Computational protocol: An integrated computational and experimental study uncovers FUT9 as a metabolic driver of colorectal cancer

Similar protocols

Protocol publication

[…] FUT9 belongs to the glycosyltransferase family and catalyzes the last step in the biosynthesis of Ley glycolipids in the carbohydrate antigen Lex (Nishihara et al, ; Gouveia et al, ). This reaction takes place in the Golgi compartment, and the product is transported to the cytosol and secreted out from the cell (Duarte et al, ). The Ley glycolipid was previously reported to inhibit the procoagulant activity and metastasis of human adenocarcinoma (Nudelman et al, ; Suzuki et al, ; Inufusa et al, ). The loss of FUT9 in the metabolic model prevents Ley glycolipid formation and secretion. To chart the network‐wide metabolic alterations induced by FUT9 inactivation, we performed a Minimization Of Metabolic Adjustment (MOMA) (Segrè et al, ) analysis to predict the metabolic state after FUT9 KD in late‐stage colorectal cancers, simulated by the Gene Inactivity Moderated by Metabolism and Expression (GIMME) algorithm (Becker & Palsson, ) (). This pinpoints reactions whose flux is predicted to be most afflicted by FUT9 inactivation in advanced‐stage cancer. We found that the loss of FUT9 in late‐stage colorectal cancers is predicted to cause an increase in the flux of 25 reactions, and a decrease in the flux of six reactions (). The flux is predicted to increase in reactions associated with glucose metabolism, and particularly TCA cycle (hyper‐geometric P‐value = 1.3676e‐09, Fig D, ). We find that the expression of metabolic genes associated with reactions predicted to increase following FUT9 loss is significantly upregulated in stage 4 vs. stage 3 colon tumors when compared by their expression in TCGA data (hyper‐geometric P‐value = 0.0046, ). Experimental evaluation of these predictions using the Human Glucose Metabolism, RT² Profiler™ PCR Array revealed a good correlation with our computational prediction (Fig D). In particular, 12 genes, including FH and SDHD, proved to be upregulated in FUT9‐silenced cells as expected from our computational analyses ().To evaluate the effect of FUT9 knockdown (KD) and overexpression (OE) on biomass production, glucose consumption, lactate production, and oxygen consumption in the benign colon adenoma state, we (i) simulated the wild‐type metabolic state associated with colon adenoma. This was done by incorporating adenoma gene expression data from Sabates‐Bellver et al () using the GIMME algorithm. (ii) We then sampled 100 flux distributions in the resulting predicted adenoma wild‐type state. In each such sample, we applied the MOMA (Segrè et al, ) algorithm to predict the metabolic state after FUT9 KD and OE in adenoma, summing up the results overall 100 samples (). We find that the biomass production predicted is significantly higher under FUT9 OE than its KD, as well as lactate secretion rate (Wilcoxon rank‐sum P‐value = 0.0081 and 0.0173, respectively, Fig E). While oxygen consumption rate is significantly higher under FUT9 KD (Wilcoxon rank‐sum P‐value = 6.79e‐8, Fig E). These predictions imply that FUT9 activity is required for supporting cancer proliferation in the adenoma state, which are consistent with the genomic findings we reported above that, while FUT9 expression is strongly downregulated in colon cancer, it is not significantly downregulated at early‐stage colon adenomas.We next evaluated the metabolic effects of FUT KD and OE in the colon tumor state. To this end, we performed a similar analysis as described above for adenoma, while first inferring the likely metabolic state of colon tumors (). Strikingly, we find that the predicted biomass production in the cancerous state is significantly higher under FUT9 KD than its OE (Wilcoxon rank‐sum P‐value = 0.0245, Fig F) and that lactate production rate is also increased under FUT9 KD (Wilcoxon rank‐sum P‐value = 0.0859, Fig F), opposite to the observed in simulated colon adenoma state. These predictions imply that the loss of FUT9, while hampering the growth of adenomas, is required for the proliferation of colon tumors, while its overexpression significantly reduces proliferation in that state.Given the opposite predicted effects of KD perturbation in colon adenomas vs. tumors, we performed an additional GSMM analysis to study whether FUT9 inactivation at early colorectal cancer stages can induce the metabolic state observed at advanced tumors, or only its inactivation at late stages can induce this transformation. To this end, we first inferred the likely metabolic state of advanced colorectal tumors using the GIMME algorithm (Becker & Palsson, ), as done above in the adenoma analysis. We then predicted the likely metabolic states after the loss if FUT9 in each of the four different stages of colorectal cancer progression, asking how similar is the metabolic state induced after the loss of FUT9 in each of these stages to the advanced, late cancerous state. The metabolic state after the KD of FUT9 in each stage‐specific context was predicted using the MOMA algorithm (Segrè et al, ) (). This analysis revealed that the loss of FUT9 at early stages does not bring the metabolic state close to that observed in advanced cancer. Rather, for the FUT9 loss to cause such an effect, it has to occur in later stages of the disease (Fig G). This indicates that FUT9 downregulation is a tumor‐transformative event only if occurs at later stages of tumor progression. To study this further from a genomic perspective, we analyzed the correlation between FUT9 copy number and the copy number levels of known early and late genetic markers of colorectal cancer. We find that FUT9 expression levels negatively correlate with the loss of the early markers APC and MCC (Spearman ρ = −0.1726 and −0.1707, P‐value < 0.05, respectively), while it is positively correlated with the loss of TP53, a marker of the advanced stage (Fearon, ; Lurje et al, ) (Spearman ρ = 0.1759, P‐value < 0.05, Fig H). […]

Pipeline specifications

Software tools MOMA, GIMME
Application Metabolic engineering
Organisms Mus musculus, Homo sapiens/Mus musculus xenograft
Diseases Colonic Neoplasms, Neoplasms, Colorectal Neoplasms