Large-scale in-silico bacterial cryptochrome/photolyase screening identifies an extremophile radical-pair candidate with expanded flavin–tryptophan wiring

Technical Summary: Large-scale in-silico bacterial cryptochrome/photolyase screening identifies an extremophile radical-pair candidate with expanded flavin–tryptophan wiring

Problem Statement
Cryptochrome/photolyase-family (CPF) proteins are flavin adenine dinucleotide (FAD)-dependent blue-light proteins involved in DNA repair and light sensing. While radical-pair photochemistry and magnetic-field effects have been experimentally demonstrated in the model organism Escherichia coli cyclobutane pyrimidine dimer (CPD) photolyase, the broader diversity of bacterial CPF architectures remains poorly explored. Current knowledge is limited to a small number of model systems, and it is unknown how widely FAD-proximal aromatic architectures—essential for efficient flavin-to-tryptophan electron transfer—occur across non-model bacterial proteins. Standard annotation based on family names is insufficient to predict whether a specific protein possesses the structural environment required for productive radical-pair chemistry.

Methodology
The study employed a multi-stage in-silico screening pipeline to identify bacterial CPF candidates with architectures suitable for radical-pair photochemistry:

1. Data Retrieval and Filtering: The authors retrieved 22,081 bacterial protein sequences from the UniProt Knowledgebase (UniProtKB) belonging to four CPF groups: CPD/DNA photolyases, cryptochrome-like proteins, cryptochrome DASH (Cry-DASH) proteins, and 6–4 photolyases.
2. Sequence Feature Prioritization: Using the experimentally characterized E. coli CPD photolyase (P00914) as a benchmark (15 tryptophans, 48 total aromatic residues), the dataset was filtered to retain candidates with comparable or enriched tryptophan and aromatic residue counts.
3. Deduplication and Domain Validation: The candidate set was deduplicated at the species level to avoid over-representation, resulting in 5,060 candidates. A domain-support filter was applied, yielding a final dataset of 5,016 domain-supported, species-deduplicated proteins.
4. Structural and Graph-Based Analysis: Predicted structures for representative candidates were retrieved from the AlphaFold Protein Structure Database. FAD coordinates were transferred from experimental reference structures onto these AlphaFold models.
   • Proximity Analysis: Distances were calculated between the transferred FAD isoalloxazine ring and aromatic residues (tryptophan, tyrosine, phenylalanine).
   • Wiring Analysis: A graph-based topology analysis was performed to map connected networks of aromatic residues. Nodes represented FAD and aromatic side chains; edges were drawn if centroid distances were ≤6 Å. Metrics included the number of FAD-connected tryptophans and the number of simple FAD-to-tryptophan paths within four graph steps.
5. Phylogenetic Context: Maximum-likelihood phylogenetic analysis was conducted to ensure selected candidates represented broad CPF diversity and grouped correctly within their respective families.

Key Results
The screening identified a broad reservoir of bacterial CPF proteins with enriched aromatic content. The analysis highlighted A0A6M1RS91, a CPD photolyase from the thermophilic bacterium Limisphaera ngatamarikiensis, as the strongest candidate.

• Sequence and Structural Conservation: A0A6M1RS91 retains the conserved CPD photolyase fold (TM-score 0.903 vs. E. coli P00914) but exhibits significantly higher aromatic content. It contains 31 total tryptophans compared to 15 in the E. coli benchmark.
• Expanded FAD-Proximal Network: Relative to the E. coli benchmark, A0A6M1RS91 shows a dense aromatic environment around the predicted FAD site:
  • 12 tryptophans within 12 Å of FAD (vs. 7 in E. coli).
  • 24 FAD-connected tryptophans (vs. 9 in E. coli).
  • 73 simple FAD-to-tryptophan paths (vs. 33 in E. coli).
• Optimal Pathway: The best predicted electron-transfer route in A0A6M1RS91 is FAD → F373 → W389 → W323 → W372.
• Family-Matched Comparison: When compared against family-matched references rather than just the E. coli benchmark, A0A6M1RS91 remained the clearest outlier, showing substantial expansion in path topology metrics (+40 simple paths, +30 mixed aromatic paths) that was not observed in other high-tryptophan candidates from different CPF subfamilies.

Significance and Claims
The paper claims to move bacterial radical-pair discovery beyond the limited scope of model photolyases by providing a structure-informed route to identify non-model candidates. The primary significance of this work is the identification of A0A6M1RS91 as a focused experimental target.

The authors posit that the expanded FAD-connected aromatic topology in A0A6M1RS91 provides a mechanistic hypothesis: the protein may possess alternative or additional aromatic routes that could influence flavin photoreduction, radical migration, or back-electron-transfer kinetics. The study does not claim to have proven radical-pair formation or magnetic-field sensitivity in this protein; rather, it identifies the protein's structural features as highly suggestive of such capabilities. The authors explicitly state that the ecological origin of L. ngatamarikiensis (thermophilic, geothermal mud) supports the biological plausibility of robust flavin-dependent photochemistry, but they emphasize that the mechanistic hypothesis must be validated through future in-vitro photochemical experiments.

The work serves as a "focused shortlist" for testing flavin photoreduction, flavin–tryptophan radical formation, and magnetic-field-sensitive photochemistry in a bacterial system with an expanded aromatic network.