Machine learning optimization of peptides for presentation by class II MHCs

T cells play a critical role in cellular immune responses to pathogens and cancer and can be activated and expanded by MHC-presented antigens contained in peptide vaccines. We present a machine learning method to optimize the presentation of peptides by class II MHCs by modifying their anchor residues. Our method first learns a model of peptide affinity for a class II MHC using an ensemble of deep residual networks, and then uses the model to propose anchor residue changes to improve peptide affinity. We use a high throughput yeast display assay to show that anchor residue optimization improves peptide binding. Supplementary information: Supplementary data are available at Bioinformatics online.

Context contribution to the intermolecular recognition of human ACE2-derived peptides by SARS-CoV-2 spike protein: implications for improving the peptide affinity but not altering the peptide specificity by optimizing indirect readout

Disrupting the intermolecular interaction of SARS-CoV-2 S protein with its cell surface receptor hACE2 is a therapeutic strategy against COVID-19. The protein context plays an essential role in hACE α1-helix recognition by viral S protein.

Peptide-Affinity Precipitation of Extracellular Vesicles and Cell-Free DNA Improves Sequencing Performance for the Detection of Pathogenic Mutations in Lung Cancer Patient Plasma

Liquid biopsy is a minimally-invasive diagnostic method that may improve access to molecular profiling for non-small cell lung cancer (NSCLC) patients. Although cell-free DNA (cf-DNA) isolation from plasma is the standard liquid biopsy method for detecting DNA mutations in cancer patients, the sensitivity can be highly variable. Vn96 is a peptide with an affinity for both extracellular vesicles (EVs) and circulating cf-DNA. In this study, we evaluated whether peptide-affinity (PA) precipitation of EVs and cf-DNA from NSCLC patient plasma improves the sensitivity of single nucleotide variants (SNVs) detection and compared observed SNVs with those reported in the matched tissue biopsy. NSCLC patient plasma was subjected to either PA precipitation or cell-free methods and total nucleic acid (TNA) was extracted; SNVs were then detected by next-generation sequencing (NGS). PA led to increased recovery of DNA as well as an improvement in NGS sequencing parameters when compared to cf-TNA. Reduced concordance with tissue was observed in PA-TNA (62%) compared to cf-TNA (81%), mainly due to identification of SNVs in PA-TNA that were not observed in tissue. EGFR mutations were detected in PA-TNA with 83% sensitivity and 100% specificity. In conclusion, PA-TNA may improve the detection limits of low-abundance alleles using NGS.

Machine learning optimization of peptides for presentation by class II MHCs

T cells play a critical role in normal immune responses to pathogens and cancer and can be targeted to MHC-presented antigens via interventions such as peptide vaccines. Here, we present a machine learning method to optimize the presentation of peptides by class II MHCs by modifying the peptide’s anchor residues. Our method first learns a model of peptide affinity for a class II MHC using an ensemble of deep residual networks, and then uses the model to propose anchor residue changes to improve peptide affinity. We use a high throughput yeast display assay to show that anchor residue optimization successfully improved peptide binding.

Tumor-Targeting Peptides: The Functional Screen of Glioblastoma Homing Peptides to the Target Protein FABP3 (MDGI)

We recently identified the glioblastoma homing peptide CooP (CGLSGLGVA) using in vivo phage display screen. The mammary-derived growth inhibitor (MDGI/FABP3) was identified as its interacting partner. Here, we present an alanine scan of A-CooP to investigate the contribution of each amino acid residue to the binding to FABP3 by microscale thermophoresis (MST) and surface plasmon resonance (SPR). We also tested the binding affinity of the A-CooP-K, KA-CooP, and retro-inverso A-CooP analogues to the recombinant FABP3. According to the MST analysis, A-CooP showed micromolar (KD = 2.18 µM) affinity to FABP3. Alanine replacement of most of the amino acids did not affect peptide affinity to FABP3. The A-CooP-K variant showed superior binding affinity, while A-[Ala5]CooP and A-[Ala7]CooP, both replacing a glycine residue with alanine, showed negligible binding to FABP3. These results were corroborated in vitro and in vivo using glioblastoma models. Both A-CooP-K and A-CooP showed excellent binding in vitro and homing in vivo, while A-[Ala5]CooP and control peptides failed to bind the cells or home to the intracranial glioblastoma xenografts. These results provide insight into the FABP3–A-CooP interaction that may be important for future applications of drug conjugate design and development.

Improvement of Peptide Affinity and Stability by Complexing to Cyclodextrin-Grafted Ammonium Chitosan

Cyclodextrin-grafted polymers are attractive biomaterials that could bring together the host–guest complexing capability of pristine cyclodextrin and the pharmaceutical features of the polymeric backbone. The present paper is aimed at characterizing the potential application of ammonium–chitosan grafted with 2-methyl-β-cyclodextrin (N+-rCh-MCD) as the functional macromolecular complexing agent for the oral administration of the neuropeptide dalargin (DAL). Specific NMR characterization procedures, along with UV and fluorescence techniques, as well as biological in vitro assessments have been performed. The results indicate that N+-rCh-MCD forms water-soluble complexes with DAL, with a prevalent involvement of Tyr or Phe over Leu and Ala residues. The association constant of DAL with the polymeric derivative is one order of magnitude higher than that with the pristine cyclodextrin (Ka: 2600 M−1 and 120 M−1, respectively). Additionally, N+-rCh-MCD shields DAL from enzymatic degradation in gastrointestinal in vitro models with a three-fold time delay, suggesting a future pharmaceutical exploitation of the polymeric derivative. Therefore, the greater affinity of N+-rCh-MCD for DAL and its protective effect against enzymatic hydrolysis can be attributed to the synergistic cooperation between cyclodextrin and the polymer, which is realized only when the former is covalently linked to the latter.

Anti-Leishmanial Vaccines: Assumptions, Approaches, and Annulments

Leishmaniasis is a neglected protozoan parasitic disease that occurs in 88 countries but a vaccine is unavailable. Vaccination with live, killed, attenuated (physically or genetically) Leishmania have met with limited success, while peptide-, protein-, or DNA-based vaccines showed promise only in animal models. Here, we critically assess several technical issues in vaccination and expectation of a host-protective immune response. Several studies showed that antigen presentation during priming and triggering of the same cells in infected condition are not comparable. Altered proteolytic processing, antigen presentation, protease-susceptible sites, and intracellular expression of pathogenic proteins during Leishmania infection may vary dominant epitope selection, MHC-II/peptide affinity, and may deter the reactivation of desired antigen-specific T cells generated during priming. The robustness of the memory T cells and their functions remains a concern. Presentation of the antigens by Leishmania-infected macrophages to antigen-specific memory T cells may lead to change in the T cells’ functional phenotype or anergy or apoptosis. Although cells may be activated, the peptides generated during infection may be different and cross-reactive to the priming peptides. Such altered peptide ligands may lead to suppression of otherwise active antigen-specific T cells. We critically assess these different immunological issues that led to the non-availability of a vaccine for human use.