Serine Metabolic Network: Gene Expression & Flux in Cancer

The serine, glycine and one-carbon (SGOC) network is a complex, cyclic system that converts serine and glycine into metabolic outputs necessary for cells to accomplish a myriad of biological functions. The SGOC network has also been linked to cancer pathogenesis, although the precise implicated functions remain unknown. Using computational reconstruction, Mehrmohamadi et al. characterized SGOC expression across a broad range of cancer tissues. Then they experimentally highlighted several context-specific uses of serine in human cancers.¹

To do this, the team connected “nodes” made up of the genes from the Kyoto Encyclopedia of Genes and Genomes (KEGG)-defined pathways—glycine-serine and threonine metabolism, cysteine and methionine metabolism, and folate biosynthesis—with adjacent chemical reactions, or “edges.” Pruning left a 64-gene set associated with key metabolic reactions and enzymes/isoenzymes. They also employed stable isotopes to label serine, followed by high-resolution mass spectrometry and mathematical modeling to trace serine’s metabolic fate. They relied on an Ultimate 3000 UHPLC coupled to a Q Exactive mass spectrometer for data acquisition and Sieve software revision 2.0 for data processing (all Thermo Scientific).

The team used the network to analyze expression in normal tissues and tissues compromised by breast, ovarian, lung or colorectal cancer drawn from two repositories, the Gene Expression in Normal and Tumor (GENT) database for normal and tumor tissues and The Cancer Genome Atlas (TCGA) for tumor samples.

Hierarchical clustering showed tissue type-dependent grouping, except for lung cancer samples and a subset of breast cancer samples without estrogen receptors, which instead clustered with ovarian cancer samples. Globally, the team observed higher expression of SGOC network genes in ovarian, colon and lung cancer samples as compared to normal tissue; however, breast cancer samples demonstrated higher variability in SGOC expression levels. All cancer types showed differential expression of some kind between tumor and normal tissues.

Then, the team turned to functional analysis, including de novo serine biosynthesis as a distinct pathway because of previous research showing its association with cancer. Examining the range of expression for each pathway, the researchers observed high within-cancer variability, particularly for breast tumors. They noted that some pathways, such as methylation, showed similar expression patterns across cancer types, while others, including de novo serine biosynthesis, demonstrated significant variations. Heterogeneity emerged as a major finding in this study.

Next, they took functional pathway analysis to a gene set level and compared expression levels in several ways: one tumor type relative to other tumor types, variability among single tumor types, overexpression in tumor relative to normal tissue, expression in normal tissue relative to other tissue types, and variability in tumor relative to corresponding normal tissue. Notable observations include several overexpressed pathways related to sulfur metabolism among breast and ovarian cancer samples. Across tumor types, they again noted high variability throughout the entire network, with the most notable pathways including taurine, methylation and NADPH pathways.

Interestingly, when the team compared tumor samples to normal tissues, they observed upregulation only in pathways related to nucleotide and redox metabolism, which is consistent with their individual gene data. Furthermore, when the researchers compared pathways and individual genes, they found that high expression levels in normal tissue did not necessarily correlate with high expression levels in tumor tissues of the same origin, suggesting that metabolism shifts do not result directly from tissue-specific differences.

Overall, Mehrmohamadi et al. report successful use of pathway-level SGOC network gene expression for the prediction of metabolic fluxes. This is valuable because a great deal of gene expression data is already publicly available, and the results of this paper would allow researchers to avoid the technical issues that come with large-scale metabolomics analysis. The team offers its findings as the first comprehensive study of its kind and calls for further comparative analysis of the relationship between gene expression and flux in biological samples.

Reference

1. Mehrmohamadi, M., et al. (2014) “Characterization of the usage of the serine metabolic network in human cancer,” Cell Reports, 9 (pp. 1–13).