Re‐engineering of a carotenoid‐binding protein based on NMR structure DOI
Andrey Nikolaev,

Daria A. Lunegova,

Roman I. Raevskii

и другие.

Protein Science, Год журнала: 2024, Номер 33(12)

Опубликована: Ноя. 16, 2024

Abstract Recently, a number of message passing neural network (MPNN)‐based methods have been introduced that, based on backbone atom coordinates, efficiently recover native amino acid sequences proteins and predict modifications that result in better expressing, more soluble, stable variants. However, usually, X‐ray structures, or artificial structures generated by algorithms trained were employed to define target conformations. Here, we show commonly used ProteinMPNN SolubleMPNN display low sequence recovery determined using NMR. We subsequently propose computational approach successfully apply re‐engineer AstaP, protein natively binds large hydrophobic ligand astaxanthin (C 40 H 52 O 4 ), for which only structure NMR is currently available. The engineered variants, designated NeuroAstaP, are 51 shorter than the 22 kDa parent protein, 38%–42% identity it, exhibit good yields, expressed mostly monomeric form, demonstrate efficient binding carotenoids vitro cells. Altogether, our work further tests limits machine learning engineering paves way MPNN‐based modification NMR‐derived structures.

Язык: Английский

How well do contextual protein encodings learn structure, function, and evolutionary context? DOI
Sai Pooja Mahajan, Fátima A. Dávila-Hernández, Jeffrey A. Ruffolo

и другие.

Cell Systems, Год журнала: 2025, Номер 16(3), С. 101201 - 101201

Опубликована: Март 1, 2025

Язык: Английский

Процитировано

0
Language models for protein design DOI
Jin Seop Lee, Osama Abdin, Philip M. Kim

и другие.

Current Opinion in Structural Biology, Год журнала: 2025, Номер 92, С. 103027 - 103027

Опубликована: Март 6, 2025

Язык: Английский

Процитировано

0
Scaling unlocks broader generation and deeper functional understanding of proteins DOI Creative Commons

Aadyot Bhatnagar,

Sarthak Jain,

Joel Beazer

и другие.

bioRxiv (Cold Spring Harbor Laboratory), Год журнала: 2025, Номер unknown

Опубликована: Апрель 16, 2025

Abstract Generative protein language models (PLMs) are powerful tools for designing proteins purpose-built to solve problems in medicine, agriculture, and industrial processes. Recent work has trained ever larger models, but there been little systematic study of the optimal training distributions influence model scale on sequences generated by PLMs. We introduce ProGen3 family sparse generative PLMs, we develop compute-optimal scaling laws up a 46B-parameter pre-trained 1.5T amino acid tokens. ProGen3’s pre-training data is sampled from an optimized distribution over Profluent Protein Atlas v1, carefully curated dataset 3.4B full-length proteins. evaluate first time wet lab find that generate viable much wider diversity families. Finally, both computationally experimentally more responsive alignment with laboratory data, resulting improved fitness prediction sequence generation capabilities. These results indicate PLMs like ProGen3-46B larger, well-curated datasets foundation push frontier design. 1

Язык: Английский

Процитировано

0
Tracing the stepwise Darwinian evolution of a plant halogenase DOI
Colin Y. Kim, David W. Kastner, Andrew J. Mitchell

и другие.

bioRxiv (Cold Spring Harbor Laboratory), Год журнала: 2024, Номер unknown

Опубликована: Дек. 11, 2024

Abstract Halogenation chemistry is rare in plant metabolism, with the chloroalkaloid acutumine produced by Menispermaceae species being only well characterized example, involving a specialized dechloroacutumine halogenase (DAH) from iron(II)- and 2-oxoglutarate-dependent dioxygenase (2ODD) superfamily. While DAH presumed to have evolved an ancestral 2ODD enzyme, broader question of how new enzymes arise through Darwinian processes, such as birth Menispermaceae, remains fundamental challenge understanding metabolic evolution. Here, we investigate DAH’s evolutionary trajectory using chromosomal-level genome assembly Menispermum canadense . By analyzing genomic context M. syntenic regions related plants, show that tandem duplication flavonol synthase ( FLS ) gene, followed series neofunctionalization gene loss events. Through structural modeling, molecular dynamics simulations, site-directed mutagenesis, identify residue changes enabling transition DAH. This functional switch required traversing complex landscape where adaptive peaks were separated deep fitness valleys. Our work illustrates enzymatic functions can lineage-specific pathways gradually reshape active site architecture permissive mutations, ultimately mechanism-switching mutations establish novel catalytic activities.

Язык: Английский

Процитировано

1
AI-generated small binder improves prime editing DOI Creative Commons
Ju-Chan Park, Heesoo Uhm, Yong Woo Kim

и другие.

bioRxiv (Cold Spring Harbor Laboratory), Год журнала: 2024, Номер unknown

Опубликована: Сен. 14, 2024

Abstract The prime editing 2 (PE2) system comprises a nickase Cas9 fused to reverse transcriptase utilizing guide RNA (pegRNA) introduce desired mutations at target genomic sites. However, the PE efficiency is limited by mismatch repair (MMR) that excises DNA strand containing edits. Thus, inhibiting key components of MMR complex through transient expression dominant negative MLH1 (MLH1dn) exhibited approximately 7.7-fold increase in over PE2, generating PE4. Herein, generative artificial intelligence (AI) technologies, RFdiffusion and AlphaFold 3, we ultimately generated de novo small binder (named MLH1-SB), which bind dimeric interface PMS2 disrupt formation components. MLH1-SB’s size (82 amino acids) allowed it be integrated into pre-existing architectures via 2A system, creating novel PE-SB platform. Resultantly, incorporating MLH1-SB PE7, have developed an improved architecture called PE7-SB, demonstrates highest date (29.4-fold PE2 2.4-fold PE7 HeLa cells), providing insight AI technologies will boost up improvement genome tools.

Язык: Английский

Процитировано

0
Ai-Generated Small Binder Improves Prime Editing DOI
Ju-Chan Park, Heesoo Uhm, Yong Woo Kim

и другие.

Опубликована: Янв. 1, 2024

Язык: Английский

Процитировано

0
Re‐engineering of a carotenoid‐binding protein based on NMR structure DOI
Andrey Nikolaev,

Daria A. Lunegova,

Roman I. Raevskii

и другие.

Protein Science, Год журнала: 2024, Номер 33(12)

Опубликована: Ноя. 16, 2024

Abstract Recently, a number of message passing neural network (MPNN)‐based methods have been introduced that, based on backbone atom coordinates, efficiently recover native amino acid sequences proteins and predict modifications that result in better expressing, more soluble, stable variants. However, usually, X‐ray structures, or artificial structures generated by algorithms trained were employed to define target conformations. Here, we show commonly used ProteinMPNN SolubleMPNN display low sequence recovery determined using NMR. We subsequently propose computational approach successfully apply re‐engineer AstaP, protein natively binds large hydrophobic ligand astaxanthin (C 40 H 52 O 4 ), for which only structure NMR is currently available. The engineered variants, designated NeuroAstaP, are 51 shorter than the 22 kDa parent protein, 38%–42% identity it, exhibit good yields, expressed mostly monomeric form, demonstrate efficient binding carotenoids vitro cells. Altogether, our work further tests limits machine learning engineering paves way MPNN‐based modification NMR‐derived structures.

Язык: Английский

Процитировано

0