Explore the vast range of titles available.

Background Although clinical guidelines recommend vancomycin-based combination therapy for patients with postoperative intracranial infections in neurosurgery, the trend of global bacterial resistance and the management of antimicrobial agents have made monotherapy a common treatment option for some patients. This study aims to compare the efficacy of single-drug therapy (SDT) versus vancomycin combination therapy (VCT) in treating central nervous system infections (CNSIs) following neurosurgery. Methods A retrospective cohort study was conducted, adjusting for various covariates such as length of stay (LoS), admission status, age, comorbidity status (Charlson Comorbidity Index, CCI), surgical and incision levels, and duration of surgery (DOS) using propensity score matching (PSM) with a 1:2 ratio. The treatment effects of the two empirical treatment regimens were evaluated through PSM and logistic regression for dual robustness. Results A total of 539 patients met the inclusion criteria, with 177 cases in SDT and 101 cases in VCT after PSM. The clinical cure rate was 76% in the SDT compared to 90% in the VCT ( p  = 0.007) after PSM. Of the result of antibiotic susceptibility testing, only 13.9% of cases identified specific pathogens, of which gram-positive cocci were the dominant. VCT was significantly more effective than SDT, both in unadjusted (OR 2.941, 95% CI 1.434–6.607, p  = 0.005) and adjusted models (OR 3.605, 95% CI 1.611–8.812, p  = 0.003). Gender, race, and surgical complexity were significant factors influencing treatment choice; female patients and those with complex surgeries were less likely to receive SDT. Although SDT was practically effective for treating CNSIs, VCT proved superior for complex infections. Conclusion The findings of this study suggest that, given concerns about antibiotic resistance and the varying complexities of infections, while SDT is effective in certain cases, VCT remains the preferred choice for complex CNSIs. This research provides important references for clinical practice, highlighting the need to consider multiple factors when selecting treatment options and advancing the understanding of treatment strategies for postoperative central nervous system infections.

Duramycin is a heavily post-translationally modified peptide that binds phosphatidylethanolamine. It has been investigated as an antibiotic, an inhibitor of viral entry, a therapeutic for cystic fibrosis, and a tumor and vasculature imaging agent. Duramycin contains a β-hydroxylated Asp (Hya) and four macrocycles, including an essential lysinoalanine (Lal) cross-link. The mechanism of Lal formation is not known. Here we show that Lal is installed stereospecifically by DurN via addition of Lys19 to a dehydroalanine. The structure of DurN reveals an unusual dimer with a new fold. Surprisingly, in the structure of duramycin bound to DurN, no residues of the enzyme are near the Lal cross-link. Instead, Hya15 of the substrate makes interactions with Lal, suggesting it acts as a base to deprotonate Lys19 during catalysis. Biochemical data suggest that DurN preorganizes the reactive conformation of the substrate, such that the Hya15 of the substrate can serve as the catalytic base for Lal formation.

Protein sequence design in the context of small molecules, nucleotides and metals is critical to enzyme and small-molecule binder and sensor design, but current state-of-the-art deep-learning-based sequence design methods are unable to model nonprotein atoms and molecules. Here we describe a deep-learning-based protein sequence design method called LigandMPNN that explicitly models all nonprotein components of biomolecular systems. LigandMPNN significantly outperforms Rosetta and ProteinMPNN on native backbone sequence recovery for residues interacting with small molecules (63.3% versus 50.4% and 50.5%), nucleotides (50.5% versus 35.2% and 34.0%) and metals (77.5% versus 36.0% and 40.6%). LigandMPNN generates not only sequences but also sidechain conformations to allow detailed evaluation of binding interactions. LigandMPNN has been used to design over 100 experimentally validated small-molecule and DNA-binding proteins with high affinity and high structural accuracy (as indicated by four X-ray crystal structures), and redesign of Rosetta small-molecule binder designs has increased binding affinity by as much as 100-fold. We anticipate that LigandMPNN will be widely useful for designing new binding proteins, sensors and enzymes. LigandMPNN is a deep-learning-based protein sequence design method that can explicitly model all nonprotein components of biomolecular systems.

In pseudocyclic proteins, such as TIM barrels, β barrels, and some helical transmembrane channels, a single subunit is repeated in a cyclic pattern, giving rise to a central cavity that can serve as a pocket for ligand binding or enzymatic activity. Inspired by these proteins, we devised a deep-learning-based approach to broadly exploring the space of closed repeat proteins starting from only a specification of the repeat number and length. Biophysical data for 38 structurally diverse pseudocyclic designs produced in Escherichia coli are consistent with the design models, and the three crystal structures we were able to obtain are very close to the designed structures. Docking studies suggest the diversity of folds and central pockets provide effective starting points for designing small-molecule binders and enzymes. Here, the authors constructed a deep-learning approach to design closed repeat proteins with central binding pockets—a step towards designing proteins to specifically bind small molecules.

Natural products and their derivatives have been significant resources for the development of therapeutic compounds. They attracted interests from both academia and industry because of their high structural diversity and potential applications. Lanthipeptides are one class of natural products that have provided antibiotics to the food industry and drug candidates for treating human diseases. Lanthipeptides are polypeptides enzymatically decorated with lanthionine rings and sometimes other post-translational modifications, which dramatically elevate their protease-resistances, improved their chemical stabilities, and increased their structural complexity. To add onto the structure knowledge and biosynthetic toolkits for lanthipeptides, I investigated the biosynthesis and discovery of lanthipeptides during my Ph.D training. Duramycin/cinnamycin-type of lanthipeptides interact tightly with phosphatidylethanolamine and several of their members displayed high potential to be drug candidates. Duramycin contains an activity-essential lysinoalanine ring which is installed by a previously unknown hypothetical protein, DurN. In Chapter 2, I described the mechanism of action studies on DurN. I reconstituted the in vitro activity of DurN. Together with Dr. Cogan, we obtained the co-crystal structures of DurN with its product or substrate analog. We demonstrated that DurN catalyzes the lysinoalanine formation through a unique substrate-assisted catalysis mechanism. Enlightened by the biosynthesis of lanthipeptide, I further designed and initiated a proof-of-concept lanthipeptide discovery project based on the predictions for the potential mode of action of natural products, which is described in Chapter 3. I hypothesized that if the gene encoding a small molecule-processing enzyme locates in the biosynthetic gene cluster of a natural product on the bacterial genome, this processing enzyme may function as the immunity protein to prevent producer viability loss during the production of the natural products, and the natural product may target the small molecule. Following this hypothesis, I identified multiple lanthipeptide biosynthesis gene cluster candidates, and selected kib cluster from Kibdelsporangium phytohabitant KLBMP 1111T for verification. The lanthipeptide was produced in heterologous expression system and displayed an interlocking ring topology with a succinimide moiety as potential warhead. The activity assays of this new lanthipeptide will be carried out in the future studies. To further understand the modes of action for lanthipeptides, in Chapter 4, I investigated the mode of action of lipid II-targeting lanthipeptides. The binding event between nisin–lipid II and Halα–lipid II were characterized using isothermal titration calorimetry. Collectively, these studies further expanded our knowledge on lanthipeptides biosynthesis and discovery.

Using N-(2-Aminoethyl)maleimide-cysteine(StBu) (Mal-Cys) as a medium, protein thiols were converted into N-terminal cysteines. After a biocompatible condensation reaction between the N-terminal cysteine and fluorescent probe 2-cyanobenzothiazole-Gly-Gly-Gly-fluorescein isothiocyanate (CBT-GGG-FITC), a new fluorogenic structure Luciferin-GGG-FITC was obtained. The latter exhibits near one order of magnitude (7 folds) enhanced fluorescence emission compared to the precursor moiety due to fluorescence resonance energy transfer (FRET) effect between the newly formed luciferin structure and the FITC motif. Theoretical investigations revealed the underlying mechanism that satisfactorily explained the experimental results. With this method, enhanced fluorescence imaging of thiols on proteins, outer membranes of living cells, translocation of membrane proteins and endothelial cell layers of small arteries was successfully achieved.

Modeling the conformational heterogeneity of protein-small molecule interactions is important for understanding natural systems and evaluating designed systems, but remains an outstanding challenge. We reasoned that while residue level descriptions of biomolecules are efficient for de novo structure prediction, for probing heterogeneity of interactions with small molecules in the folded state an entirely atomic level description could have advantages in speed and generality. We developed a graph neural network called PLACER (Protein-Ligand Atomistic Conformational Ensemble Resolver) trained to recapitulate correct atomic positions from partially corrupted input structures from the Cambridge Structural Database and the Protein Data Bank; the nodes of the graph are the atoms in the system. PLACER accurately generates structures of diverse organic small molecules given knowledge of their atom composition and bonding, and given a description of the larger protein context, builds up structures of small molecules and protein side chains for protein-small molecule docking. Because PLACER is rapid and stochastic, ensembles of predictions can be readily generated to map conformational heterogeneity. In enzyme design efforts described here and elsewhere, we find that using PLACER to assess the accuracy and pre-organization of the designed active sites results in higher success rates and higher activities; we obtain a preorganized retroaldolase with a / of 11000 M min , considerably higher than any pre-deep learning design for this reaction. We anticipate that PLACER will be widely useful for rapidly generating conformational ensembles of small molecule and small molecule-protein systems, and for designing higher activity preorganized enzymes.

Modeling the conformational heterogeneity of protein-small molecule systems is an outstanding challenge. We reasoned that while residue level descriptions of biomolecules are efficient for de novo structure prediction, for probing heterogeneity of interactions with small molecules in the folded state an entirely atomic level description could have advantages in speed and generality. We developed a graph neural network called ChemNet trained to recapitulate correct atomic positions from partially corrupted input structures from the Cambridge Structural Database and the Protein Data Bank; the nodes of the graph are the atoms in the system. ChemNet accurately generates structures of diverse organic small molecules given knowledge of their atom composition and bonding, and given a description of the larger protein context, and builds up structures of small molecules and protein side chains for protein-small molecule docking. Because ChemNet is rapid and stochastic, ensembles of predictions can be readily generated to map conformational heterogeneity. In enzyme design efforts described here and elsewhere, we find that using ChemNet to assess the accuracy and pre-organization of the designed active sites results in higher success rates and higher activities; we obtain a preorganized retroaldolase with a k cat/K M of 11000 M-1min-1, considerably higher than any pre-deep learning design for this reaction. We anticipate that ChemNet will be widely useful for rapidly generating conformational ensembles of small molecule and small molecule-protein systems, and for designing higher activity preorganized enzymes.Modeling the conformational heterogeneity of protein-small molecule systems is an outstanding challenge. We reasoned that while residue level descriptions of biomolecules are efficient for de novo structure prediction, for probing heterogeneity of interactions with small molecules in the folded state an entirely atomic level description could have advantages in speed and generality. We developed a graph neural network called ChemNet trained to recapitulate correct atomic positions from partially corrupted input structures from the Cambridge Structural Database and the Protein Data Bank; the nodes of the graph are the atoms in the system. ChemNet accurately generates structures of diverse organic small molecules given knowledge of their atom composition and bonding, and given a description of the larger protein context, and builds up structures of small molecules and protein side chains for protein-small molecule docking. Because ChemNet is rapid and stochastic, ensembles of predictions can be readily generated to map conformational heterogeneity. In enzyme design efforts described here and elsewhere, we find that using ChemNet to assess the accuracy and pre-organization of the designed active sites results in higher success rates and higher activities; we obtain a preorganized retroaldolase with a k cat/K M of 11000 M-1min-1, considerably higher than any pre-deep learning design for this reaction. We anticipate that ChemNet will be widely useful for rapidly generating conformational ensembles of small molecule and small molecule-protein systems, and for designing higher activity preorganized enzymes.

Fluorescent proteins and small molecule dyes have complementary strengths for biological imaging: the former are genetically manipulatable enabling tagging of specific proteins and detection of protein interactions, while the latter have greater photostability and brightness but are difficult to target. To combine these strengths, we used de novo protein design to generate binders to three bright, stable, cell-permeable dyes spanning the visible spectrum: JF657 (far red), JF596 (orange-red) and JF494 (green). For each dye, we obtain nanomolar binders with weak or no binding to the other two dyes; the accuracy of the design approach is confirmed by a crystal structure of one binder which is very close to the design model. Fusion of the JF567, JF596 and JF494 binders to three different targets followed by staining with the three dyes simultaneously enables multiplex imaging. We further expand functionality by incorporating an active site carrying out nucleophilic aromatic substitution to form a covalent linkage with the dye, and developing split versions which reconstitute fluorescence at subcellular locations where both halves are present, enabling both protein-protein interaction detection and chemically induced dimerization with fluorescence reporting. Our designs combine the advantages of fluorescent proteins and small molecule dyes and should be broadly useful for cellular imaging.

A general method for designing proteins to bind and sense any small molecule of interest would be widely useful. Due to the small number of atoms to interact with, binding to small molecules with high affinity requires highly shape complementary pockets, and transducing binding events into signals is challenging. Here we describe an integrated deep learning and energy based approach for designing high shape complementarity binders to small molecules that are poised for downstream sensing applications. We employ deep learning generated psuedocycles with repeating structural units surrounding central pockets; depending on the geometry of the structural unit and repeat number, these pockets span wide ranges of sizes and shapes. For a small molecule target of interest, we extensively sample high shape complementarity pseudocycles to generate large numbers of customized potential binding pockets; the ligand binding poses and the interacting interfaces are then optimized for high affinity binding. We computationally design binders to four diverse molecules, including for the first time polar flexible molecules such as methotrexate and thyroxine, which are expressed at high levels and have nanomolar affinities straight out of the computer. Co-crystal structures are nearly identical to the design models. Taking advantage of the modular repeating structure of pseudocycles and central location of the binding pockets, we constructed low noise nanopore sensors and chemically induced dimerization systems by splitting the binders into domains which assemble into the original pseudocycle pocket upon target molecule addition.