Transcriptome Profile of Human Colorectal Adenomas

Abstract

Colorectal cancers are believed to arise predominantly from adenomas. Although these precancerous lesions have been subjected to extensive clinical, pathologic, and molecular analyses, little is currently known about the global gene expression changes accompanying their formation. To characterize the molecular processes underlying the transformation of normal colonic epithelium, we compared the transcriptomes of 32 prospectively collected adenomas with those of normal mucosa from the same individuals. Important differences emerged not only between the expression profiles of normal and adenomatous tissues but also between those of small and large adenomas. A key feature of the transformation process was the remodeling of the Wnt pathway reflected in patent overexpression and underexpression of 78 known components of this signaling cascade. The expression of 19 Wnt targets was closely correlated with clear up-regulation of KIAA1199, whose function is currently unknown. In normal mucosa, KIAA1199 expression was confined to cells in the lower portion of intestinal crypts, where Wnt signaling is physiologically active, but it was markedly increased in all adenomas, where it was expressed in most of the epithelial cells, and in colon cancer cell lines, it was markedly reduced by inactivation of the B-catenin/T-cell factor(s) transcription complex, the pivotal mediator of Wnt signaling. Our transcriptomic profiles of normal colonic mucosa and colorectal adenomas shed new light on the early stages of colorectal tumorigenesis and identified KIAA1199 as a novel target of the Wnt signaling pathway and a putative marker of colorectal adenomatous transformation.

An integrated computational and experimental study uncovers FUT 9 as a metabolic driver of colorectal cancer

“…FUT9 is the most highly scored gene and is also strongly supported by the earlier genomic analysis: Its expression is strongly downregulated in colon cancer (Rank‐sum P ‐value = 1e‐22, Fig 2A), it is significantly deleted in colon cancer while not in other cancer types ( Q ‐value = 0.0356, Fig 2B), its low expression is associated with poor survival in colon cancer (Kaplan–Meier (KM) ∆AUC = −0.1206, Fig 2C) (Table 1) [The resulting KM log‐rank P ‐value is 0.1942, likely due to the small sample size of patients expressing FUT9 (only ~15% of patients.)]. Interestingly though, while MTA highly scores FUT9 for all three transformations, FUT9 is not significantly downregulated at early‐stage colon adenomas using paired gene expression of healthy/adenoma samples from Sabates‐Bellver et al (2007) (Paired Student's t ‐test, P ‐value = 0.47, Appendix Fig S1). This suggests that its inactivation may play a significant role only at later stages of colon cancer progression.…”

“…We analyzed three independent transcriptomic datasets including 27 paired healthy/tumor samples from TCGA, 17 paired healthy/tumor samples from Khamas et al (Khamas et al , 2012), and 32 paired healthy/adenoma samples from Sabates‐Bellver et al (2007). In the first step, we ran an MTA analysis on each pair of matched healthy and tumor gene expression samples, yielding a ranked list of genes according to their oncogenic transformation scores ( OTS ) (Materials and Methods).…”

“…Following that, in a second step, an aggregate OTS was assigned to each metabolic gene by considering its scores across all samples and then, in a third step we aggregated the OTS scores of each gene across all three datasets analyzed. We additionally analyzed colon polyp data from Sabates‐Bellver et al (2007), which includes 32 matched healthy and polyp samples. These data enabled us to perform two complementary MTA analyses, one predicting metabolic genes whose knockdown may cause the transformation to the polyp state, and one predicting metabolic genes whose inactivation may cause a further malignant transformation into colon cancer (Materials and Methods, Table EV4).…”

“…This was done by incorporating adenoma gene expression data from Sabates‐Bellver et al (2007) using the GIMME algorithm. (ii) We then sampled 100 flux distributions in the resulting predicted adenoma wild‐type state.…”

“…To predict the effect of FUT9 levels on biomass production, glucose consumption, lactate production, and oxygen consumption, we utilized the GIMME algorithm to simulate metabolic flux of (i) colon adenoma state using the 32 adenoma samples from Sabates‐Bellver et al (2007) (ii) Colon cancer state using 268 cancerous samples from the TCGA. For each of the adenoma and cancer predicted flux distributions, we sampled 100 flux distributions for FUT9 KD and another 100 for FUT9 OE (defined by setting the lower bound of FUT9‐associated reactions to 80% of their maximum), using MOMA algorithm, aiming to minimize the metabolic adjustments after FUT9 perturbations, from the initial adenoma or cancerous metabolic state.…”