Biology subjects

Pini, A.

Publications and source records attributed to Pini, A..

7 recordsLinked to original sources

bioRxiv2026

Branching Suppresses Self-Aggregation while Preserving Membrane Association in SET-M33: An Atomistic Molecular Dynamics Study

Branched antimicrobial peptides offer a promising strategy for improving the stability and activity of membrane-active peptides, but how branching alters self-association and membrane binding remains unclear. We used all-atom molecular dynamics simulations to compare tetra-branched SET-M33 with its linear branch sequence Q33 (KKIRVRLSA) in water and at an anionic bacterial-membrane model, to do this, systems containing equivalent numbers of Q33 sequence units were simulated. Q33 formed dimers and higher-order oligomers, particularly at the membrane interface and under hydration-tuned enhanced sampling. In contrast, SET-M33 remained more dispersed, as the branched scaffold and its high cationic charge limited peptide-peptide association while allowing multivalent membrane binding. Both architectures associated rapidly with the bilayer but remained predominantly surface-bound, without persistent pore formation or complete insertion. Q33 showed stronger membrane-induced helical stabilization, whereas SET-M33 retained a more heterogeneous conformational ensemble. Overall, branching primarily altered lateral peptide organization while preserving close membrane association, suppressing uncontrolled self-aggregation and providing an atomistic rationale for the distinct behavior of SET-M33 relative to Q33.

biophysics↗

bioRxiv2026

In vitro efficacy of synthetic antimicrobial peptide SET-M33 against poultry isolates with diverse antimicrobial resistance phenotypes

Antimicrobial resistance is an impactful One Health issue. One of its drivers is the extensive use of antibiotics in both human and animal production systems, and despite regulatory restrictions on antibiotic use in poultry farming, antimicrobial resistance remains a major challenge. Consequently, animals are at higher risk of harder-to-treat diseases and play a role as resistance reservoirs, highlighting the need for alternative antimicrobial strategies. Towards this end, antimicrobial peptides (AMPs) have emerged as promising candidates due to their broad-spectrum activity and lower propensity to induce resistance. However, the effectiveness of AMPs against poultry pathogens, and in particular multi drug-resistant strains, is largely unclear. To tackle this question, we evaluated the synthetic AMP SET-M33 against four species of clinically relevant pathogens in poultry, namely Escherichia coli, Salmonella enterica, Enterococcus faecalis and Enterococcus cecorum. Using a panel of 141 field isolates, we found that SET-M33 broadly inhibited bacterial growth at low micromolar concentrations (median MICs of 2.5 M and 5 M for Gram-negative and Gram-positive strains, respectively), including in multi drug-resistant isolates. To examine the potential impact of SET-M33 on the host, we established a new in vitro co-cultivation system using chicken intestinal organoids. We found that SET-M33 retains its antimicrobial activity in organoid-microbe co-cultures at concentrations that preserved host viability. These findings demonstrate the potential of SET-M33 as a new antimicrobial agent against pathogens in poultry.

microbiology↗

bioRxiv2026

A Conditional Variational Autoencoder with QSAR-Guided Surrogate-Weighted Fine-Tuning and Cross-Entropy Optimization for Targeted Antimicrobial Peptide Generation

Machine Learning frameworks have emerged as a promising tool for antimicrobial peptide design; however, generative models remain limited by two persistent problems: the limited availability of experimentally validated peptides and the circular dependency of the models. In this work we present a conditional variational autoencoder pipeline that addresses both limitations through a modular architecture that combines both binary and quantitative experimental data and implements a multimodal approach to externally guide the generation. A transformer-based encoder successfully generated a discriminative 64-dimensional latent space (test AUROC 0.968, F1 0.919) separating antimicrobial from non-antimicrobial sequences. This latent representation conditions a species-specific LoRA fine-tuned ProtGPT2 decoder through a scalar gating function, which generates balanced antimicrobial peptides through two different modes; prior and perturb, depending on their generation starting points. We introduced a Surrogate Weighted Fine-Tuning (SWF) ensemble to eliminate the circular dependency and a Cross-Entropy Method to explore and exploit the latent space, leading to successful antimicrobial peptide generation. The best candidates exhibited competitive physicochemical characteristics, a mean helical fraction of 0.874 (mean pLDDT 83.7), and externally predicted efficacy evaluated by APEX.

bioinformatics↗

bioRxiv2025

SET-M33 peptide as a selective in vitro antimicrobial agent against the porcine respiratory pathogen Glaesserella parasuis

As we face the threat from bacterial pathogens that are resistant to many conventional antibiotics, many current research efforts focus on expanding our arsenal of antimicrobial compounds. However, identifying use cases in which such new antimicrobials can effectively target pathogens while minimizing collateral damage in the commensal microbiota remains a challenge. To tackle this challenge, we focused on one new antimicrobial, the synthetic antimicrobial peptide SET-M33, and examined its ability to target porcine respiratory pathogens and a collection of porcine commensal nasal microbiota members in vitro. Our experiments revealed three key results. First, there were large differences in SET-M33 sensitivity across the tested strains. In particular, pathogenic Glaesserella parasuis was highly sensitive to SET-M33 at concentrations that did not affect the growth of most commensal strains. Second, some of the tested commensal strains (i.e. Rothia nasimurium and Staphylococcus aureus) were able to inactivate SET-M33 during in vitro cultivation. Third, despite this potential for SET-M33 inactivation by commensal strains, SET-M33 was still able to selectively eliminate pathogenic G. parasuis from in vitro co-cultures that also contained R. nasimurium. Overall, this study highlights the substantial complexity that emerges from the interplay between antimicrobials, pathogens, and commensals, even within a comparatively simple in vitro system, and provides a template for identifying suitable use cases for newly developed antimicrobials.

microbiology↗

bioRxiv2024

Branched Oncolytic Peptides Target HSPGs, Inhibit Metastasis, and Trigger the Release of Molecular Determinants of Immunogenic Cell Death in Pancreatic Cancer.

Immunogenic cell death (ICD) can be exploited to treat non-immunoreactive tumors that do not respond to current standard and innovative therapies. Not all chemotherapeutics trigger ICD, among those that do exert this effect, there are anthracyclines, irinotecan, some platinum derivatives and oncolytic peptides. We studied two new branched oncolytic peptides, BOP7 and BOP9 that proved to elicit the release of damage-associated molecular patterns DAMPS, mediators of ICD, in pancreatic cancer cells. The two BOPs selectively bound and killed tumor cells, particularly PANC-1 and Mia PaCa-2, but not cells of non-tumor origin such as RAW 264.7, CHO-K1 and pgsA-745. The cancer selectivity of the two BOPs may be attributed to their repeated cationic sequences, which enable multivalent binding to heparan sulfate glycosaminoglycans (HSPGs), bearing multiple anionic sulfation patterns on cancer cells. This interaction of BOPs with HSPGs not only fosters an anti-metastatic effect in vitro, as demonstrated by reduced adhesion and migration of PANC-1 cancer cells, but also shows promising tumor-specific cytotoxicity and low hemolytic activity. Remarkably, the cytotoxicity induced by BOPs triggers the release of DAMPs, particularly HMGB1, IFN-{beta} and ATP, by dying cells, persisting longer than the cytotoxicity of conventional chemotherapeutic agents such as irinotecan and daunorubicin. An in vivo assay in nude mice showed an encouraging 20% inhibition of tumor grafting and growth in a pancreatic cancer model by BOP9.

biochemistry↗

bioRxiv2024

A compact prism-based microscope for high sensitive measurements in fluid biopsy

The increasing demand for sensitive, portable, and cost-effective disease detection methods has raised significant interest in the development of biosensors for rapid and early-stage diagnosis, population mass-screening, and bedside monitoring. Although a certain number of sensing devices with high sensitivity has experienced considerable progress, their practical application has been hindered by challenges in replicating instrumental systems, obtaining rapid and multiplexed signals, and the high cost of manufacturing. In response, we present a streamlined prism-based Total Internal Reflection system which, in combination with surface functionalization techniques on gold nanoparticles, is capable of facilitating Evanescent Wave scattering for the highly sensitive and rapid detection of specific analytes in synthetic liquids and real human samples. The system achieves a remarkable limit of detection of 1 fg/mL concentration for the targeted pathological biomarker. Our innovative design addresses the limitations of existing technologies by reducing costs, minimising overall size, and ensuring swift biofluid analysis with remarkable sensitivity. To validate its efficacy, we conducted scattering experiments in synthetic and human serum samples, exploiting functionalized AuNPs to recognize bacterial lipopolysaccharides as biomarkers for sepsis disease. The cohesive integration of these techniques hopefully makes this biosensing setup a promising candidate for potential clinical deployment, meeting the pressing requirements for rapid personalised diagnosis, large-scale population screening and bed-monitoring.

biophysics↗

bioRxiv2018

New graph-theoretical-multimodal approach using temporal and structural correlations reveal disruption in the thalamo-cortical network in patients with schizophrenia

Schizophrenia has been understood as a network disease with altered functional and structural connectivity in multiple brain networks compatible to the extremely broad spectrum of psychopathological, cognitive and behavioral symptoms in this disorder.\n\nWhen building brain networks, functional and structural networks are typically modelled independently: functional network models are based on temporal correlations among brain regions, whereas structural network models are based on anatomical characteristics. Combining both features may give rise to more realistic and reliable models of brain networks.\n\nIn this study, we applied a new flexible graph-theoretical-multimodal model called FD (F, the functional connectivity matrix, and D, the structural matrix) to construct brain networks combining functional, structural and topological information of MRI measurements (structural and resting state imaging) to patients with schizophrenia (N=35) and matched healthy individuals (N=41). As a reference condition, the traditional pure functional connectivity (pFC) analysis was carried out.\n\nBy using the FD model, we found disrupted connectivity in the thalamo-cortical network in schizophrenic patients, whereas the pFC model failed to extract group differences after multiple comparison correction. We interpret this observation as evidence that the FD model is superior to conventional connectivity analysis, by stressing relevant features of the whole brain connectivity including functional, structural and topological signatures. The FD model can be used in future research to model subtle alterations of functional and structural connectivity resulting in pronounced clinical syndromes and major psychiatric disorders. Lastly, FD is not limited to the analysis of resting state fMRI, and can be applied to EEG, MEG etc.

neuroscience↗