• We are sorry, but NCBI web applications do not support your browser and may not function properly. More information
Logo of gutGutCurrent TOCInstructions to authors
Gut. Mar 2005; 54(3): 374–384.
PMCID: PMC1774427

Differential gene expression in colon cancer of the caecum versus the sigmoid and rectosigmoid

Abstract

Background and aims: There are epidemiological, morphological, and molecular differences between normal mucosa as well as between adenocarcinomas of the right and left side of the large bowel. The aim of this study was to investigate differences in gene expression.

Methods: Oligonucleotide microarrays (GeneChip) were used to compare gene expression in 45 single samples from normal mucosa and sporadic colorectal carcinomas (Dukes’ B and C) of the caecum compared with the sigmoid and rectosigmoid. Findings were validated by real time polymerase chain reaction.

Results: Fifty eight genes were found to be differentially expressed between the normal mucosa of the caecum and the sigmoid and rectosigmoid (p<0.01), including pS2, S100P, and a sialyltransferase, all being expressed at higher levels in the caecum. A total of 118 and 186 genes were differentially expressed between normal and right or left sided tumours of the colon, showing more pronounced differences in Dukes’ C than B tumours. Thirty genes differentially expressed in tumour tissue were common to adenocarcinomas of both sides, including known tumour markers such as the matrix metalloproteinases. Keratins 8, 19, and 20 as well as carbonic anhydrases (II, IV, VII) showed side specific expression and were downregulated in left sided tumours whereas teratocarcinoma growth factor and cyclooxygenase 2 (COX-2) were upregulated in left sided adenocarcinomas. Immunohistochemical analysis confirmed differences in side specific expression for cytokeratin 20 and COX-2.

Conclusions: Differences in gene expression between normal mucosa as well as between adenocarcinomas of the caecum and sigmoid or rectosigmoid exist and should be taken into account when examining new targeted therapeutic regimens.

Keywords: gene expression, colon cancer, caecum, sigmoid, rectosigmoid, large bowel, microarray, biomarkers

Multiple differences between right sided (RCC) and left sided (LCC) sporadic colon adenocarcinomas with regard to epidemiological, morphological, and molecular characteristics suggest that the mechanisms of sporadic colorectal carcinogenesis may differ according to tumour location.1 Cancers of the right and left colon may form different but related groups of tumours because of their different embryological origin (midgut and hindgut, respectively) and different exposure to bowel content. Colon cancer has a different prevalence at varying ages, in high and low incidence nations, and in men and women. RCCs are more common in females, LCCs in males.2 There is also a difference in clinical presentation, in prognosis, and possibly in genetic and environmental epidemiology (see review by Iacopetta3). Furthermore, it has been suggested that a mechanism exists that promotes the progression of mucosal lesions to invasive cancers in the left colon and rectum whereas a de novo pathway from depressed type lesions may be implicated in cancers of the right colon.4 No difference has been found in the distribution of Dukes’ stages or in operative mortality between right and left sided sporadic colon cancers. Despite their higher tumour diameter and twofold higher rate of undifferentiated carcinomas, the prognosis of right sided tumours is relatively better than that of left sided tumours, and it has been hypothesised that this could be due to the better blood and lymph supply providing more efficient local tumour defence.5 Recurrence and survival are similar between RCC and LCC6 whereas response to 5-fluorouracil treatment is significantly better in RCC.7

Two studies suggest that molecular differences in gene expression exist between right and left sided colon cancers. Kapiteijn et al showed significantly higher expression of nuclear β-catenin and p53 in rectal cancers compared with proximal cancers.8 Fric et al showed significantly higher expression of cytoplasmic c-erbB2, epidermal growth factor receptor (EGFR), proliferating cell nuclear antigen (PCNA), and dipeptidylpeptidase IV (DPP IV) in right sided sporadic colon cancers compared with left sided cancers.9 Distal tumours display a higher frequency of 17p and 18q allelic loss, p53 accumulation, c-myc expression, and aneuploidy than proximal tumours. Recently, Glebov et al distinguished proximal from distal normal colon mucosa based on gene expression analysis.10

To the best of our knowledge there are no expression data available on differences between adenocarcinomas originating from the proximal or distal part of the colon. As this could have a strong impact on molecularly targeted cancer treatment, we wished to elucidate this aspect and gain insight into differential expression of approximately 7000 human genes of right sided and left sided Dukes’ stage B and C adenocarcinomas as well as normal colon mucosa.

MATERIALS AND METHODS

Tissue samples, patient information, and RNA isolation

Tissue samples, patient information, and RNA isolation are provided in detail as supplementary data (these data can be viewed on the Gut website at http://www.gut.com/supplemental). Samples from the caecum and rectosigmoid or sigmoid were obtained fresh from surgery and immediately transferred to a solution containing sodium dodecyl sulphate and guanidinium isothiocyanate, snap frozen in liquid nitrogen, and stored at −80°C.

Samples consisted of biopsies from the superficial non-necrotic part of tumours and/or normal mucosa biopsies taken from the oral resection margin. All tumour samples were staged as either Dukes’ B (eight from the left colon, five from the caecum) or Dukes’ C (seven from the left colon, five from the caecum).

Supplementary table 1 [triangle] shows detailed clinicopathological information—for example, location of samples in the colon and TNM status (see the Gut website at http://www.gut.com/supplemental). All 15 left sided and 9/10 right sided tumours (90%) were invasive adenocarcinomas; one was an invasive mucinous adenocarcinoma. Six of 10 right sided tumours (60%) were moderately differentiated, 3/10 (30%) were poorly differentiated, and 1/10 (10%) was well differentiated. Ten of 15 left sided tumours (67%) were moderately differentiated, 4/15 (27%) were poorly differentiated, and 1/15 (6%) was well differentiated. The approximate percentages of the volume fractions of tumour cells and stromal cells were semi quantitatively estimated using paraffin embedded diagnostic tissue sections. More than half of the tumour samples showed more than 70% malignant cells. We hypothesise that the percentage of tumour cells is probably higher in the arrayed samples than in the screened paraffin embedded diagnostic histological tissue sections, as the latter represents the whole invasive tumour in the bowel wall. Informed consent was obtained from all patients. All tumours were sporadic. The local scientific ethics commission approved the project.

Table 1
 Fifty eight genes differentially expressed more than threefold (p<0.01), comparing normal mucosa from the caecum to that of the sigmoid or rectosigmoid

Total RNA was isolated from approximately 50 mg of single tissue samples using a Polytron homogeniser followed by treatment with Trizol (Invitrogen, Carlsbad, California, USA) according to the manufacturer’s instructions. GeneChip (Affymetrix Inc., Santa Clara, California, USA) analysis of single samples was carried out on 10 samples from the caecum (65B, 66B, 73B, 120B, 137B, 90C, 126C, 145C, 138C, and 162C), five Dukes’ stage B (median age 76 years) and five Dukes’ stage C (median age 66 years). Each of the tumours was accompanied by a corresponding matched normal mucosa sample at the same location from the same patient (median age 70 years). Matched samples were given the same sample number, differentiated by “N” for normal and “B” or “C” for Dukes’ B or Dukes’ C tumours. Left sided colon samples comprised eight Dukes’ stage B (median age 76 years), seven Dukes’ stage C (median age 68 years), and 10 “normal mucosa” samples (median age 69 years). Five of these tumours (201C, 202B, 203B, 204C, and 208C) were accompanied by a corresponding matched normal sample at the same location from the same patient. The remaining five normal mucosa samples (157N, 161N, 179N, 195N, and 205N) and 10 tumour samples (16B, 237B, 239B, 54B, 127B, 58C, 74C, 85C, 91C, and 96C) were obtained from an independent set of samples of individual patients who underwent resection of the sigmoid or rectosigmoid colon.

cRNA preparation, array hybridisation and scanning, and RT-PCR

cRNA preparation, array hybridisation, and scanning are provided in detail as supplementary data, including supplementary tables 1 [triangle]–7 (see the Gut website at http://www.gut.com/supplemental).

Data analysis and selection of genes

Data analysis and selection of genes is provided in detail as supplementary data (see the Gut website at http://www.gut.com/supplemental). Comparison analysis was done using Microarray Suite 5.0 (MAS 5.0), MicroDB 3.0 (MDB 3.0), and Datamining Tool 3.0 (DMT 3.0) (Affymetrix) applying the Affymetrix specific software “Statistical Expression Algorithms”. Five different comparison groups (A–E) were established and a schematic overview in given in fig 1 [triangle] and described in detail in the supplementary data (see the Gut website at http://www.gut.com/supplemental).

Figure 1
 Schematic overview of the five different comparison groups (A–E). Comparison (A): normal right sided colon mucosa (NR) from the caecum versus normal left sided (NL) colon ...

For all comparisons, several filterings were made to obtain solid and consistent data. To exclude genes with minor or only individual importance, genes were excluded if more than 80% (comparison A) or 70% (B and C) of all datasets were accompanied by a “detection” call of “absent”. Genes were included if more than 80% (B and C) or 70% (D) of the comparisons were accompanied by a “change” call of increased or decreased. For statistical analysis, an Affymetrix software integrated Mann-Whitney U test was applied to the signal data of the groups compared with each other. Significance was set at a p value of p[less-than-or-eq, slant]0.05.

Real time PCR, normalisation of RT-PCR data, and microsatellite analysis

Real time PCR, normalisation of RT-PCR data, and microsatellite analysis are described in detail as supplementary data (see the Gut website at http://www.gut.com/supplemental).

Immunohistochemistry

Formalin fixed paraffin embedded sections from the normal mucosa and matched tumour tissue were stained with monoclonal mouse antihuman cyclooxygenase 2 (COX-2) (cat No 35-8200; Zymed, AH-diagnostisk, Denmark), diluted 1:300, or monoclonal mouse antihuman cytokeratin 20 (cat. No M7019; Dako Cytomation, Denmark), diluted 1:100, as described in detail in the supplementary data (see the Gut website at http://www.gut.com/supplemental).

RESULTS

Using Affymetrix GeneChip oligonucleotide microarrays, we analysed gene expression of 45 colonic samples. The expression profile of 10 sporadic adenocarcinomas of Dukes’ B and C from the right side and 15 from the left side were compared with 20 normal colon mucosa samples, 10 matched samples from the right and 10 partly matched samples from the left side. Gene expression differences were determined between: (A) normal mucosa of the right and left side; (B) normal mucosa and Dukes’ B or C adenocarcinomas of the right side; (C) normal mucosa and Dukes’ B or C adenocarcinomas of the left side; (D) Dukes’ B or C adenocarcinomas of the right and left side; and finally (E), differentially expressed genes in the right sided colon from comparison B with those in the left sided colon from comparison C (fig 1 [triangle]).

Comparison A: normal caecum versus sigmoid/rectosigmoid

By comparing normal mucosa samples from 20 different patients—namely, 10 right sided from the caecum to 10 left sided from the sigmoid or rectosigmoid—we identified 160 genes showing site specific differential gene expression, being increased or decreased more than 1.5 fold (p<0.05, Mann-Whitney U test). Fifty eight genes with a p value of <0.01 are shown in table 1 [triangle]; 12 of these genes with fold changes more than threefold the median signal are labelled with an asterisk.

The gene encoding the pS2 protein, maintaining the mucosal surface barrier and stimulating repair processes, showed 7.7-fold higher expression in the left than in the right colon. Other differentially expressed genes with a consistent difference were calcium binding protein S100P (7.6-fold), homeodomain protein HOXB13 (7.2-fold), defensin 5 (6.1-fold), Gal-beta (1-3/1-4) GlcNAc alpha-2.3-sialyltransferase (5.6-fold), 3-beta-hydroxysteroid dehydrogenase gene (5.4-fold), and HE4 extracellular proteinase inhibitor homologue (5.3-fold). Also, the beta subunit of creatine kinase-B, fibrinogen A alpha polypeptide alt. splice 3 E (3.7-fold), cathepsin E, and protein tyrosine phosphatase were among the genes showing more than threefold significantly different expression between the two groups.

Comparison B: normal versus tumour caecum

By comparing normal mucosa of the caecum to matching caecum adenocarcinomas staged as Dukes’ B or C and derived from the same patient, we identified 118 genes significantly up or downregulated more than 2.8-fold (p<0.05, Mann-Whitney U test) in adenocarcinomas compared with normal mucosa (see supplementary table on the Gut website at http://www.gut.com/supplemental). Seventy three showed fold changes of more than fourfold, and of these, 22 genes with a p value of <0.01 are shown in table 2 [triangle].

Table 2
 Twenty two genes differentially expressed more than fourfold (p<0.01), comparing normal mucosa to matched Dukes’ B or C adenocarcinomas of the caecum

A characteristic finding was that most genes (n = 15) were downregulated in carcinomas compared with normal mucosa, and only a few were upregulated (n = 7). Several matrix metalloproteinases, such as MMP1, MMP3, and MMP10, located in the extracellular space and involved in proteolysis and peptidolysis were highly upregulated in carcinomas, as well as E1A enhancer binding protein (E1A-F) (fourfold) and calcium binding protein S100P. TRPM-2 protein (fivefold), complement protein component C7 (fivefold), and NAD+ dependent 15 hydroxyprostaglandin dehydrogenase (PGDH; 16-fold) showed decreased expression.

Comparison C: normal versus tumour sigmoid/rectosigmoid

We compared normal mucosa from the left side of the colon to matching adenocarcinomas of Dukes’ B and C from the same patient in five cases, and in those 10 cases where a matching normal sample was not present, we compared each of the 10 tumours to each of five single normal samples (for details see material and methods). We identified 186 genes significantly differentially expressed more than 2.8 fold (p<0.05, Mann-Whitney U test) from the normal mucosa to Dukes’ B or Dukes’ C tumours (see supplementary table 3 [triangle] on the Gut website at http://www.gut.com/supplemental). The majority confirmed our recently published findings made on pools of colorectal cancer samples11; for example, downregulation of nuclear encoded mitochondrial genes such as TST thiosulfate sulfurtransferase (rhodanese) (4.5-fold) and the SCAD gene 5′ UTR exon 1 and 2 (sevenfold). The 42 most important genes with a fold change [gt-or-equal, slanted]4 (p<0.01) in both Dukes’ B and Dukes’ C are shown in table 3 [triangle]. Other genes (for example, osteopontin) which showed changes have been omitted here because changes were found to be [gt-or-equal, slanted]4 fold in either Dukes’ C or Dukes’ B but not in both.

Table 3
 Forty two genes differentially expressed more than fourfold (p<0.01), comparing normal mucosa to Dukes’ B or C tumours from the sigmoid or rectosigmoid

Thirty genes were found to be downregulated in cancer, such as GCAP-II (33-fold), carbonic anhydrase IV (33-fold), and DTD sulfate transporter gene (20-fold). Only 12 genes were upregulated, among these microsomal dipeptidase (MDP4, MDP7; 20-fold) and interleukin 8/MDNCF (15-fold). As a novel finding we found that carbonic anhydrase VII (CA VII) was decreased more than fourfold from normal to Dukes’ B and C adenocarcinomas.

Comparison D: tumours from the caecum versus sigmoid/rectosigmoid

Within each of the Dukes’ B and C stages, we compared all adenocarcinomas from the left side with all of those from the right side of the colon. We identified five genes in Dukes’ B, 39 in Dukes’ C, and five genes in both B and C, that showed significant differences in expression levels (p<0.05) with an average fold change of 2.8, corresponding to a total of 44 genes differentially expressed in left and right sided tumours (see supplementary table 4 [triangle] on the Gut website at http://www.gut.com/supplemental). Among these 44 genes, 16 showed more than threefold upregulation or more than fourfold downregulation (table 4 [triangle]).

Table 4
 Sixteen genes differentially expressed more than threefold (p<0.05), comparing Dukes’ B and C adenocarcinomas of the caecum with those of the sigmoid or rectosigmoid

Differential gene expression was more common in Dukes’ C than in Dukes’ B, and among the genes were caldesmon 1, involved in cellular mitosis and receptor capping, modulator recognition factor 2 (a DNA binding factor), ARHB, involved in signal transduction, transgelin 11 (SM22-alpha), and D component of complement (adipsin, involved in proteolysis and peptidolysis), all five showing higher expression in left sided carcinomas. In contrast, homeobox A5 protein, a sequence specific transcription factor, was more strongly expressed in Dukes’ C adenocarcinomas of the right side of the colon.

Comparison E: comparison to identify genes in common or differentially expressed in right sided versus left sided tumours

A total of 186 genes previously identified to be differentially expressed from normal mucosa to tumour in the left side of the colon were compared with 118 genes identified in the right colon. This resulted in 30 common cancer genes being significantly differentially expressed more than threefold (accompanied by a p value of <0.05) in at least one of the Dukes’ in both right sided as well as left sided tumours. These may make ideal colonic tumour markers (table 5 [triangle]).

Table 5
 Thirty common cancer genes differentially expressed more than threefold accompanied by a p value of <0.05 in at least one of the Dukes’ compared with normal mucosa

Validation of the results by real time PCR applied to aminopeptidase N/CD13, SCAD, and PCK1 is shown in fig 2 [triangle] where single GeneChip analyses were compared with real time PCR analyses. Additionally, we identified cancer genes being characteristic for one side of the colon only. Eighty eight genes shown in supplementary table 5 [triangle] (on the Gut website at http://www.gut.com/supplemental) were significantly differentially expressed exclusively in right sided tumours, such as factor XIII subunit a and calcium binding protein S100P (fig 2 [triangle]), suggesting a more crucial role in caecal adenocarcinomas. A total of 156 genes shown in supplementary table 6 (on the Gut website at http://www.gut.com/supplemental) were significantly differentially expressed only in left sided tumours. Among these were MDP4/MDP7 and the interferon inducible protein “9-27”. Differences in expression in most of the growth factors were seen in the left colon such as upregulation of teratocarcinoma derived growth factor (>7 fold). Furthermore, the COX-2 gene was more than sixfold higher in Dukes’ C tumours of the left colon, and did not show a significant difference in right sided tumours. Most strikingly, expression of keratins 8, 19, and 20 was severely reduced in the left colon but did not show significant differences in the caecum.

Figure 2
 Comparison of single GeneChip analyses with real time polymerase chain reaction (PCR) analyses. Expression analyses of five selected genes using single samples of normal colon mucosa and adenocarcinomas of Dukes’ stages ...

Microsatellite analysis

Microsatellite analysis was performed on microdissected tumour tissue, as described in materials and methods in the supplementary data (on the Gut website at http://www.gut.com/supplemental). Of 10 samples, where the amount of tissue allowed microdissection, only one sample (No 120B) was found to be highly microsatellite instable, the other nine samples being microsatellite stable (MSS), as listed in supplementary table 1 [triangle] (on the Gut website at http://www.gut.com/supplemental). The fact that all except one of the tumours were stable with regard to microsatellites BAT25 and BAT26 (MSS) strongly supports the conclusion that the differences described here do not result from differences in microsatellite stability but have to be regarded as differences characterising the function and behaviour of tumours originating from the caecum or sigmoid and rectosigmoid.

Immunohistochemical analysis

Immunostaining was applied to paraffin embedded specimen from eight of the 10 right sided and 11 of the 15 left sided tumours where snap frozen material had been previously analysed on microarrays to enable a comparison of RNA and protein expression. The 19 tumours where selected based on the availability of their matching normal mucosa from the oral resection edge.

Figure 3 [triangle] (A, B) shows five right and five left sided tumours with their matching normal mucosa stained with COX-2. In the right colon, COX-2 was moderately to strongly expressed in normal mucosa, mostly throughout the entire epithelium as well as in right sided tumours. Comparing normal tissue with tumour, we detected upregulation (in one of eight tissue sections), downregulation (1/8), or about equal expression in normal tissue and tumour (6/8). In the left side of the colon, COX-2 was not or only very weakly expressed in normal mucosa and was upregulated from normal mucosa to tumour. Comparing normal mucosa to tumour, we observed strong upregulation in more than 50% of cells (3/11), moderate upregulation in more than 50% of cells (5/11), and very strong upregulation in single cell groups corresponding to less than 10% of cells (3/11).

Figure 3
 Immunohistochemistry of formalin fixed paraffin embedded sections of Dukes’ B and C adenocarcinomas and their matching normal mucosa (N). Sample numbers of tumours refer to samples previously analysed on microarrays. Cyclooxygenase ...

Figure 3 [triangle] (C, D) shows five right and five left sided tumours with their matching normal mucosa stained with cytokeratin 20 (KRT20). KRT20 was strongly expressed in the luminal epithelium of normal mucosa of both sides. Comparing normal tissue to tumour of the right side, we detected strong upregulation with staining of more than 50% of cells (2/8), downregulation (4/8) with staining of less than 10% of cells, or about equal expression in normal and tumour with staining of approximately 30–40% of tumour cells (2/8). Comparing normal mucosa to tumour on the left side, we observed upregulation with staining of more than 50% of cells (2/11), downregulation with staining of less than 10% of cells (7/11), or about equal expression in normal mucosa and tumour with staining of approximately 30–40% of tumour cells (2/11). Staining of tumour cells was very heterogeneous in most of the tumours.

DISCUSSION

While published data on right sided versus left sided colon cancers are lacking, colon cancers per se have previously been compared with normal mucosa. In this study, we identified differences in gene expression in the colon that characterised left and right sided normal mucosa and adenocarcinomas. Using statistical algorithms provided by the Affymetrix software, we identified sets of genes differentially expressed, as well as genes in common, between right sided and left sided adenocarcinomas.

In this study, we analysed a total of 45 samples (20 normal and 25 tumour samples). The complexity of our study is comparable with colon cancer expression analyses previously described by Alon et al, analysing 22 normal and 40 tumour samples, and by Notterman et al, analysing 18 adenocarcinomas and four adenomas with paired normal tissue, both using Affymetrix GeneChips, as previously discussed.12,13 The reliability of our data (for example, with regard to comparisons of left sided normal mucosa to Dukes’ B and C tumours) is supported by the fact that we confirmed identification of various genes previously identified by other techniques. Metallothionein, fibronectin, and SPARC, for example, had previously been shown to be differentially expressed in normal tissue and tumour by Zhang et al, using the SAGE technique on two normal and two tumour samples.14 Furthermore, we confirmed differential expression of more than 70% of genes previously identified by GeneChip analyses of pooled samples of the left side of the colon.11 In addition, these results were also highly comparable with data previously published by Notterman et al (for example, upregulation of MGSA from normal to tumour tissue and downregulation of guanylin or chromogranin A).13 Reliability of the results with regard to differences between the right and left colon was further supported by expression analysis using RT-PCR showing high reproducibility of expression levels detected by the arrays. Previous studies have, in most cases, not taken into account the Dukes’ stage or location within the colon where the samples originated. Obviously, adding more subclasses to the material inevitably leads to fewer samples per class and the main findings of this paper should be repeated on larger material.

A grouped Mann-Whitney test intrinsic to the Affymetrix software DMT 3.0 was used for statistical analyses. As some of the data were paired (different tissues from the same patient) a Wilcoxon matched pairs test may have been more appropriate for these cases and this may have been a limitation of our statistical analyses. On the other hand, some of the samples were grouped (tissue from different patients) and a Mann-Whitney test had to be applied. However, such “breaking of matching” is more likely to make the results more, rather than less, comparable between tumour and normal tissue, and so this limitation is not of major importance as it is not likely to explain any of the observed differences.

We focused the analysis on adenocarcinomas of Dukes’ stages B and C as these are the most challenging stages in colon cancer, with the possibility of curative treatment. Most of the factors that may influence gene expression were taken into account but for array analysis it was not possible to match all samples with their normal mucosa (as could be achieved for immunostainings) or to match samples with regard to sex, as most of our right sided colon cancer patients were female. In general, colon cancer affects males and females equally but some studies indicate that right sided colon cancer affects more women than men.2 Comparison of right versus left sided normal mucosa showed 58 genes differentially expressed, 12 with fold changes more than threefold and none located on the Y chromosome. A comparison of right and left sided adenocarcinomas showed two genes located on the Y chromosome and significantly higher expressed in the group of left sided Dukes’ C but not Dukes’ B. RPS4Y and SMCY show high fold changes of 23- and 12-fold but SMCY increases only up to a signal of 50, which is close to the detection level. From these data there is no evidence that the imbalance between males and females influences the results profoundly.

Genes such as β-catenin, c-erbB2, EGFR, PCNA, or DPP IV, previously shown to be differentially expressed in right and left colon cancers,9 were not identified as significant in this study but did show side differences when less stringent selection criteria were applied. There are many factors affecting gene expression analysis, such as ischaemic delay, defined as the period of time from clamping of blood vessels to snap freezing, ratio of tumour versus non-tumour cells, RNA extraction method and quality of RNA (28s/18s ratio), type of array used (c-DNA arrays, nylon membrane, oligonucleotide arrays), amplification, labelling (Cy3/Cy5 or d-UTP-Biotin/SAPE) and labelling efficiency, sensitivity and detection threshold, software used for analysis, and statistical significance criteria.

The most predominant differences between normal left and right colon mucosa were higher expression in left sided mucosa of genes such as pS2 protein, calcium binding protein S100P, HOXB13, SIAT4C, and WFDC2. This agrees with previous findings of a 7.7-fold higher expression of pS2 protein15 and approximately fourfold higher expression of HOXB13 and S100P10 in the left colon. This agreement is remarkable because different platforms have been used for analysis and two thirds of the samples in the study of Glebov et al were HNPCC samples. Homeobox proteins such as HOXB13 or HOXA5 encode transcription factors and upregulate tumour suppressor p53 and may therefore be involved in side specific tumorigenesis.

Defensin 5 was found to be expressed sixfold higher in right sided mucosa which matches the proposal that the right colon provides more efficient local tumour defence, maintaining the mucosal barrier.5,16 We hypothesise that the right sided colon mucosa provides protection against carcinogens by defensin 5 expression, leading to less frequent carcinogenesis compared with the left side. Remarkably, the site on chromosome 8p housing the defensin gene is frequently lost in liver metastases from primary colon cancers.15

The majority of genes found to be differentially expressed from normal mucosa to Dukes’ B or C of the left side confirmed our recently published findings performed on pooled samples.11 Genes such as MDP4/MDP7 and interleukin 8/MDNCF were strongly upregulated, and several nuclear encoded mitochondrial proteins such as rhodanese or SCAD were strongly downregulated in tumours. We also identified reduced levels of several carbonic anhydrases (CA) such as CAVII or CAIV which have not previously been described indepth in colon cancer. CAIV, downregulated by up to 33-fold in left sided tumours, is responsible for maintenance of pH and ion equilibrium. Takenawa et al showed that low level expression of CAIV and aquaporin 1 in renal cell carcinomas was associated with poor survival.17

Notterman et al analysed differential gene expression between the normal colon and tumour, without discriminating between the right and left side.13 In terms of expression differences between normal mucosa and tumour of the left colon, our study is highly comparable with that of Notterman et al. In both studies, prior to analyses samples were defined with regard to Dukes’ stage, snap frozen bulk tissue samples yielded high quality RNA, identical labelling and GeneChips were used, and the data were analysed using the Mann-Whitney U test. Notterman et al identified CAIV as being downregulated by 38-fold from normal colon to tumour, which is identical to our results. In summary, this strongly supports the hypothesis that a decrease in CAIV expression is linked to carcinogenesis and colon cancer progression.

Expression of genes such as COX-2, caldesmon 1, adipsin, transgelin 11, and ARHB was found to be higher in left sided compared with right sided adenocarcinomas. A previous study showed a better effect of chemoprevention with non-steroidal anti-inflammatory drugs on right sided than on left sided adenocarcinomas,3 and the sixfold higher expression of COX-2 may explain failure to prevent this, as a higher dose may be needed to inhibit the high levels of this molecule in left sided malignant lesions. Loss of transgelin gene expression may be an important early event in tumour progression as a consequence of deregulation of RAS gene expression through RAF independent pathways.18 Interestingly, ARHB (RhoB), located on chromosome 2pter-p12, is one of three RAS homologue gene family members and is known as an oncogene.19

From normal mucosa to Dukes’ B and C of the caecum, we found that TRPM-2 (clusterin) and PGDH were strongly downregulated whereas several matrix metalloproteinases such as MMP1, MMP3, and MMP10 were upregulated, as seen previously in left sided tumours. MMPs are enzymes responsible for extracellular matrix degradation, playing a role in cancer progression and metastatic spreading. MMP1 expression is associated with a poor prognosis in colorectal cancer.20 One possible therapeutic approach for patients with colon cancer, mainly Dukes’ C, could therefore be administration of specific MMP inhibitors to prevent distant metastases and prolong survival,21 as has been shown by inhibition of MMP2 expression in mouse xenograft experiments.22 Circulating proenzymes of MMPs have been described as possible serum markers, and proMMP-9, but not proMMP-2 identified here, was found to be significantly higher in cancer sera versus normal sera.23 From a clinical approach, we suggest analysis of sera levels of the MMPs identified here, as these molecules seem to be ideal general colonic tumour markers reflecting the presence of both left and right sided colonic tumours.

In conclusion, the 30 genes identified in adenocarcinomas of both sides have to be regarded as general tumour markers. The present data and our previously published LOH analyses11 strongly support the hypothesis that genes such as aminopeptidase N (CD 13), sulfate transporter DTD, SCAD, or PCK1 should be regarded as potential new tumour suppressors requiring further investigation.

Paraffin embedded tissue sections from Dukes’ B and C tumours and their matching normal mucosa were subjected to immunohistochemical analysis for COX-2 and cytokeratin 20 (KRT20). Microarray analysis showed a significant decrease in KRT20 from normal mucosa to tumour in the left side of the colon. Immunostaining confirmed the difference seen between the two sides of the colon as KRT20 was strongly downregulated in 80% of left sided tumours compared with 50% of right sided tumours. In general, KRT20 staining was found to be very heterogeneous within the tumours.

COX-2 microarray data showed that COX-2 was upregulated from normal to Dukes’ C in right as well as left sided tumours, but the increase was significant only for left side (p<0.005). Immunostainings support the microarray based findings to date, that COX-2 is not or very weakly expressed in left sided mucosa but upregulated in matching tumours. In contrast, COX-2 is expressed with the same intensity in right sided normal mucosa compared with matching tumours. COX-2 is heterogeneously expressed within a tumour, as only some groups of cells within a tumour are stained. In conclusion, the microarray based findings were confirmed by immunohistochemistry but an absolute quantitative comparison between RNA expression on microarrays and protein expression on tissue specimen is not possible for KRT20 and COX-2 due to their heterogeneous staining patterns.

The existence of a side specific expression difference for COX-2, having been identified by microarray analysis and confirmed by immunohistochemistry in this study, has recently been reported by Nasir and colleagues.24 Immunohistochemical staining applying a COX-2 polyclonal antibody on 18 right sided versus 18 left sided adenocarcinomas showed that COX-2 positivity was significantly higher for left compared with right sided tumours.

We conclude that differences in gene expression between normal mucosa as well as adenocarcinomas of the caecum and sigmoid and rectosigmoid colon clearly exist, and we hypothesise that the difference in gene expression could be related to differences in tumour development and the prognosis of patients.

The emerging treatments directed towards specific molecular targets should emphasise the differences seen in right and left sided tumours of the colon. We suggest that some of the highly expressed molecules that are in both left and right sided colonic adenocarcinomas may be promising new potential serum markers and therapy targets.

Supplementary Material

[Web-only Tables]

Acknowledgments

We thank Bente Devantier, Ing Lis Thorsens, and Annette B. Nielsen for their technical assistance, and also as project-nurse Edith Kirkedahl Nielsen at Aarhus Sygehus for collection of colon tissue samples. The study was supported by funds from the Karen Elise Jensen Foundation, the Danish Research Council, AROS Applied Biotechnology Aps, Aarhus, the University and County of Aarhus, the Nordic Cancer Union, and the European Union’s 5th frameprogram (European Community, No QLG2-CT-2001-01861).

Abbreviations

  • LCC, left sided colon cancer
  • RCC, right sided colon cancer
  • UG cluster, UniGene cluster (http://www.ncbi.nlm.nih.gov/UniGene)
  • EGFR, epidermal growth factor receptor
  • PCNA, proliferating cell nuclear antigen
  • DPP IV, dipeptidylpeptidase IV
  • RT-PCR, reverse transcription-polymerase chain reaction
  • COX-2, cyclooxygenase 2
  • MSS, microsatellite stable
  • CA, carbonic anhydrase
  • MMP, matrix metalloproteinase

Notes

Conflict of interest: None declared.

REFERENCES

1. Campbell F, Appleton MA, Shields CJ, et al. No difference in stem cell somatic mutation between the background mucosa of right- and left-sided sporadic colorectal carcinomas. J Pathol 1998;186:31–5. [PubMed]
2. Distler P, Holt PR. Are right- and left-sided colon neoplasms distinct tumors? Dig Dis 1997;15:302–11. [PubMed]
3. Iacopetta B . Are there two sides to colorectal cancer? Int J Cancer 2002;101:403–8. [PubMed]
4. Konishi K, Fujii T, Boku N, et al. Clinicopathological differences between colonic and rectal carcinomas: are they based on the same mechanism of carcinogenesis? Gut 1999;45:818–21. [PMC free article] [PubMed]
5. Reifferscheid M, Fass J, Hartung R, et al. Special aspects of right colon cancer. Langenbecks Arch Chir 1987;371:193–200. [PubMed]
6. Tomoda H, Taketomi A, Baba H, et al. The clinicopathological characteristics and outcome of patients with right colon cancer. Oncol Rep 1998;5:481–3. [PubMed]
7. Elsaleh H, Joseph D, Grieu F, et al. Association of tumour site and sex with survival benefit from adjuvant chemotherapy in colorectal cancer. Lancet 2000;355:1745–50. [PubMed]
8. Kapiteijn E, Liefers GJ, Los LC, et al. Mechanisms of oncogenesis in colon versus rectal cancer. J Pathol 2001;195:171–8. [PubMed]
9. Fric P, Sovova V, Sloncova E, et al. Different expression of some molecular markers in sporadic cancer of the left and right colon. Eur J Cancer Prev 2000;9:265–8. [PubMed]
10. Glebov OK, Rodriguez LM, Nakahara K, et al. Distinguishing right from left colon by the pattern of gene expression. Cancer Epidemiol Biomarkers Prev 2003;12:755–62. [PubMed]
11. Birkenkamp-Demtroder K, Christensen LL, Olesen SH, et al. Gene expression in colorectal cancer. Cancer Res 2002;62:4352–63. [PubMed]
12. Alon U, Barkai N, Notterman DA, et al. Broad patterns of gene expression revealed by clustering analysis of tumor and normal colon tissues probed by oligonucleotide arrays. Proc Natl Acad Sci U S A 1999;96:6745–50. [PMC free article] [PubMed]
13. Notterman DA, Alon U, Sierk AJ, et al. Transcriptional gene expression profiles of colorectal adenoma, adenocarcinoma, and normal tissue examined by oligonucleotide arrays. Cancer Res 2001;61:3124–30. [PubMed]
14. Zhang L, Zhou W, Velculescu VE, et al. Gene expression profiles in normal and cancer cells. Science 1997;276:1268–72. [PubMed]
15. Paredes-Zaglul A, Kang JJ, Essig YP, et al. Analysis of colorectal cancer by comparative genomic hybridization: evidence for induction of the metastatic phenotype by loss of tumor suppressor genes. Clin Cancer Res 1998;4:879–86. [PubMed]
16. Wehkamp J, Schwind B, Herrlinger KR, et al. Innate immunity and colonic inflammation: enhanced expression of epithelial alpha-defensins. Dig Dis Sci 2002;47:1349–55. [PubMed]
17. Takenawa J, Kaneko Y, Kishishita M, et al. Transcript levels of aquaporin 1 and carbonic anhydrase IV as predictive indicators for prognosis of renal cell carcinoma patients after nephrectomy. Int J Cancer 1998; %20 79:1–7. [PubMed]
18. Shields JM, Rogers-Graham K, Der CJ. Loss of transgelin in breast and colon tumors and in RIE-1 cells by Ras deregulation of gene expression through Raf-independent pathways. J Biol Chem 2002;277:9790–9. [PubMed]
19. Madaule P, Axel R. A novel ras-related gene family. Cell 1985;41:31–40. [PubMed]
20. Murray GI, Duncan ME, O’Neil P, et al. Matrix metalloproteinase-1 is associated with poor prognosis in colorectal cancer. Nat Med 1996;2:461–2. [PubMed]
21. Aparicio T, Kermorgant S, Dessirier V, et al. Matrix metalloproteinase inhibition prevents colon cancer peritoneal carcinomatosis development and prolongs survival in rats. Carcinogenesis 1999;20:1445–51. [PubMed]
22. Oba K, Konno H, Tanaka T, et al. Prevention of liver metastasis of human colon cancer by selective matrix metalloproteinase inhibitor MMI-166. Cancer Lett 2002;175:45–51. [PubMed]
23. Pucci-Minafra I, Minafra S, La Rocca G, et al. Zymographic analysis of circulating and tissue forms of colon carcinoma gelatinase A (MMP-2) and B (MMP-9) separated by mono- and two-dimensional electrophoresis. Matrix Biol 2001;20:419–27. [PubMed]
24. Nasir A, Kaiser HE, Boulware D, et al. Cyclooxygenase-2 expression in right- and left-sided colon cancer: a rationale for optimization of cyclooxygenase-2 inhibitor therapy. Clin Colorectal Cancer 2004;3:243–7. [PubMed]

Articles from Gut are provided here courtesy of BMJ Group
PubReader format: click here to try

Formats:

Related citations in PubMed

See reviews...See all...

Cited by other articles in PMC

See all...

Links

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...