The cyanobacterial circadian clock is driven by a core post-translational oscillator composed of KaiA, KaiB, and KaiC, with the histidine kinase (HK), SasA, acting as the primary transducer of timing signals for downstream gene expression. Despite its central role, the structural architecture, oligomeric assembly of full-length SasA, and its stoichiometric interactions with the KaiC hexamer remain unclear. Herein, we determined the crystal structure of the SasA C-terminal ATPase domain (SasAC) from Thermosynechococcus vestitus BP-1 at 2.31 & Aring; resolution. SasAC adopts a canonical GHKL family fold but features a distinct G2-box (ATP lid) orientation that differs markedly from other prototypical HKs. Mass photometry (MP) analysis of wild-type SasA and a truncated sensory domain mutant revealed that SasA undergoes concentration-dependent, even-numbered oligomerization (monomer-dimer-tetramer), establishing that SasA is predominantly a dimeric HK rather than the previously proposed trimeric or hexameric forms. Interaction assays demonstrated that the KaiC hexamer formed discrete complexes containing two or four SasA molecules, corresponding to 2:6 and 4:6 (SasA:KaiC) stoichiometries. These results provide a structural and biophysical framework for understanding SasA assembly with KaiC in the cyanobacterial circadian output pathway.
The hydrolytic breakdown of cellobiose into glucose, catalyzed by beta-glucosidases, is the last and rate-limiting step in cellulose saccharification for producing fermentable glucose in the bioethanol industry. This limitation arises because beta-glucosidase activity is inhibited by factors such as temperature, pH, and glucose accumulation in reactors. Enzyme inactivation leads to the buildup of cello-oligosaccharides, which, in turn, inhibit upstream cellulases. Therefore, glucose-tolerant beta-glucosidases are preferred for the formulation of industrial cellulase cocktails. In this study, we have recombinantly expressed, purified, and biochemically characterized a beta-glucosidase from the cellulolytic fungus Fusarium odoratissimum (FoBgl-WT). FoBgl-WT exhibits optimal cellobiose hydrolysis over a broad pH range (4.5-7.5), an important and industrially desirable property for its application in bioreactors. However, the glucose tolerance of FoBgl-WT was similar to 0.56 M. Structure-based analyses were carried out to map the residues lining the active site of FoBgl, and their roles in stabilizing the product glucose (or even the substrate, cellobiose) were elucidated through a series of site-specific mutations, followed by biochemical characterization of the resulting FoBgl mutants. Among all the mutants generated, FoBgl-K256I and FoBgl-K256W exhibit similar to 2-2.5-fold enhancement in glucose tolerance. Further, the FoBgl-K256W variant displayed a two-fold improvement in catalytic efficiency compared to FoBgl-WT. Double mutants of FoBgl-WT targeting the +2 subsite of glucose binding further enhanced the glucose tolerance levels; however, the catalytic efficiencies decreased considerably. The structure-based rational engineering efforts improve glucose tolerance and the kinetic properties of FoBgl mutants, making it a useful and promising candidate enzyme for industrial cellulase cocktails.
ABSTRACT Predicted protein structures now enable fold discovery beyond the reach of sequence homology, but scalable and interpretable methods remain limited. We present Cooper‐Beta, a resource‐efficient, chain‐resolved framework for detecting β ‐barrel‐like folds from strand connectivity. Cooper‐Beta converts DSSP‐assigned physical β ‐strands into an adjacency graph and evaluates four measurements using calibrated direct rules or logistic regression. On a grouped held‐out benchmark of 466 positive and 1063 negative chains, the direct classifier achieved 87.77% recall and 97.65% specificity, while the learned classifier reached 94.42% recall and 97.08% specificity. A separate evaluation of 114 manually classified CATH Class II non‐barrels yielded specificities of 95.61% and 92.98%, respectively. Both classifiers retained prediction‐flip rates below 3% through 1.0 Å Gaussian C α perturbation, with stronger learned‐model sensitivity at higher noise. Screening 351 242 structures from the Big Fantastic Virus Database yielded 5803 joint‐positive candidates; sequential Swiss‐Prot and UniRef50 searches identified 1568 sequence‐available candidates without a passing Swiss‐Prot or UniRef50 match. Cooper‐Beta provides an interpretable route from predicted structural repositories to β ‐barrel‐like candidates for experimental follow‐up.
ABSTRACT Accurate prediction of nucleic acid‐binding sites is essential for understanding protein function and facilitating drug discovery. However, current computational methods are limited by single‐layer language model representations, Cα point clouds that discard local‐frame geometry, and feature fusion in which high‐dimensional sequence embeddings overshadow weaker evolutionary and structural signals. Here, we present PNBind, a multimodal deep learning framework that leverages protein language models, evolutionary information, and three‐dimensional structural features via a geometric vector perceptron (GVP) backbone with an invariant point attention (IPA) layer for nucleic acid‐binding site prediction. By integrating sequence representations from the ESM‐series language models with alignment‐derived evolutionary features, secondary‐structure data, and geometric features extracted from five‐atom local frames, PNBind enables effective late‐stage multimodal signal fusion. Comprehensive benchmarking on the DNA‐129, DNA‐181, RNA‐117, and RNA‐285 datasets reveals that PNBind achieves strong performance, with F 1 scores of 0.611, 0.443, 0.370, and 0.472 and MCC values of 0.586, 0.425, 0.351, and 0.416, respectively. Ablation studies confirm the dominant contribution of the protein language model backbone, while systematic evaluation of ESM‐family protein language models demonstrates that proper PLM selection significantly boosts prediction accuracy. This study demonstrates that the effective integration of protein language models and geometric deep learning offers a practical approach for accurate nucleic acid‐binding site prediction.
Metamorphic proteins challenge the classical view of protein folding by reversibly interconverting between two distinct, stable native structures. XCL1 (lymphotactin) is a prototypical example, transitioning between a monomeric mixed- chemokine fold and an alternate all- fold that forms a stable dimer. Here we develop and apply a coarse-grained, structure-based model to investigate how temperature and electrostatic screening control this fold switch. By tuning the relative interaction strength of native contacts in the two folds, we show that the model qualitatively reproduces the experimentally observed temperature dependence of XCL1 fold populations, including an increasing alternate-fold population with increasing temperature. We introduce a sequence-dependent treatment of salt effects, which reveals a strong sensitivity of the fold equilibrium to salt concentration and yields a temperature-salt phase diagram with a broad coexistence region. Focusing on conditions near coexistence, kinetic simulations of the chemokine-to-alternate fold transition show that productive dimerization tends to occur when both chains are at least partially unfolded, and that formation of the final alternate dimer may proceed via an intermediate state characterized by a partially formed dimer interface. Finally, we use the model to examine how charge patterning in ancestral variants of XCL1 modulates the population balance between the two folds.
