Hepatocellular carcinoma (HCC) is the sixth most common cancer and the third most common cause of cancer-related death, with tumour associated liver endothelial cells being thought to be major drivers in HCC progression. This study aims to compare the gene expression profiles of tumour endothelial cells from the liver with endothelial cells from non-tumour liver tissue, to identify perturbed biologic functions, co-expression modules, and potentially drugable hub genes that could give rise to novel therapeutic targets and strategies. Gene Set Variation Analysis (GSVA) showed that cell growth-related pathways were upregulated, whereas apoptosis induction, immune and inflammatory-related pathways were downregulated in tumour endothelial cells. Weighted Gene Co-expression Network Analysis (WGCNA) identified several modules strongly associated to tumour endothelial cells or angiogenic activated endothelial cells with high endoglin (ENG) expression. In tumour cells, upregulated modules were associated with cell growth, cell proliferation, and DNA-replication, whereas downregulated modules were involved in immune functions, particularly complement activation. In ENG+ cells, upregulated modules were associated with cell adhesion and endothelial functions. One downregulated module was associated with immune system-related functions. Querying the STRING database revealed known functional-interaction networks underlying the modules. Several possible hub genes were identified, of which some (for example FEN1, BIRC5, NEK2, CDKN3, and TTK) are potentially druggable as determined by querying the Drug Gene Interaction database. In summary, our study provides a detailed picture of the transcriptomic differences between tumour and non-tumour endothelium in the liver on a co-expression network level, indicates several potential therapeutic targets and presents an analysis workflow that can be easily adapted to other projects.
Glioblastoma (GBM) is characterized by a particularly invasive phenotype, supported by oncogenic signals from the fibroblast growth factor (FGF)/ FGF receptor (FGFR) network. However, a possible role of FGFR4 remained elusive so far. Several transcriptomic glioma datasets were analyzed. An extended panel of primary surgical specimen-derived and immortalized GBM (stem)cell models and original tumor tissues were screened for FGFR4 expression. GBM models engineered for wild-type and dominant-negative FGFR4 overexpression were investigated regarding aggressiveness and xenograft formation. Gene set enrichment analyses of FGFR4-modulated GBM models were compared to patient-derived datasets. Despite widely absent in adult brain, FGFR4 mRNA was distinctly expressed in embryonic neural stem cells and significantly upregulated in glioblastoma. Pronounced FGFR4 overexpression defined a distinct GBM patient subgroup with dismal prognosis. Expression levels of FGFR4 and its specific ligands FGF19/FGF23 correlated both in vitro and in vivo and were progressively upregulated in the vast majority of recurrent tumors. Based on overexpression/blockade experiments in respective GBM models, a central pro-oncogenic function of FGFR4 concerning viability, adhesion, migration, and clonogenicity was identified. Expression of dominant-negative FGFR4 resulted in diminished (subcutaneous) or blocked (orthotopic) GBM xenograft formation in the mouse and reduced invasiveness in zebrafish xenotransplantation models. In vitro and in vivo data consistently revealed distinct FGFR4 and integrin/extracellular matrix interactions. Accordingly, FGFR4 blockade profoundly sensitized FGFR4-overexpressing GBM models towards integrin/focal adhesion kinase inhibitors. Collectively, FGFR4 overexpression contributes to the malignant phenotype of a highly aggressive GBM subgroup and is associated with integrin-related therapeutic vulnerabilities.
Rational-design methods have proven to be a valuable toolkit in the field of protein design. Numerical approaches such as free-energy calculations or QM/MM methods are fit to widen the understanding of a protein-sequence space but require large amounts of computational time and power. Here, we apply an efficient method for free-energy calculations that combines the one-step perturbation (OSP) with the third-power-fitting (TPF) approach. It is fit to calculate full free energies of binding from three different end states only. The nonpolar contribution to the free energies are calculated for a set of chosen amino acids from a single simulation of a judiciously chosen reference state. The electrostatic contributions, on the other hand, are predicted from simulations of the neutral and charged end states of the individual amino acids. We used this method to perform in silico saturation mutagenesis of two sites in human Caspase-2. We calculated relative binding free energies toward two different substrates that differ in their P1′ site and in their affinity toward the unmutated protease. Although being approximate, our approach showed very good agreement upon validation against experimental data. 76% of the predicted relative free energies of amino acid mutations was found to be true positives or true negatives. We observed that this method is fit to discriminate amino acid mutations because the rate of false negatives is very low (<1.5%). The approach works better for a substrate with medium/low affinity with a Matthews correlation coefficient (MCC) of 0.63, whereas for a substrate with very low affinity, the MCC was 0.38. In all cases, the combined TPF + OSP approach outperformed the linear interaction energy method.
The existence of covalent heme to protein bonds is the most striking structural feature of mammalian peroxidases, including myeloperoxidase and lactoperoxidase (LPO). These autocatalytic posttranslational modifications (PTMs) were shown to strongly influence the biophysical and biochemical properties of these oxidoreductases. Recently, we reported the occurrence of stable LPO-like counterparts with two heme to protein ester linkages in bacteria. This study focuses on the model wild-type peroxidase from the cyanobacterium Lyngbya sp. PCC 8106 (LspPOX) and the mutants D109A, E238A, and D109A/E238A that could be recombinantly produced as apoproteins in Escherichia coli, fully reconstituted to the respective heme b proteins, and posttranslationally modified by hydrogen peroxide. This for the first time allows not only a direct comparison of the catalytic properties of the heme b and PTM forms but also a study of the impact of D109 and E238 on PTM and catalysis, including Compound I formation and the two-electron reduction of Compound I by bromide, iodide, and thiocyanate. It is demonstrated that both heme to protein ester bonds can form independently and that elimination of E238, in contrast to exchange of D109, does not cause significant structural rearrangements or changes in the catalytic properties neither in heme b nor in the PTM form. The obtained findings are discussed with respect to published structural and functional data of human peroxidases.
Despite the wealth of knowledge generated through epigenome-wide association studies our understanding of the relationships of CpG sites is still limited, as analysis of DNA methylation data remains difficult due its high dimensionality. To combat this problem, deep learning algorithms, such as autoencoders, are increasingly applied to capture the complex patterns and reduce dimensionality into latent space. We believe that the way an autoencoder groups together CpGs in its latent dimensions has biological meaning and might reveal novel insights regarding the relationship of CpGs. Therefore, in this work, we propose a chromosome-wise autoencoder for interpretable dimensionality reduction of methylation data (mEthAE). Our framework shows an impressive reduction in dimensions of up to 400-fold compared to the provided input, without compromising on reconstruction accuracy or predictive power in the latent space. Through our perturbation-based interpretability approach we revealed groups of CpGs which are highly connected across all latent dimensions (global CpGs) and were significantly more often reported in EWAS studies, indicating our interpretability method can successfully identify CpGs with biological relevance. In an attempt to gain a deeper understanding of the relationship between individual CpG sites, we focused on interpreting individual latent features and found that CpGs connected to a common feature do not share biological associations, correlation patterns, or are located in close proximity on the chromosome. We conclude that while there is evidence that the autoencoder does not group CpGs randomly, the logic behind the observed CpG relationships can not be delineated easily. With regards to the analyses done in this work, we believe that the autoencoder groups CpGs according to long range non-linear interaction patterns that lack characterisation in the current epigenetic research landscape.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.