PATIENTS AND TISSUES

Colonic biopsy specimens were obtained by endoscopy from 21 patients with ulcerative colitis (UC) (in three patients colonic biopsy specimens were obtained at different time points during the active and inactive phases of the disease), from nine patients with Crohn's disease, five patients with ischaemic colitis, one patient with amoebic colitis, and six normal controls (table 1). The diagnosis of UC and Crohn's disease was based on established endoscopic and histological criteria.17 The diagnosis of ischaemic colitis was based on clinical features and established endoscopic criteria,18 while that of amoebic colitis was based on the presence of the organism. Inactive UC represented cases with endoscopically healed areas and tissue specimens featuring crypt atrophy or distortion with no neutrophil infiltration. On the other hand, active UC included cases with endoscopically inflamed areas of low to moderate grade (increased granularity and friability of the mucosa) but areas of erosion and ulceration were excluded from analysis. Histological inflammatory activity was evaluated based on the degree of neutrophil infiltration according to Matts' grade19: grades 1 and 2 were classified as active in this study. Informed consent was obtained from all patients and control subjects (patients with colorectal cancer and polyps) before biopsy. The study protocol was approved by the Institutional Review Committee for Research on Human Subjects. Demographic features of the patients and history of drug therapy up to the time of colonoscopy are summarised in table 1.

CELL LINE, CYTOKINES, AND REAGENTS

The human colon cancer cell line (HT29) was obtained from the American Type Culture Collection (Rockville, Maryland, USA). Recombinant human (rh) IL-1β, rhIL-6, interferon γ (rhIFN-γ), and tumour necrosis factor α (rhTNF-α) were purchased from Biosource (Camarillo, California, USA). rhIL-4 was purchased from PharMingen (San Diego, California, USA) and 3-morpholinosydnonimine (SIN-1) from Sigma (St Louis, Missouri, USA).

ISOLATION OF COLONIC CRYPT AND SURFACE EPITHELIUM

Colonic crypts and surface epithelia were isolated from colonic biopsy specimens in sheets consisting of both crypt and surface epithelium using the method previously described by our laboratory.20 Briefly, immediately after obtaining the colonic biopsy, the specimen was immersed in 10 ml of Ca²⁺and Mg²⁺ free Hank's balanced salt solution (Life Technologies, Gaithersburg, Maryland, USA) containing 30 mM EDTA and incubated at room temperature for 30 minutes. After incubation the colonic crypts and surface epithelia were isolated mechanically by microscopy using fine needles. The isolated epithelium was used to extract RNA. That the isolated samples contained only crypt and surface epithelium was demonstrated morphologically using haematoxylin and eosin stained sections examined under a light microscope.

cDNA REPRESENTATION DIFFERENCE ANALYSIS (RDA)

RDA for cDNA was performed as described previously.21In brief, cDNA from normal or active UC colonic epithelium was digested with DpnII and ligated to the R-Bgl-12/24 adapters. Representations were made by polymerase chain reaction (PCR) amplification of the R ligated cDNA fragments for 20 cycles using the R-Bgl-24mer as primer. The R linked sequences in tester representations were replaced with J-Bgl-12/24 adaptors. Firstly, subtractive hybridisation was set up using 0.4 μg of J ligated testers and 40 μg of linker removed drivers (tester:driver=1:100). An aliquot of the hybridisation mixture was amplified by PCR for 10 cycles using the J-24mer as primers. The PCR products were then digested with mung bean nuclease (New England Biolabs, Beverly, Massachusetts, USA) for 35 minutes at 30°C. Digested PCR products were further amplified for 18 cycles, and the products of this amplification were the first difference products (DP1). The procedure was repeated twice using different primers and tester:driver ratios of 1:800 and 1:80 000.

IN SITU HYBRIDISATION (ISH)

Biopsy specimens were fixed with 20% neutral formalin immediately after harvesting. Formalin fixed materials were embedded in paraffin using standard procedures. Deparaffinised and rehydrated sections (4 μm) were treated with 10 μg/ml of proteinase K at 37°C for 20 minutes. Hybridisation was performed overnight using 5 μg/ml of digoxigenin (Dig)-UTP labelled sense and antisense RNA probes at 60°C. To produce Dig labelled RNA, cDNAs of Reg 1α and GW112 inserted into pCRII vector (Invitrogen, San Diego, California, USA) with both orientation was transcribed in vitro with T7 RNA polymerase in the presence of Dig-UTP using the DIG RNA labelling kit (Boehringer Mannheim, Mannheim, Germany), as explained in the protocol provided by the manufacturer. After appropriate washing with 50% formamide in 0.5× SSC at 45°C, sections were reacted with alkaline phosphatase (AP) sheep anti-Dig antibody, and the sites of AP were visualised with NBT/BCIP (Boehringer Mannheim). To verify the specificity of signals, Dig labelled sense RNA probes were hybridised with adjacent sections as negative controls in every experiment.

DNA SEQUENCING

The primer set used to amplify full length Reg 1α cDNA was 5′-AGCATGGCTCAGACCA GCTCATAC-3′ for 5′ primer and 5′-CCT CTAGTTTTTGAACTTGCAGAC-3′ for 3′ primer. The cycling conditions were 94°C for 30 seconds, 55°C for 30 seconds, and 68°C for two minutes for 35 cycles with platinum Taq DNA polymerase High Fidelity (Life Technologies). Amplified Reg 1α cDNA was cloned into pCRII vector (Invitrogen) followed by DNA sequencing using M13 reverse and forward primers on ALF automated DNA sequencer II (Pharmacia Biotech).

REAL TIME QUANTITATIVE PCR WITH FLUOROGENIC PROBES

Total RNA was extracted as previously described22from the isolated specimens of colonic epithelium. Reverse transcription of extracted RNA (500 ng) was performed using RNase H deficient reverse transcriptase (Superscript II; Life Technologies) and oligo(dT) primers (Life Technologies). An aliquot (2 μl) of diluted reverse transcription reaction mixture (200 μl) was used for quantitation of Reg 1α, GW112, Annexin-1, and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) gene expression by real time PCR assay. The primer set used to amplify Reg 1α was 5′-TCCCTGG TCTCCTACAAGTCCT-3′ for 5′ primer and 5′-GGAATCCTGTGCTTGAGGTCAG-3′ for 3′ primer. The primer set used to amplify GW112 was 5′-TGGTGTGGTGAAC ATCAGCA-3′ for 5′ primer and 5′-CCTACCCCAAGCACCATATAGA-3′ for 3′ primer. The primer set used to amplify Annexin-1 was 5′-AGTTCTGGACCTG GAGTTGAAAG-3′ for 5′ primer and 5′-TGCAAAGAAAGCTGGTTTGC-3′ for 3′ primer. The primer set used to amplify GAPDH was 5′-GAAGGTGAAGGTCGGA GTC-3′ for 5′ primer and 5′-GAAG ATGGTGATGGGATTTC-3′ for 3′ primer. The FAM conjugated fluorogenic probes used to quantify Reg 1α, GW112, and Annexin-1 gene expression were 5′-FAM-TGGAG CCCCAAGCAGTGTTAATCCTG-3′, 5′-FAM-ACCGTCTGTGGTTCAGCTCAAC TGGA-3′, and 5′-FAM-AGAAATGCCTCAC AGCTATCGTGAAGTGC-3′, respectively. The JOE conjugated fluorogenic probe used to quantify GAPDH gene expression was 5′-JOE-CAAGCTTCCCGTTCTCAGCC-3′. The fluorogenic probes were synthesised by PE Applied Biosystems (Foster City, California, USA). PCR was performed using the Taq-Man PCR kit (PE Applied Biosystems), as previously described.23 Following activation of the AmpliTaq Gold (PE Applied Biosystems) for 10 minutes at 95°C, 35 cycles of 15 seconds at 95°C and one minute at 62°C were carried out using a Sequence Detector (model 7700, PE Applied Biosystems). Real time fluorescence measurements were taken and a threshold cycle (C_T) value for each sample was calculated by the above sequence detector.24 For Reg 1α, GW112, and Annexin-1, standard curves of C_T values obtained from serially diluted pCRIIReg1α, pCRIIGW112, and pCRIIAnn-1 were constructed. C_T values for Reg 1α, GW112, and Annexin-1 transcripts from clinical specimens were plotted on the standard curves, and the amounts (fg, femtogram) of these genes were calculated automatically by Sequence Detector version 1.6 (PE Applied Biosystems). The amounts are expressed as the amount of Reg 1α, GW112, and Annexin-1 transcripts and not as the amount of plasmid. For GAPDH, a standard curve of C_T values obtained from serially diluted standard samples was constructed. The C_T values for GAPDH transcripts from clinical specimens were plotted on the standard curve and the relative amounts of GAPDH transcripts to the standard sample were calculated. Each sample was tested in duplicate, and the average of the two values was used for calculation. The difference between the two values of the samples was almost zero. After standardisation, the amount of transcripts of these genes was expressed (fg) relative to that of GAPDH. Samples were considered negative if C_T values exceeded 35 cycles. Group data were expressed as mean (SEM) and differences in transcript levels between groups were analysed by the Student's t test. A p value less than 5% denoted the presence of a significant difference between the tested groups.

IDENTIFICATION OF CANDIDATE GENES SELECTIVELY EXPRESSED IN NORMAL AND UC COLONIC MUCOSA

To identify candidate genes selectively expressed in normal and UC colonic mucosa, we performed cDNA RDA between normal and active UC colonic epithelium (fig 1). Representations of normal and active UC colonic epithelium were prepared from cDNA derived from normal and active UC colonic epithelial cells. RDA was performed with the active UC representation as “tester” and an excess amount of the normal representation as “drivers” (fig 1A) or with the normal representation as “tester” and an excess amount of the active UC representation as “drivers” (fig 1B). After one round of subtraction, the first difference products (DP1) showed a broad smear band in both cases. After two rounds of subtraction, the second difference products (DP2) showed some discrete bands in both cases. After three rounds of subtraction, the third difference products (DP3) showed two discrete bands with no polydisperse DNAs in RDA with the active UC representation as “tester” and an excess amount of the normal representation as “drivers” (fig 1A). In RDA with the normal representation as “tester” and an excess amount of the active UC representation as “drivers”, no discrete bands were seen (fig 1B). DP3 in RDA with the active UC representation as “tester” and the normal representation as “drivers” were randomly cloned. Of 14 clones examined, five were found to be identical to Reg 1α, four were found to be GW112, one was zinc finger protein 17, one was zinc finger protein 173, and one was Annexin-1 by DNA sequencing. The other two clones were not identical to any of the genes submitted to GenBank (fig1C). Of these, we further examined for Reg 1α and Annexin-1, functions of which are reported, and for GW112, which is the dominant species in DP3.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

cDNA representation difference analysis (RDA) of gene expression in normal and active ulcerative colitis (UC) colonic epithelia. (A) cDNA was derived from normal and active UC colonic epithelial cells, and used for RDA with active UC representation as the “tester” and normal representation as the “driver.” The original representations and the first (DP1), second (DP2), and third (DP3) difference products are shown. Note the presence of two discrete bands with no polydisperse DNAs in DP3. (B) cDNA was derived from normal and active UC colonic epithelial cells and used for RDA with normal representation as the “tester” and active UC representation as the “driver.” The original representations and the first (DP1), second (DP2), and third (DP3) difference products are shown. Note the lack of discrete bands in DP3. (C) Identification of active UC DP3 by DNA sequencing. Active UC DP3 was randomly cloned. Of 14 clones examined, five were found to be identical to Reg 1α, four were GW112, one was zinc finger protein 17, one was zinc finger protein 173, and one was Annexin-1 by DNA sequencing. The other two clones were not identical to any of the genes in the GenBank. Numbers in parentheses represent the number of identified clones. The GenBank accession numbers of Reg 1α, GW112, zinc finger proteins 17 and 173, and Annexin-1 are M18963, AF097021, NM-000700, AA205091, and NM-003449, respectively.

LEVELS OF REG 1α, ANNEXIN-1, AND GW112 mRNA IN COLONIC EPITHELIUM OF IBD, NON-IBD, AND CONTROL SUBJECTS

To examine if Reg 1α, Annexin-1, and GW112 are preferentially expressed in active UC colonic epithelium, expression of these genes was further examined by real time PCR. In these assays, the relative amounts of gene transcripts were quantified in colonic epithelia of control subjects (n=6), active UC (n=14), inactive UC (n=10), active Crohn's disease (n=4), inactive Crohn's disease (n=5), and non-IBD patients (ischaemic colitis n=5, amoebic colitis n=1) using GAPDH gene transcripts as an internal control for standardisation (fig 2). The mean C_T values of GAPDH gene transcripts of the colonic epithelium for each group were 26.0 (1.1) in normal, 24.6 (2.3) in active UC, 25.3 (1.9) in inactive UC, 25.4 (0.8) in active Crohn's, 24.8 (1.7) in inactive Crohn's, and 24.9 (1.2) in non-IBD. There were no significant differences in C_T values of GAPDH gene transcripts among groups. The relative amounts of Reg 1α gene transcripts [fg/GAPDH (C_T=25.6)] of colonic epithelium in each of the above groups were: not detectable, 112.6 (44.1), 9.9 (8.0), 384.0 (366.5), 0.4 (0.4), and 26.7 (15.0), respectively (fig 2A). Statistical analysis showed that the relative amount of Reg 1α gene transcripts in active UC colonic epithelium was significantly higher than in inactive UC and normal samples (active UCv inactive UC, p<0.001; active UCv normal, p<0.001). We also detected significant amounts of Reg 1α gene transcripts in active Crohn's and in non-IBD active lesions. In inactive Crohn's disease, we detected small amounts (2.0) of Reg 1α gene transcripts only in one specimen. These results confirmed upregulation of Reg 1α gene expression in the colonic epithelium of inflamed colonic mucosa, including active UC, active Crohn's, and non-IBD active lesions. The relative amounts of GW112 gene transcripts in colonic epithelium in each group were 13.8 (2.9) in active UC, 5.0 (2.3) in inactive UC, 2.5 (2.2) in active Crohn's, 8.8 (5.7) in inactive Crohn's, and 3.2 (2.4) in non-IBD. In normal tissues, GW112 gene transcripts were not detectable (fig 2B). The relative amount of GW112 gene transcripts in active UC colonic epithelium was significantly higher than in inactive UC and normal specimens (active UC v inactive UC, p<0.05; active UC v normal, p<0.001). We detected GW112 gene transcripts in two of four active Crohn's and in two of five inactive Crohn's and in three of six non-IBD active lesions. However, only in active UC was upregulation of GW112 gene expression confirmed to be statistically significant. We detected small amounts (1.0–7.0) of Annexin-1 gene transcripts in only five of 14 active UC and one of 10 inactive UC samples. Annexin-1 gene transcripts were not detectable in normal, active Crohn's disease, inactive Crohn's disease, or non-IBD (fig 2C).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Relative quantitation of (A) Reg 1α, (B) GW112, and (C) Annexin-1 gene transcripts in inflammatory bowel disease (IBD), non-IBD, and normal colonic epithelia by real time polymerase chain reaction (PCR) assay. Total RNA was extracted from colonic epithelia isolated from active ulcerative colitis (A-UC), inactive UC (I-UC), active Crohn's disease (A-CD), inactive Crohn's disease (I-CD), non-IBD, and normal controls (NC) using glyceraldehyde-3-phosphate dehydrogenase (GAPDH) gene transcripts as internal control for standardisation. Each sample was tested in duplicate and the average of the two values was used for calculation. The difference between the two values was close to zero in all samples. After standardisation, the relative amounts of gene transcripts were expressed as the amount (fg) per GAPDH (C_T=25.6). Horizontal bars represent the mean value of each group.

DETECTION OF REG 1α AND GW112 GENE EXPRESSION IN ACTIVE UC COLONIC MUCOSA BY IN SITU HYBRIDISATION

The results of real time PCR suggested that expression of Reg 1α and GW112 genes was upregulated in the epithelium of inflamed colonic mucosa in active UC. To evaluate the pathological implications of Reg 1α and GW112 gene expression in inflamed colonic mucosa, it is necessary to determine the cellular basis of expression of these genes. In the next step, we examined localisation of Reg 1α and GW112 gene expression in the inflamed colonic mucosa in UC by ISH (fig 3). Paraffin embedded sections were hybridised to Dig labelled sense and antisense RNA probes complementary to Reg 1α or GW112 template DNA. Hybridisation of sections with Dig labelled antisense RNA probes of these genes showed that Reg 1α (fig 3A) and GW112 (fig 3B) signals were confined to the crypt epithelial cells whereas epithelial cells on the luminal surface were almost free of expression of these genes. Furthermore, expression of these genes was almost absent in mesenchymal cells. Hybridisation of serial sections with Dig labelled sense RNA probes of these genes showed sparse and weak background labelling of cells. Thus ISH confirmed the specific expression of Reg 1α and GW112 in crypt epithelial cells in active UC colonic mucosa. In normal colonic mucosa, Reg 1α or GW112 gene expression was not detected by ISH (data not shown).

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

In situ hybridisation of Reg 1α and GW112 mRNAs in active UC colonic mucosa. Paraffin embedded sections were hybridised to digoxigenin (Dig) labelled sense and antisense RNA probes complementary to Reg 1α or GW112 template DNA. Hybridisation of sections with Dig labelled antisense RNA probes of these genes showed that (A) Reg 1α and (B) GW112 signals were confined to the crypt epithelial cells. Hybridisation of serial sections with Dig labelled sense RNA probes of these genes resulted in sparse and weak background labelling of cells. Original magnification ×25 (A) and ×50 (B).

COMPARISON OF SEQUENCES OF REG 1α cDNA DERIVED FROM IBD AND NON-IBD COLONIC EPITHELIA

To further evaluate the pathological implications of Reg 1α gene expression in UC, we compared the sequences of Reg 1α cDNA derived from inflamed colonic epithelium of UC with that from ischaemic colitis. The 500 bp full length cDNA sequence of each four clones derived from colonic epithelium of active UC (n=2) and ischaemic colitis (n=2) was determined. In all clones examined, the Reg 1α cDNA sequences were not different and were identical to wild-type Reg 1α (GenBank accession number M18963) in regenerating pancreatic islet cells.25

EFFECTS OF CELL CONFLUENCE AND INFLAMMATORY CYTOKINES ON REG 1α GENE EXPRESSION IN COLONIC EPITHELIAL CELL LINE (HT29)

Finally, we examined whether upregulation of Reg 1α gene expression in inflamed colonic epithelium was mediated by inflammatory mediators. Recent studies have shown that Reg 1α expression in HT29 was associated with the cell growth period.26 We first examined if the level of Reg 1α gene expression depends on cell confluence by real time PCR. Increased confluence of these cells was associated with further downregulation of Reg 1α gene expression. Although at 30% and 60% cell confluence we detected significant amounts of Reg 1α gene transcripts (130 and 41 fg Reg 1α transcripts/GAPDH, respectively), at 100% cell confluence Reg 1α gene expression was not detectable (fig 4).

• Download figure
• Open in new tab
• Download powerpoint

Figure 4

Effects of cell confluence and inflammatory cytokines on Reg 1α gene expression in a colonic epithelial cell line (HT29). Cells were harvested under 30%, 60%, and 100% confluent cell culture conditions and total RNA was extracted followed by real time polymerase chain reaction (PCR) assay to detect Reg 1α transcripts. The effects of various cytokines and nitrous oxide on Reg 1α expression in HT29 were examined under 100% confluent cell culture conditions. One day after HT29 cells became confluent, cells were cultured for 24 hours without or with interleukin (IL)-1β (2 ng/ml), IL-4 (50 U/ml), IL-6 (10 ng/ml), tumour necrosis factor α (TNF-α) (10 ng/ml), interferon γ (IFN-γ) (1 μg/ml), or 3-morpholinosydnonimine (SIN-1) (10 mM) in 2 ml of RPMI1640 containing 10% fetal calf serum. Cells were harvested and total RNA was extracted followed by real time PCR to detect Reg 1α transcripts. Each sample was tested in duplicate, and the average of the two values was used for calculation. The difference between the two values was almost zero in all samples. After standardisation, the relative amount of gene transcripts was expressed as the amount (fg) per GAPDH (C_T=15.5). Results are from one of two experiments with identical results. nd, not detectable.

It was difficult to evaluate the effects of inflammatory cytokines on Reg 1α gene expression because the level of expression was associated with cell confluence. For this reason we examined the effects of various cytokines and nitrous oxide on Reg 1α expression in HT29 under 100% confluent cell culture conditions. One day after HT29 cells become confluent, cells were cultured without or with IL-1β (0.5, 2, or 10 ng/ml), IL-4 (5, 50, or 500 U/ml), IL-6 (1,10, or 100 ng/ml), TNF-α (1,10, or 100 ng/ml), IFN-γ (0.1, 1, or 10 μg/ml), or SIN-1 (1, 10, or 100 mM) in 2 ml of RPMI1640 containing 10% fetal calf serum for 12, 24, or 48 hours. Cells were harvested and total RNA was extracted at 12, 24, and 48 hours of culture followed by real time PCR assay to detect Reg 1α transcripts. We could not detect Reg 1α gene expression under any condition or culture period studied at 100% confluency (fig 4 and data not shown).