RESUMEN
Prediction and understanding of virus-host protein-protein interactions (PPIs) have relevance for the development of novel therapeutic interventions. In addition, virus-like particles open novel opportunities to deliver therapeutics to targeted cell types and tissues. Given our incomplete knowledge of PPIs on the one hand and the cost and time associated with experimental procedures on the other, we here propose a deep learning approach to predict virus-host PPIs. Our method (Siamese Tailored deep sequence Embedding of Proteins [STEP]) is based on recent deep protein sequence embedding techniques, which we integrate into a Siamese neural network. After showing the state-of-the-art performance of STEP on external datasets, we apply it to two use cases, severe acute respiratory syndrome coronavirus 2 and John Cunningham polyomavirus, to predict virus-host PPIs. Altogether our work highlights the potential of deep sequence embedding techniques originating from the field of NLP as well as explainable artificial intelligence methods for the analysis of biological sequences.
RESUMEN
Hollow viral vectors, such as John Cunningham virus-like particles (JC VLPs), provide a unique opportunity to deliver drug cargo into targeted cells and tissue. Current understanding of the entry of JC virus in brain cells has remained insufficient. In particular, interaction of JC VLPs with the blood-brain barrier (BBB) has not been analyzed in detail. Thus, JC VLPs were produced in this study for investigating the trafficking across the BBB. We performed a carotid artery injection procedure for mouse brain to qualitatively study JC VLPs' in vivo binding and distribution and used in vitro approaches to analyze their uptake and export kinetics in brain endothelial cells. Our results show that clathrin-dependent mechanisms contributed to the entry of VLPs into brain endothelial cells, and exocytosis or transcytosis of VLPs across the BBB was observed in vitro. VLPs were found to interact with sialic acid glycans in mouse brain endothelia. The ability of JC VLPs to cross the BBB can be useful in developing a delivery system for transport of genes and small molecule cargoes to the brain.
RESUMEN
JC polyomavirus (JCV) infection is highly prevalent and usually kept in a persistent state without clinical signs and symptoms. It is only during immunocompromise and especially impaired CD4(+) T cell function in the brain, as seen in AIDS patients or natalizumab-treated multiple sclerosis patients, that JCV may cause progressive multifocal leukoencephalopathy (PML), an often life-threatening brain disease. Since CD4(+) T cells likely play an important role in controlling JCV infection, we here describe the T cell response to JCV in a group of predominantly HLA-DR-heterozygotic healthy donors (HD) by using a series of overlapping 15-mer peptides spanning all JCV-encoded open reading frames. We identified immunodominant epitopes and compared T cell responses with anti-JCV VP1 antibody production and with the presence of urinary viral shedding. We observed positive JCV-specific T cell responses in 28.6% to 77.6%, humoral immune response in 42.6% to 89.4%, and urinary viral shedding in 36.4% to 45.5% of HD depending on the threshold. Four immunodominant peptides were mapped, and at least one immunogenic peptide per HLA-DRB1 allele was detected in DRB1*01(+), DRB1*07(+), DRB1*11(+), DRB1*13(+), DRB1*15(+), and DRB1*03(+) individuals. We show for the first time that JCV-specific T cell responses may be directed not only against JCV VP1 and large T antigen but also against all other JCV-encoded proteins. Heterozygotic DRB1*04:01(+) individuals showed very low T cell responses to JCV together with normal anti-VP1 antibody levels and no urinary viral shedding, indicating a dominant-negative effect of this allele on global JCV-directed T cell responses. Our data are potentially relevant for the development of vaccines against JCV.