Back

Multiomics data analysis identifies TM4SF4 as a cell surface target in hepatocellular carcinoma

Wong, K. K.; Ab Hamid, S. S.

2024-07-03 cancer biology
10.1101/2024.07.01.601537 bioRxiv
Show abstract

The clinical application of cellular immunotherapy in hepatocellular carcinoma (HCC) is impeded by the lack of a cell surface target frequently expressed in HCC cells and with minimal presence in normal tissues to reduce on-target, off-tumor toxicity. In this in silico analysis, a longlist of genes (n=12,948) expressed in HCCs according to The Human Protein Atlas database were examined. The aim was to identify the optimal cell surface target, without being shed into circulation, in HCCs and with restrictive expression profile in other normal human tissues. A total of eight genes were shortlisted and they were subsequently ranked according to the combination of their transcript and protein expression levels in HCC cases (n=791) derived from four independent datasets (TCGA, CNHPP, GSE14520, and CHCC). TM4SF4 was the top-ranked target with the highest expression in HCCs. TM4SF4 demonstrated more favorable expression profile with significantly lower expression in normal human tissues but more highly expressed in HCC cases compared with seven other common HCC therapeutic targets. Furthermore, scRNA-seq (GSE149614 and GSE134355) and immunohistochemistry (HPA) datasets showed that TM4SF4 was absent in immune cell populations but highly expressed in the bile duct canaliculi of hepatocytes, a region that is inaccessible to immune cells. In scRNA-seq dataset of HCCs (GSE149614), TM4SF4 expression was positively associated with mitochondrial components and oxidative phosphorylation Gene Ontologies in HCC cells (n=15,787 cells), suggesting its potential roles in mitochondrial-mediated oncogenic effects in HCC. Taken together, TM4SF4 is proposed as a promising cell surface target in HCC.

Matching journals

The top 11 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.