The identification of BACE-1, a key enzyme in the production of Amyloid-β (Aβ) peptides, generated by the proteolytic processing of amyloid precursor protein, was a major advance in the field of Alzheimer's disease as this pathology is characterized by the presence of extracellular senile plaques, mainly comprised of Aβ peptides. Hydroxyethylamines have demonstrated a remarkable potential, like candidate drugs, for this disease using BACE-1 as target. Density Functional Theory calculations were employed to estimate interaction energies for the complexes formed between the hydroxyethylamine derivated inhibitors and 24 residues in the BACE-1 active site. The collected data offered not only a general but a particular quantitative description that gives a deep insight of the interactions in the active site, showing at the same time how ligand structural variations affect them. Polar interactions are the major energetic contributors for complex stabilization and those ones with charged aspartate residues are highlighted, as they contribute over 90% of the total attractive interaction energy. Ligand-ARG296 residue interaction reports the most repulsive value and decreasing of the magnitude of this repulsion can be a key feature for the design of novel and more potent BACE-1 inhibitors. Also it was explained why sultam derivated BACE-1 inhibitors are better ones than lactam based. Hydrophobic interactions concentrated at S1 zone and other relevant repulsions and attractions were also evaluated. The comparison of two different theory levels (X3LYP and M062X) allowed to confirm the relevance of the detected interactions as each theory level has its own strength to depict the forces involved, as is the case of M062X which is better describing the hydrophobic interactions. Those facts were also evaluated and confirmed by comparing the quantitative trend, of selected ligand-residue interactions, with MP2 theory level as reference standard method for electrostatic plus dispersion energies.

Meningitis is a potentially life-threatening infection characterized by the inflammation of the leptomeningeal membranes. Many different viral and bacterial pathogens can cause meningitis, with differences in mortality rates, risk of developing neurological sequelae and treatment options. Here we constructed a compendium of digital cerebrospinal fluid (CSF) proteome maps to define pathogen-specific host response patterns in meningitis. The results revealed a drastic and pathogen-type specific influx of tissue-, cell- and plasma proteins in the CSF, where in particular a large increase of neutrophil derived proteins in the CSF correlated with acute bacterial meningitis. Additionally, both acute bacterial and viral meningitis result in marked reduction of brain-enriched proteins. Generation of a multi-protein LASSO regression model resulted in an 18-protein panel of cell and tissue associated proteins capable of classifying acute bacterial meningitis and viral meningitis. The same protein panel also enabled classification of tick-borne encephalitis, a subgroup of viral meningitis, with high sensitivity and specificity. The work provides insights into pathogen specific host response patterns in CSF from different disease etiologies to support future classification of pathogen-type based on host response patterns in meningitis.

Generating and analyzing overlapping peptides through multienzymatic digestion is an efficient procedure for de novo protein using from bottom-up mass spectrometry (MS). Despite improved instrumentation and software, de novo MS data analysis remains challenging. In recent years, deep learning models have represented a performance breakthrough. Incorporating that technology into de novo protein sequencing workflows require machine-learning models capable of handling highly diverse MS data. In this study, we analyzed the requirements for assembling such generalizable deep learning models by systemcally varying the composition and size of the training set. We assessed the generated models’ performances using two test sets composed of peptides originating from the multienzyme digestion of samples from various species. The peptide recall values on the test sets showed that the deep learning models generated from a collection of highly N- and C-termini diverse peptides generalized 76% more over the termini-restricted ones. Moreover, expanding the training set’s size by adding peptides from the multienzymatic digestion with five proteases of several species samples led to a 2–3 fold generalizability gain. Furthermore, we tested the applicability of these multienzyme deep learning (MEM) models by fully de novo sequencing the heavy and light monomeric chains of five commercial antibodies (mAbs). MEMs extracted over 10000 matching and overlapped peptides across six different proteases mAb samples, achieving a 100% sequence coverage for 8 of the ten polypeptide chains. We foretell that the MEMs’ proven improvements to de novo analysis will positively impact several applications, such as analyzing samples of high complexity, unknown nature, or the peptidomics field.

No abstract

No abstract

No abstract

No abstract

Generating and analyzing overlapping peptides through multienzymatic digestion is an efficient procedure for de novo protein using from bottom-up mass spectrometry (MS). Despite improved instrumentation and software, de novo MS data analysis remains challenging. In recent years, deep learning models have represented a performance breakthrough. Incorporating that technology into de novo protein sequencing workflows require machine-learning models capable of handling highly diverse MS data. In this study, we analyzed the requirements for assembling such generalizable deep learning models by systematically varying the composition and size of the training set. We assessed the generated models' performances using two test sets composed of peptides originating from the multienzyme digestion of samples from various species. The peptide recall values on the test sets showed that the deep learning models generated from a collection of highly N- and C-termini diverse peptides generalized 76% more over the termini-restricted ones. Moreover, expanding the training set's size by adding peptides from the multienzymatic digestion with five proteases of several species samples led to a 2-3 fold generalizability gain. Furthermore, we tested the applicability of these multienzyme deep learning (MEM) models by fully de novo sequencing the heavy and light monomeric chains of five commercial antibodies (mAbs). MEM models extracted over 10000 matching and overlapped peptides across six different proteases mAb samples, achieving a 100% sequence coverage for 8 of the ten polypeptide chains. We foretell that the MEM models' proven improvements to de novo analysis will positively impact several applications, such as analyzing samples of high complexity, unknown nature, or the peptidomics field.