largely unknown. Here we report five cryo-electron microscopy structures of ABCD3: the apo state, the eicosapentaenoyl coenzyme A (C20:5-CoA)-bound state, the cholesteryl hemisuccinate (CHS)-bound state, the ATP-bound inward-facing state, and the ATP-bound outward-facing state. In the C20:5-CoA-bound state, the CoA moiety is recognized by ABCD3 in a similar manner to ABCD1, but the long, unsaturated bent acyl chain moiety is accommodated in a large cavity between protomers, distinct from the saturated straight acyl chain moiety-binding site observed in ABCD1. Furthermore, we found that the structurally and chemically different substrates, ATP and CHS, bind to distinct hydrophilic and hydrophobic sites, respectively, in the transmembrane domain of ABCD3. Combined with molecular dynamics simulations, these structures, encompassing both inward- and outward-facing conformations, reveal the molecular mechanism by which ABCD3 recognizes and transports a broad range of substrates.Main text
- Open Access
- 8 May 2025
- Published by Springer Nature in Research Square
GenSPARC: Generalized Structure- and Property-Aware Representations of Language Models for Compound-Protein Interaction Prediction
Compound-protein interaction (CPI) prediction plays a crucial role in drug discovery by aiding the identification of binding and affinities between small molecules and proteins. Current deep learning models rely heavily on sequence-based representations and suffer from a lack of labeled data, which restricts their accuracy and generalizability. To overcome these challenges, we propose GenSPARC ( Gen eralized S tructure and P roperty A ware R epresentation for C PI prediction), a deep learning model that leverages structure-aware protein representations derived from AlphaFold2 predictions and Foldseek’s 3D interaction alphabet. Compound features were extracted using graph convolutional networks and a pretrained chemical language model, thereby ensuring comprehensive multimodal representation. A novel attention mechanism further enhanced interaction modeling by capturing intricate binding patterns. GenSPARC was validated successfully with multiple CPI benchmark datasets, demonstrating strong generalizability across challenging data splits and competitive results in virtual screening tasks. Therefore, GenSPARC will substantially advance artificial intelligence-driven drug discovery.
- 8 May 2025
- Published by Springer Nature in Research Square
Open AccessGenSPARC: Generalized Structure- and Property-Aware Representations of Language Models for Compound-Protein Interaction Prediction
Compound-protein interaction (CPI) prediction plays a crucial role in drug discovery by aiding the identification of binding and affinities between small molecules and proteins. Current deep learning models rely heavily on sequence-based representations and suffer from a lack of labeled data, which restricts their accuracy and generalizability. To overcome these challenges, we propose GenSPARC ( Gen eralized S tructure and P roperty A ware R epresentation for C PI prediction), a deep learning model that leverages structure-aware protein representations derived from AlphaFold2 predictions and Foldseek’s 3D interaction alphabet. Compound features were extracted using graph convolutional networks and a pretrained chemical language model, thereby ensuring comprehensive multimodal representation. A novel attention mechanism further enhanced interaction modeling by capturing intricate binding patterns. GenSPARC was validated successfully with multiple CPI benchmark datasets, demonstrating strong generalizability across challenging data splits and competitive results in virtual screening tasks. Therefore, GenSPARC will substantially advance artificial intelligence-driven drug discovery.
- Open Access
- 7 June 2023
- Published by Springer Nature in Nature
- Vol. 618 (7967) , 1085-1093
Class B1 GPCR activation by an intracellular agonist
G protein-coupled receptors (GPCRs) generally accommodate specific ligands in the orthosteric-binding pockets. Ligand binding triggers a receptor allosteric conformational change that leads to the activation of intracellular transducers, G proteins and β-arrestins. Because these signals often induce adverse effects, the selective activation mechanism for each transducer must be elucidated. Thus, many orthosteric-biased agonists have been developed, and intracellular-biased agonists have recently attracted broad interest. These agonists bind within the receptor intracellular cavity and preferentially tune the specific signalling pathway over other signalling pathways, without allosteric rearrangement of the receptor from the extracellular side1–3. However, only antagonist-bound structures are currently available1,4–6, and there is no evidence to support that biased agonist binding occurs within the intracellular cavity. This limits the comprehension of intracellular-biased agonism and potential drug development. Here we report the cryogenic electron microscopy structure of a complex of Gs and the human parathyroid hormone type 1 receptor (PTH1R) bound to a PTH1R agonist, PCO371. PCO371 binds within an intracellular pocket of PTH1R and directly interacts with Gs. The PCO371-binding mode rearranges the intracellular region towards the active conformation without extracellularly induced allosteric signal propagation. PCO371 stabilizes the significantly outward-bent conformation of transmembrane helix 6, which facilitates binding to G proteins rather than β-arrestins. Furthermore, PCO371 binds within the highly conserved intracellular pocket, activating 7 out of the 15 class B1 GPCRs. Our study identifies a new and conserved intracellular agonist-binding pocket and provides evidence of a biased signalling mechanism that targets the receptor–transducer interface.
- 7 June 2023
- Published by Springer Nature in Nature
- Vol. 618 (7967) , 1085-1093
Open AccessClass B1 GPCR activation by an intracellular agonist
G protein-coupled receptors (GPCRs) generally accommodate specific ligands in the orthosteric-binding pockets. Ligand binding triggers a receptor allosteric conformational change that leads to the activation of intracellular transducers, G proteins and β-arrestins. Because these signals often induce adverse effects, the selective activation mechanism for each transducer must be elucidated. Thus, many orthosteric-biased agonists have been developed, and intracellular-biased agonists have recently attracted broad interest. These agonists bind within the receptor intracellular cavity and preferentially tune the specific signalling pathway over other signalling pathways, without allosteric rearrangement of the receptor from the extracellular side1–3. However, only antagonist-bound structures are currently available1,4–6, and there is no evidence to support that biased agonist binding occurs within the intracellular cavity. This limits the comprehension of intracellular-biased agonism and potential drug development. Here we report the cryogenic electron microscopy structure of a complex of Gs and the human parathyroid hormone type 1 receptor (PTH1R) bound to a PTH1R agonist, PCO371. PCO371 binds within an intracellular pocket of PTH1R and directly interacts with Gs. The PCO371-binding mode rearranges the intracellular region towards the active conformation without extracellularly induced allosteric signal propagation. PCO371 stabilizes the significantly outward-bent conformation of transmembrane helix 6, which facilitates binding to G proteins rather than β-arrestins. Furthermore, PCO371 binds within the highly conserved intracellular pocket, activating 7 out of the 15 class B1 GPCRs. Our study identifies a new and conserved intracellular agonist-binding pocket and provides evidence of a biased signalling mechanism that targets the receptor–transducer interface.
- Open Access
- 7 December 2022
- Published by EMBL-EBI
Cryo-EM structures of the β3 adrenergic receptor bound to solabegron and isoproterenol
EMPIAR, the Electron Microscopy Public Image Archive centered at EMBL-EBI, is a public resource for raw electron microscopy images related to EMDB, contains micrographs, particle sets and tilt-series.
- 7 December 2022
- Published by EMBL-EBI
Open AccessCryo-EM structures of the β3 adrenergic receptor bound to solabegron and isoproterenol
EMPIAR, the Electron Microscopy Public Image Archive centered at EMBL-EBI, is a public resource for raw electron microscopy images related to EMDB, contains micrographs, particle sets and tilt-series.
- Open Access
- 7 December 2022
- Published by EMBL-EBI
Dog beta3 adrenergic receptor bound to mirabegron in complex with a miniGs heterotrimer
EMPIAR, the Electron Microscopy Public Image Archive centered at EMBL-EBI, is a public resource for raw electron microscopy images related to EMDB, contains micrographs, particle sets and tilt-series.
- 7 December 2022
- Published by EMBL-EBI
Open AccessDog beta3 adrenergic receptor bound to mirabegron in complex with a miniGs heterotrimer
EMPIAR, the Electron Microscopy Public Image Archive centered at EMBL-EBI, is a public resource for raw electron microscopy images related to EMDB, contains micrographs, particle sets and tilt-series.
- Open Access
- 31 March 2022
- Published by Elsevier in Molecular Cell
- Vol. 82 (10) , 1865-1877.e4
Structure of the type V-C CRISPR-Cas effector enzyme
- 31 March 2022
- Published by Elsevier in Molecular Cell
- Vol. 82 (10) , 1865-1877.e4
Open AccessStructure of the type V-C CRISPR-Cas effector enzyme
- Open Access
- 17 February 2022
- Published by EMBL-EBI
Endogenous ligand recognition and structural transition of a human PTH receptor.
EMPIAR, the Electron Microscopy Public Image Archive centered at EMBL-EBI, is a public resource for raw electron microscopy images related to EMDB, contains micrographs, particle sets and tilt-series.
- 17 February 2022
- Published by EMBL-EBI
Open AccessEndogenous ligand recognition and structural transition of a human PTH receptor.
EMPIAR, the Electron Microscopy Public Image Archive centered at EMBL-EBI, is a public resource for raw electron microscopy images related to EMDB, contains micrographs, particle sets and tilt-series.
- Open Access
- 18 November 2021
- Published by Elsevier in Molecular Cell
- Vol. 81 (23) , 4799-4809.e5
Cryo-EM reveals mechanistic insights into lipid-facilitated polyamine export by human ATP13A2
- 18 November 2021
- Published by Elsevier in Molecular Cell
- Vol. 81 (23) , 4799-4809.e5
Open AccessCryo-EM reveals mechanistic insights into lipid-facilitated polyamine export by human ATP13A2