table of contents table of contents

Accumulating databases in human genome research have enabled integrated genome-wide study on …

Home » Biology Articles » Anatomy & Physiology » Anatomy, Human » A transcriptome anatomy of human colorectal cancers » Discussion

- A transcriptome anatomy of human colorectal cancers

The release of the human genome sequences and accumulating databases as well as technological development for detecting large-scale gene expression profile have enabled integrated genome-wide study on complicated pathological states, such as cancers. Tissue or organ-specific gene expression patterns as well as genome mapping have been reported via database mining including dbEST and SAGEMap [12,18]. These computational methods were proved to be effective for comparisons among tissues under different physiological or pathological states [19]. We thus developed an integrated dbEST mining procedure, GetUni. In this study, we utilized GetUni for genome-wide transcriptome analysis on normal mucosa, cancers and related precancerous lesions of colon. The efficiency of this software package was cross validated by the profiles of UniGene cDNA based libraries using the Xprofile online tool. However, we did not focus on differential gene expression patterns among these states since ESTs from dbEST were made by various methods by contributors. To our knowledge, this is the first report on computational genome-wide transcriptome analysis.

In the initial analysis of these 4 sets of transcriptome, we noticed a prominent phenomenon that all genes or gene transcripts from N, IBD and A were found in cancers. Besides the fact that the total ESTs in T were much more than those in the other 3 libraries, one of the underlying paradoxical interpretations was the complicated heterogeneity of colorectal cancers with regard to histogenesis, morphology and molecular genetics, etc. ESTs were highly redundant cDNA fragments derived from either the 3'- or 5'- regions of human genes. For example, there were 20,370 and 279,163 ESTs, and finally 4,108 and 14,879 genes in N and T library, respectively. The average ESTs per gene between T and N (18.7622, 279,163/14,879 versus 4.9586, 20,370/4,108) was approximately 5 and amounted to 7 after normalization of EST clustering efficiency (51.34% in N versus 71.86% ESTs in T which could be clustered into UniGenes, respectively). Considering that there are about 13 times more ESTs in T than in N, it is reasonable to believe that at least 50% increased ESTs in T library could be ascribed to redundancy, suggesting that increased EST was not the unique answer for the broad pattern of gene expression in cancers. Multiple pathways contributing to colorectal cancers were well established including adenoma-carcinoma sequence, inflammatory bowel disease-dysplasia-cancer, hyperplastic polyposis-intraepithelial neoplasia (IEN)-cancer or juvenile polyposis-IEN-cancer as well as direct malignant transformation from normal mucosa (de novo) [1]. As for morphological features, various histological subtypes with varied differentiation were fully accepted including tubular adenocarcinoma, mucinous carcinoma, signet-ring cell carcinoma, medullary carcinoma, undifferentiated carcinoma, etc. A long standing concept in cancer biology is that tumours arise and grow as a result of "tumour stem cells" or "stem cells" (multipotent progenitor cells with the capacity for self-renewal) with multiple additional mutations [20]. Consistent with this idea, a defined minority of these cells might be able to proliferate, differentiate, dedifferentiate and transdifferentiate, resulting in heterogenous gene expression patterns in cancers.

This study also found that transcript variants were quite common since 2,355 out of 14,879 genes had at least 2 transcripts. And, we believe that with the ever increasing ESTs in dbEST, more transcript variants would be discovered in these colonic tissues. In the finished human genome, a big surprise is that there are not more than 25,000 genes in the human genome, barely more than the worm Caenorhabditis elegans [5]. Considering the myriad cellular processes that keep our body functioning, a clear and reinforcing realization is that many genes encode more than one protein, a theory replacing the old notion of one gene one protein. One way that human genome performs such complex functions with so few genes is alternative splicing, which plays important roles in development, physiology, and disease. A genome survey of human alternative pre-mRNA splicing indicated that at least 74% of human multi-exon genes are alternatively spliced [21]. Furthermore, another intriguing phenomenon in this preliminary study is that the average transcripts of individual gene in T are higher than those in the other profiles, indicating that increased alternative splicing might be an optimal option for colorectal cancer considering their more complicated biological behaviours and functions than those of normal or benign lesions in colon tissue.

Uexpectedly, the enriched genes of ribosome in KEGG pathway were highest in 2 precancerous lesions, A and IBD, and lowest in cancer as suggested by GOTM analysis when we compared among these 4 libraries. Ribosome proteins, the major components of ribosome which is the protein synthesis center in a cell, play critical roles in physiological and pathological situations. The fundamental physiological function of colonocytes is secretion, which involves many ribosomes, as evidenced by electronic microscopy. In addition, there is highly active renewal and proliferation in crypt cells. But in cancer, this secretary capability was lost or or impaired due to dedifferentiation, resulting in the limited enrichment of ribosome genes in cancers despite that other oncogenic proteins might be actively synthesized with the absolute number of ribosome genes remaining high. This hypothesis was supported by a recent immunohistochemical study, in which 10 of 12 ribosome proteins were stained stronger in normal mucosa than in colon cancers [22]. As in adenoma and IBD, 2 intermediate states between normal mucosa and cancers, we hypothesized that the ability of secretion largely remained and that there was additional new protein synthesis to maintain the transformed phenotype in IEN or dysplasia. In our previous analysis on differentially expressed genes between colonic adenoma-normal mucosa, 6 out of 62 differentially expressed genes were ribosome proteins [23]. Ribosome protein S11 and L7 were upregulated in colonic adenomas rather than in normal mucosa or cancers as indicated by microarray [24]. An early study also demonstrated that increased mRNA levels of several ribosomal proteins were present in colorectal tumors and polyps [25]. All these findings implicated that increased synthesis of ribosomes might be an important indicator of precursor lesions of colorectal cancers.

Other findings in this study were also intriguing. Genes in the process of the KEGG Glycolysis/Gluconeogenesis pathway were significantly more enriched in adenomas and IBD than in cancers. This is consistent with a recent report on ApcMin/+ mouse model of colon tumours (26). Aberrant glucose metabolism might be emerged in the precancerous stage, earlier than before expected. Five thioredoxin family members were present in library A. These enriched genes might play important roles in anti-oxidative injury, inhibition of apoptosis, cell proliferation and differentiation [27,28]. Totally, 111 7-transmembrane receptor (rhodopsin family) superfamily members were found in cancers. Small GTP-binding protein-coupled receptors, endocrinal or neuroendocrinal receptors, and cytokine receptors were included in this catalogue. Their role in colon cancer was well documented recently [29,30].

Finally, we analyzed the expression of 8 putative genes in colorectal cancers. The preliminary data suggested that all these genes were variably expressed in cancer tissues and cell lines. Particularly, SOX9 was upregulated in most colorectal cancers. It is in consistent with our latest immunostaining results (unpublished data), indicating that SOX9 may play oncogenic roles and serve as an independent adverse prognosticator in colorectal cancers. A recent study revealed that SOX9 was an intestinal crypt specific transcription factor and downstream target of β-catenin [31]. SOX9 may thus be a potential target gene for prognostic assessment and therapeutic intervention in colorectal cancers.

rating: 3.00 from 1 votes | updated on: 25 Aug 2008 | views: 9377 |

Rate article: