Wong, N. K. Y. | Search bioRxiv

bioRxiv2024

5'-S-(3-aminophenyl)-5'-thioadenosine, a novel chemoprotective agent for reducing toxic side effects of fluorouracil in treatment of MTAP-deficient cancers

BackgroundNucleobase analogue (NBA) drugs are effective chemotherapeutics, but their clinical use is limited by severe side effects. Compelling evidence suggests the use of S-methyl-5-thioadenosine (MTA) can selectively reduce NBA toxicity on normal tissues while maintaining the efficacy of NBAs on methylthioadenosine phosphorylase (MTAP)-deficient cancers. However, we found that MTA induced hypothermia at its effective dose, limiting its translational potential. We intended to find an MTA analogue that can exert MTA function while minimize the undesired side effects of MTA. Thus, such an analogue can be used in combination with NBAs in selectively targeting MTAP-deficient cancers. MethodsWe screened a library of MTA analogues for the following criteria: 1) being substrates of MTAP; 2) selectively protection on MTAP-expressing cells from NBA toxicity using MTAP-isogenic cell lines; 3) ability to protect the host from NBA toxicity without hypothermic effect; and 4) lack of interference on the tumor-suppressive effect of NBA in mice bearing MTAP-deficient tumors. ResultsWe identified 5-S-(3-aminophenyl)-5-thioadenosine (m-APTA) that did not induce hypothermia at the effective doses. We demonstrated that m-APTA could be converted to adenine by MTAP. Consequently, m-APTA selectively protected mouse hosts from 5-FU-induced toxicity (i.e. anemia); yet it did not interfere with the drug efficacy on MTAP-deficient bladder cancers. In silico docking studies revealed that, unlike MTA, m-APTA interact inefficiently with adenosine A1 receptor, providing a plausible explanation of the superior safety profile of m-APTA. Conclusionm-APTA can significantly improve the translational potential of the NBA toxicity reduction strategy in selectively targeting MTAP-deficient cancers.

cancer biology↗

bioRxiv2023

Clinical Phenotype Prediction From Single-cell RNA-seq Data using Attention-Based Neural Networks

MotivationA patients disease phenotype can be driven and determined by specific groups of cells whose marker genes are either unknown, or can only be detected at late-stage using conventional bulk assays such as RNA-Seq technology. Recent advances in single-cell RNA sequencing (scRNA-seq) enable gene expression profiling in cell-level resolution, and therefore have the potential to identify those cells driving the disease phenotype even while the number of these cells is small. However, most existing methods rely heavily on accurate cell type detection, and the number of available annotated samples is usually too small for training deep learning predictive models. ResultsHere we propose the method ScRAT for clinical phenotype prediction using scRNA-seq data. To train ScRAT with a limited number of samples of different phenotypes, such as COVID and non-COVID, ScRAT first applies a mixup module to increase the number of training samples. A multi-head attention mechanism is employed to learn the most informative cells for each phenotype without relying on a given cell type annotation. Using three public COVID datasets, we show that ScRAT outperforms other phenotype prediction methods. The performance edge of ScRAT over its competitors increases as the number of training samples decreases, indicating the efficacy of our sample mixup. Critical cell types detected based on high-attention cells also support novel findings in the original papers and the recent literature. This suggests that ScRAT overcomes the challenge of missing marker genes and limited sample number with great potential revealing novel molecular mechanisms and/or therapies.

bioinformatics↗