Induction of Angiogenesis by Hyperplastic Colonic Mucosa Adjacent to Colon Cancer

Abstract

We determined whether hyperplastic mucosa adjacent to colon cancer contributes to neoplastic angiogenesis. Surgical specimens of human colon cancer (40 Dukes’ stage B and 34 Dukes’ stage C) were analyzed by immunohistochemistry for expression of proliferative and angiogenic molecules. The mucosa adjacent to Dukes’ stage C tumors (but not Dukes’ stage B tumors) had a higher Ki-67 labeling index and a higher expression of epidermal growth factor receptor and transforming growth factor-α than distant mucosa. The expression levels of vascular endothelial growth factor, basic fibroblast growth factor, interleukin-8, and the vascular density in the adjacent mucosa were similar to those in the tumor lesions and significantly higher than those in the distant mucosa. The expression of interferon-β inversely correlated with the level of pro-angiogenic molecules and the vascular density. The injection of metastatic human colon cancer cells and murine colon cancer cells into the cecal wall of mice induced hyperplastic changes in the adjacent mucosa which expressed higher levels of epidermal growth factor receptor, basic fibroblast growth factor, and vascular endothelial growth factor, and lower levels of interferon-β than did the control mucosa, which directly correlated with the degree of hyperplasia. These data suggest that metastatic human colon cancer cells can induce hyperplasia in the adjacent mucosa, which in turn produces angiogenic molecules that contribute to neoplastic angiogenesis.

The mucosa adjacent to most human colorectal adenocarcinomas is often hyperplastic. 1-3 Several analyses concluded that this transitional mucosa contains more immature and intermediate cells and fewer differentiated cells and more sialomucin secretion than does nonhyperplastic normal mucosa, whose cells predominantly secrete sulfomucins. 4-8 Whether the hyperplastic mucosa adjacent to colon cancer is a precancerous lesion 4,9,10 or a response to the growing cancer 11-14 or to microorganisms, such as Citrobacter freundii in humans 15,16 and Citrobacter rodentium in mice 17 has been debated. Hyperplasia of the ductal epithelium is often found adjacent to mammary adenocarcinoma and pancreatic cancer. Whether this atypical ductal hyperplasia is a precancerous lesion or a reactive change 18-21 is also unclear.

The growth and survival of tumor cells depends on angiogenesis, 22,23 which also increases the likelihood that tumor cells will enter the circulation to produce metastasis. 24-27 Indeed, the number of microvessels within and adjacent to tumor lesions has been shown to be a prognostic factor in human carcinomas of the breast, 28-32 prostate, 33-35 ovaries, 36 stomach, 37 and colon. 38 The onset of angiogenesis is determined by the local balance between pro-angiogenic and anti-angiogenic molecules. 39-41 Because hyperplastic tissues can express high levels of pro-angiogenic molecules, 42,43 we sought to determine whether the transitional mucosa adjacent to colon carcinomas can contribute to neoplastic angiogenesis. We examined the expression of the pro-angiogenic molecules vascular endothelial growth factor (VEGF), basic fibroblast growth factor (bFGF), and interleukin-8 (IL-8) and the anti-angiogenic regulator interferon-β (IFN-β) in 74 surgical specimens of human colon cancer and found that the hyperplastic mucosa produces high levels of pro-angiogenic molecules. We also implanted murine and human colon cancer cells into the cecal wall of nude mice and found that the growing tumor lesions induce hyperplasia in the adjacent mucosa that in turn expresses high levels of pro-angiogenic molecules directly correlating with a high degree of vascularity.

Materials and Methods

Surgical Specimens

Seventy-four formalin-fixed, paraffin-embedded archival surgical specimens of human primary colon adenocarcinomas that invaded the subserosal layer from four patients treated at The University of Texas M. D. Anderson Cancer Center and 70 patients treated at the J. R. Hiroshima General Hospital of the West Japan Railway Company were chosen at random. In 34 of 74 cases, lymph node metastasis was detected (Dukes’ stage C), and 40 cases had no lymph node metastases (Dukes’ stage B). For each case, the tumor lesion, the adjacent mucosa (within 2 mm of the tumor), and nonpathological control mucosa (at least 10 cm from the edge of the tumor) were studied.

Cultured Cells

The highly metastatic KM12SM cell line was derived from a rare liver metastasis produced by the heterogeneous, low-metastatic KM12C human colon carcinoma cell line growing in the cecal wall of nude mice. 44,45 KM12SM and KM12C human colon cancer cell lines and CT-26 murine colon carcinoma cells syngeneic to BALB/c mice 46 were grown as monolayer cultures in modified Eagle’s medium supplemented with 10% fetal bovine serum, vitamins, sodium pyruvate, l-glutamine, and nonessential amino acids. The adherent monolayer cultures were incubated at 37°C in a humidified atmosphere containing 5% CO₂ in air. All cultures were free of mycoplasma, reovirus type 3, pneumonia virus of mice, K virus, encephalitis virus, lymphocyte choriomeningitis virus, ectromelia virus, and lactate dehydrogenase virus (assayed by M. A. Bioproducts, Walkersville, MD).

Animal Models

Specific pathogen-free male BALB/c mice and male athymic NCr-nu/nu mice were purchased from the Animal Production Area of the National Cancer Institute–Frederick Cancer Research and Development Center (Frederick, MD). Animals were maintained according to institutional guidelines in facilities approved by the American Association for Accreditation of Laboratory Animal Care in accordance with current regulation and standards of the United States Department of Agriculture, Department of Health and Human Services, and National Institutes of Health. The mice were used according to the institutional guidelines when they were 8 to 10 weeks old. Modified Eagle’s medium, Ca²⁺- and Mg²⁺-free Hanks’ balanced salt solution, and fetal bovine serum were purchased from M. A. Bioproducts.

To produce cecal tumors, 1 × 10 6 KM12SM or KM12C human colon cancer cells were implanted into the cecal wall of anesthetized nude mice after laparotomy 44,45 and 5 × 10 5 CT-26 murine colon cancer cells were injected into the cecal wall of BALB/c mice. 46 The incision was closed in one layer with wound clips. Tumors were harvested 7 to 28 days after injection. Mice were injected intravenously with 0.2 ml saline containing 250 μg anti-5-bromo-2-deoxyuridine (BrdU) (Sigma Chemical Co., St. Louis, MO) 1 hour before being killed.

Histology and Immunohistochemistry

Specimens were fixed in buffered formalin and embedded in paraffin. For both human and mouse studies, consecutive 4-μm sections were cut from each study block. The sections were immunostained by anti-proliferating cell nuclear antigen (PCNA) monoclonal antibody (DAKO Corp., Carpinteria, CA), anti-BrdU monoclonal antibody (Becton-Dickinson, Mountain View, CA), anti-epidermal growth factor receptor (EGF-R) polyclonal antibody (Santa Cruz Biotechnology, Inc., Santa Cruz, CA), anti-mouse IFN-β polyclonal antibody and anti-human IFN-β polyclonal antibody (Lee Biomolecular Research Laboratories, Inc., San Diego, CA), anti-VEGF/VPF polyclonal antibody (Santa Cruz Biotechnology, Inc.), anti-bFGF monoclonal antibody (Upstate Biotechnology, Inc., Lake Placid, NY), anti-IL-8 polyclonal antibody (Biosource, Camarillo, CA), anti-CD31 monoclonal antibody (DAKO Corp.), and anti-Factor VIII polyclonal antibody (DAKO Corp.). Immunohistochemical staining was performed by the immunoperoxidase technique after antigen retrieval: microwave treatment (1000 W) in citrate buffer for 5 minutes for PCNA; 2 N HCl at 37°C for 30 minutes for BrdU; and pepsin (Biomeda Corp., Foster City, CA) at room temperature for 20 minutes for EGF-R, mouse IFN-β, human IFN-β, VEGF/VPF, IL-8, CD31, and Factor VIII. After peroxidase block by 3% H₂O₂-methanol for 10 minutes, specimens were blocked with phosphate-buffered saline (PBS) containing 5% normal horse serum and 1% normal goat serum (Vector Laboratories, Inc., Burlingame, CA). The antibodies were used at the following dilutions: 1:50 for PCNA, BrdU, and IL-8; 1:400 for EGF-R, 1:1,000 for mouse IFN-β and human IFN-β; and 1:200 for VEGF/VPF, bFGF, CD31, and Factor VIII. After an overnight incubation at 4°C with primary antibody, specimens were briefly washed with PBS and incubated at room temperature with secondary antibody conjugated with peroxidase: anti-mouse immunoglobulin G (IgG)2a goat antibody (Serotec, Inc., Raleigh, NC) for PCNA; anti-mouse IgG1 antibody (PharMingen, San Diego, CA) for BrdU; anti-mouse IgG antibody (Jackson Immuno Research, West Grove, PA) for bFGF and CD31; and anti-rabbit IgG antibody (Jackson Immuno Research) for EGF-R, mouse IFN-β, human IFN-β, VEGF/VPF, IL-8, and Factor VIII. The specimens were then washed with PBS and color-developed by stable 3,3′-diaminobenzidine solution (Research Genetics, Huntsville, AL). After quantitation by colorimetric scanning using a computer, specimens were counterstained with Meyer-hematoxylin (Sigma Chemical Co.).

Oligonucleotide Probes

Based on published reports of the cDNA sequences of EGF-R, 47,48 bFGF, 49,50 IL-8, 51,52 and VEGF, 53,54 specific antisense oligonucleotide DNA probes were designed to complement the mRNA transcripts of these four metastasis-related genes. The specificity of the oligonucleotide sequences was initially determined by a GenEMBL database search using the FastA algorithm, 55 which showed 100% homology with the target gene and minimal homology with nonspecific mammalian gene sequences. The sequences and working dilutions of the probes were as follows: EGF-R, 5′-GGA GCG CTG CCC CGG CCG TCC CGG-3′ (1:800); bFGF, 5′-CGG GAA GGC GCC GCT GCC GCC-3′ (1:200); IL-8, 5′-CTC CAC AAC CCT CTG CAC CC-3′ (1:200); and VEGF, 5′-TGG TGA TGT TGG ACT CCT CAG TGG GC-3′ (1:200). A d(T)₂₀ oligonucleotide was used to verify the integrity of the mRNA in each sample. 56 All DNA probes were synthesized with six biotin molecules (hyperbiotinylated) at the 3′ end via direct coupling using standard phosphoramidine chemistry (Research Genetics). 57,58 The lysophilized probes were reconstituted to a 1 μg/μl stock solution in 10 mmol/L Tris-HCl (pH 7.6) and 1 mmol/L ethylenediaminetetraacetic acid. The stock solution was diluted with Probe Diluent (Research Genetics) immediately before use.

In Situ Hybridization

In situ hybridization was performed using the Microprobe manual staining system (Fisher Scientific, Pittsburgh, PA). 58 Tissue sections (4 μm) of formalin-fixed, paraffin-embedded specimens were mounted on silane-coated ProbeOn slides (Fisher Scientific). 59 The slides were placed in the Microprobe slide holder, dewaxed, and dehydrated with Autodewaxer and Autoalcohol (Research Genetics), followed by enzymatic digestion with pepsin. 56 Hybridization of the probe was performed for 60 minutes at 45°C, and the samples were then washed three times with 2× standard saline citrate for 2 minutes at 45°C. The samples were incubated in alkaline phosphatase-labeled avidin for 30 minutes at 45°C, briefly rinsed in 50 mmol/L Tris buffer (pH 7.6), rinsed with alkaline phosphatase enhancer (Biomeda Corp.) for 1 minute, and incubated with chromogen substrate FastRed (Research Genetics) for 30 minutes at 45°C. A positive reaction in this assay stained red. To provide a control for endogenous alkaline phosphatase, the samples were treated as described above but in the absence of the biotinylated probe, and chromogen was used in the absence of any oligonucleotide probes. The specificity of the hybridization signal was checked using the following controls: 1) RNase to pretreat tissue section; 2) a biotin-labeled sense probe; and 3) a competition assay with unlabeled antisense probe. A markedly decreased or absent signal was obtained after all these treatments.

TUNEL Method

Apoptotic cells in intestinal tissues were detected by the terminal deoxynucleotidyl transferase (TdT)-mediated dUTP-biotin nick end-labeling (TUNEL) method as described previously. 60

Image Analysis to Quantify Intensity of Color Reaction in Immunohistochemistry and in Situ Hybridization

Stained sections were examined in a Zeiss photomicroscope (Carl Zeiss Inc., Thornwood, NY) equipped with a three-chip charge-coupled device color camera (model DXC-960 MD; Sony Corp., Tokyo, Japan). The images were analyzed using the Optimas image analysis software (version 5.2; Bothell, WA). The slides to be analyzed were prescreened by one of the investigators to determine the range in staining intensity. Images covering the range of staining intensities were captured electronically. For immunostaining, captured images were converted to gray scale, and the threshold value was set to gray scale. All subsequent images were quantified based on this threshold. The integrated optical density (OD) of each selected field was determined based on its equivalence to the mean log inverse gray value multiplied by the area of the field. Because the samples were not counterstained before image analysis, the OD was due solely to immunoreaction. A total of eight hot spots, each consisting of 20 strong-staining cells, 61 were subjected to measurement of intensities: three in the tumor, three in the control mucosa, and three in the mucosa within 2 mm adjacent to the tumor. Staining intensities in each area were measured only on cytoplasm for in situ hybridization and on cytoplasm and/or membrane for immunohistochemistry. Staining of the cells was then quantified to derive an average value of the area. The representative OD value was the mean of three hot spots for the tumor, and the mean of two mucosa at the oral and anal edges for the adjacent mucosa. The measured OD of each in situ hybridization or immunostained specimen was standardized by comparison with the integrated OD of nonpathological control mucosa of the ascending colon in mouse specimens or of nonpathological control mucosa at least 10 cm from the cancer edge in human colon cancer specimens, which were set at 100.

Labeling Index

The labeling index for staining using PCNA, BrdU, and TUNEL methods was determined by the percentage of examined nuclei that were immunoreactive. For each area (tumor, adjacent mucosa, and control mucosa), we examined three hot spots, each containing 100 nuclei and calculated the average index.

Vascular Density

Vascular density was measured on CD31- (for mouse specimens) or factor VIII- (for human specimens) stained specimens in microscopic fields (×200 magnification) at the area with maximum vascular density (hot spot) 61 of the tumor, at the tumor-mucosa junction (for the adjacent mucosa), and at the control distant mucosa.

Statistical Analysis

The mean of the assigned expression levels for EGF-R, VEGF, bFGF, IL-8, human IFN-β, labeling indexes for Ki-67, PCNA, TUNEL, and vascular density were stratified according to the metastatic status of lymph nodes and the location of the measured area. To assess the statistical significance of differences in mean values of the parameters, nonparametric analysis by unpaired Mann-Whitney U test was performed. 62 A P value of = 0.05 was considered significant. Specimen correlation analysis (Stat View, version 4.51; SAS Institute, Inc., Cary, NC) was used to define the significant relationship between the growth factors (TGF-α or IL-15) and a proliferative marker (Ki-67 labeling index) of the adjacent mucosa. 62

Results

Proliferative and Angiogenic Properties of Autochthonous Human Colon Carcinomas and Mucosa

In the first set of studies, we measured proliferative markers (Ki-67 labeling index and expression of EGF-R, TGF-α, and IL-15) and angiogenic properties (expression of bFGF, VEGF, IL-8, and vascular density) in the tumor lesions, adjacent mucosa, and distant mucosa of 74 surgical specimens of human colon carcinomas. We compared the parameters between 34 Dukes’ stage C (lymph node metastasis) and 40 Dukes’ stage B (no evidence of metastasis) tumors. Thirty-two (94%) of the 34 Dukes’ stage C cases and 16 (40%) of the 40 Dukes’ stage B cases had evidence of morphological hyperplasia in the mucosa adjacent to the carcinoma, a change defined as crypt column height ≥1.5-fold that in the control distant mucosa. 1-4 This difference in incidence of mucosal hyperplasia was highly significant (P < 0.0001, Fisher’s exact test).

As shown in Table 1 ▶ and Figure 1 ▶ , the mucosa adjacent to Dukes’ C neoplasms expressed higher levels of EGF-R, TGF-α, VEGF, bFGF, and IL-8 than did the distant mucosa. In contrast, the expression of the anti-angiogenic molecule, IFN-β, 63 known to be expressed in differentiated epithelia, 64 was lower in the tumor tissue and the adjacent mucosa than in the distant mucosa. The Ki-67-labeling index (indicating tumor cell proliferation) and the vascular density (indicating angiogenesis) were 3.8- and 10.6-fold higher, respectively, in the mucosa adjacent to the tumors than in the distant control mucosa. No discernible differences in Ki-67-labeling index, vascular density, expression levels of EGF-R, VEGF, bFGF, IL-8, or IFN-β were found between the tumors and the adjacent mucosa. The expression of TGF-α was highest in the tumor tissue, intermediate in the adjacent mucosa, and lowest in the distant mucosa (Table 1) ▶ .

Table 1.

Proliferative and Angiogenic Profiles of Human Colon Cancers and Mucosa

Factors	Tumor	Adjacent mucosa	Distant mucosa*	P value†
				(1)	(2)	(3)
Dukes’ stage C (n = 34)
Ki-67‡	38 (18–83)	19 (6–51)	5 (0–8)	<0.0001	<0.0001	<0.0001
EGF-R§	233 (122–338)	200 (138–288)	100 (98–103)	0.0049	<0.0001	<0.0001
TGF-α§	260 (114–425)	200 (114–357)	100 (98–103)	<0.0001	<0.0001	<0.0001
VEGF§	227 (136–318)	209 (145–255)	100 (98–102)	0.0034	<0.0001	<0.0001
bFGF§	181 (119–263)	163 (113–225)	100 (98–102)	0.1145	<0.0001	<0.0001
IL-8§	138 (106–169)	131 (106–156)	100 (97–102)	0.1677	<0.0001	<0.0001
Human IFN-β§	3 (0–5)	4 (2–6)	100 (97–102)	0.4357	<0.0001	<0.0001
IL-15∥	67	0	0	—	—	—
Vascular density¶	68 (48–134)	74 (42–92)	7 (2–12)	0.6785	<0.0001	<0.0001
Dukes’ stage B (n = 40)
Ki-67‡	44 (8–85)	9 (0.3–33)	5 (0–9)	<0.0001	<0.0001	0.0224
EGF-R§	179 (109–275)	125 (89–225)	100 (97–103)	<0.0001	<0.0001	<0.0001
TGF-α§	114 (88–188)	138 (100–263)	100 (98–102)	<0.0001	<0.0001	<0.0001
VEGF§	200 (120–318)	145 (64–218)	100 (100–102)	<0.0001	<0.0001	<0.0001
bFGF§	163 (113–250)	125 (88–188)	100 (97–102)	<0.0001	<0.0001	<0.0001
IL-8§	117 (100–179)	100 (94–131)	100 (98–103)	<0.0001	<0.0001	0.0212
Human IFN-β§	7 (3–12)	63 (42–93)	100 (96–101)	<0.0001	<0.0001	<0.0001
IL-15∥	8	0	8	—	—	—
Vascular density¶	46 (32–94)	37 (20–58)	8 (2–14)	0.0061	<0.0001	<0.0001

*At least 10 cm from the edge of the neoplasm and morphologically nonpathological.

^†Significance of differences between 1) the adjacent mucosa and the tumor, 2) the tumor and the distant mucosa, and 3) the adjacent mucosa and distant mucosa (unpaired Mann-Whitney U test).

^‡Median (range). Percent Ki-67-positive nuclei in five fields of 30 nuclei each.

^§Median (range). Expression intensity was quantitated by a computer program standardized to expression at the control distant mucosa. In control mucosa, three hot spots containing at least 20 cells each were examined, and the average expression was assigned the value of 100.

^∥Percent IL-15-positive cells.

^¶Median (range). Factor VIII-positive vessels were counted in five microscopic fields (×200).

Figure 1.

Hyperplastic changes in mucosa adjacent to autochthonous human colon carcinomas. Histology: H&E and immunohistochemical staining for Ki-67, TGF-α, IL-15, and Factor VIII are shown for control distant mucosa (descending colon of case 48) and the mucosa adjacent to tumor in case 48. A well-differentiated adenocarcinoma of the sigmoid colon with lymph node metastasis (Dukes stage C) and case 12 moderately differentiated adenocarcinoma of the sigmoid colon without evidence of metastasis (Dukes’ stage B). The Ki-67 labeling indexes for the control distant mucosa, the tumor, and the adjacent mucosa for case 12 were 6, 65, and 12, respectively. For case 48, the values were 12, 45, and 30, respectively. Expression intensity values of TGF-α assessed by immunohistochemistry in the control distant mucosa, the tumor, and the adjacent mucosa of case 12 were 100, 125, and 113, respectively. For case 48, the values were 100, 314, and 200, respectively. Tumor cells in case 48 (Dukes’ C) also expressed IL-15 protein. The numbers of Factor VIII-positive vessels in the control distant mucosa, the tumor, and the adjacent mucosa of case 12 were 2, 48, and 28, respectively. The values in case 48 were 10, 80, and 74, respectively.

The Dukes’ stage B tumors, however, expressed lower levels of the proliferative and angiogenic markers than did Dukes’ stage C tumors (P < 0.001). The Ki-67-labeling index and vascular density in the mucosa adjacent to the Dukes’ stage B tumors were 1.8- and 4.6-fold higher, respectively, than in the control distant mucosa. The expression of IFN-β in the mucosa adjacent to Dukes’ stage B tumors was lower than that in the control distant mucosa. The expression levels of EGF-R, VEGF, bFGF, and IL-8 were also higher in the mucosa adjacent to Dukes’ stage B tumors than in the distant mucosa (P < 0.0001, Mann-Whitney U test).

IL-15 is a mitogen for intestinal epithelial cells, especially in inflammatory bowel disease. 65,66 Immunohistochemical analysis (Figure 1) ▶ revealed that colon cancer cells express IL-15 (intratumoral heterogeneity) and that, compared to Dukes’ stage B tumors, Dukes’ stage C tumors expressed a higher level of IL-15. Nonneoplastic epithelial cells were negative for IL-15 expression.

Induction of Hyperplastic Colonic Mucosa and Expression of Angiogenic Molecules

To determine whether the hyperplastic mucosa adjacent to the tumor represents a reactive process or a precancerous lesion, we used an orthotopic murine colon cancer model. Viable nonmetastatic or metastatic colon cancer cells were implanted into the wall of the colon in mice. The highly metastatic human KM12SM cells produced growing tumors in the cecal wall of nude mice (Figure 2) ▶ . By day 7, the implantation of 1 × 10 6 KM12SM cells produced 3.7-mm diameter submucosal tumors. The expression of EGF-R in these lesions and the adjacent mucosa was 2.6-fold higher and 1.8-fold higher, respectively, than in the uninvolved mucosa (Figures 2 and 3) ▶ ▶ . The expression of TGF-α was also up-regulated in the tumors and adjacent mucosa (3.4- and 2.2-fold higher, respectively) than in the distant mucosa. Cell proliferation was determined by PCNA and BrdU labeling at the periphery of the tumors and in the crypt columns of the adjacent mucosa (within 2 mm of the tumor). On day 7 after tumor cell injection, the highest labeling index for PCNA and BrdU was found in small tumor lesions. After day 7, proliferation of tumor cells was reduced, and the high proliferative activity resumed after day 21. In the adjacent mucosa, both PCNA and BrdU labeling indexes were increased from day 7 and reached levels similar to those in the tumor lesions. Mucosal hyperplasia was also morphologically observed from day 7 after tumor cell injection (Figure 3) ▶ .

Figure 2.

Growth and expression of angiogenic markers in metastatic human colon carcinoma KM12SM cells implanted into the cecum of nude mice. Nude mice given intracecal injections of 1 × 10 6 viable highly metastatic human colon cancer KM12SM cells were killed at the indicated times. The cecal lesions and mucosa (n = 6) were analyzed by immunohistochemistry for proliferative activities (PCNA and BrdU labeling), apoptosis (TUNEL), and expression of EGF-R, TGF-α, bFGF, mouse IFN-β, and VEGF. Vascular density was determined by immunostaining with CD31. PCNA, BrdU, and TUNEL results are expressed as the percentages calculated from 100 nuclei. Immunohistochemical expression intensity was quantitated by a computer program and standardized to the intensity in control distant mucosa (value set at 100). CD31-positive vasculature was counted in five microscopic fields (original magnification, ×200). The value represents fold increase of that in the control distant mucosa. For all measurements, the SD from the mean did not exceed 10%.

Figure 3.

Histology and immunohistochemistry of colon carcinomas induced in the cecum of nude mice by implantation of highly metastatic KM12SM cells. Mice given intracecal injections of 1 × 10 6 KM12SM cells were killed on day 28. The colon and tumors were prepared for histology and immunohistochemistry. The control distant mucosa is from the descending colon. H&E staining demonstrates that the mucosa adjacent to the tumor contains elongated glands with numerous mitotic figures in PCNA. Labeling indexes in the distant mucosa, adjacent mucosa, and the tumor were 4, 39, and 45, respectively. The values for VEGF intensity were 100, 162, and 271, respectively. Vascular density was assessed by CD31 staining. Marked neovascular formation was found at the junction between the tumor and the adjacent mucosa. The vascular densities in the adjacent mucosa and the tumor were 2.3 and 2.8, respectively, higher than that in the distant mucosa.

We compared the expression levels of VEGF, bFGF, IL-8, and IFN-β in the developing tumors and the adjacent mucosa to those in the control mucosa. The expression of bFGF was increased in the small tumors but not in the large tumors. In both the tumor and the adjacent mucosa, the expression of IFN-β inversely correlated with the expression of bFGF. The expression levels of VEGF and IL-8 were higher in the tumor lesions (regardless of size) and the adjacent mucosa than in the distant mucosa.

The relative expression levels of VEGF, bFGF, IL-8, and IFN-β throughout the experiments were confirmed using an mRNA in situ hybridization technique (data not shown). The chronological changes in the expression levels of the angiogenesis-regulating genes suggested that bFGF is responsible for early stages of angiogenesis, cell division of endothelial cells, and sprouting of capillaries, whereas VEGF and IL-8 play a major role in the maintenance of the neovasculature. 64 Indeed, vascular density in the tumors reached the highest level on day 7, whereas the vasculature at the junction between the tumor and the adjacent mucosa increased to this level 1 to 2 weeks later.

To determine whether metastatic tumors induce a higher degree of mucosal hyperplasia, we compared the cecal tumors produced by highly metastatic KM12SM cells with tumors produced by low-metastatic human KM12C colon cancer cells. Compared to KM12SM cells, KM12C cells injected into the cecal wall of nude mice produced slower-growing tumors (Figures 4 and 5) ▶ ▶ . On day 28, the average diameters of cecal tumors were 3.8 mm and 12.5 mm for the KM12C and KM12SM cells, respectively. The mucosa adjacent to KM12C tumors was less hyperplastic than that adjacent to the KM12SM tumors (Figure 6) ▶ , corresponding to a lower index of PCNA and BrdU labeling (Figure 2) ▶ . The expression levels of EGF-R, TGF-α, VEGF, and bFGF for the KM12C-injected mice, although higher in the tumors than in the distant mucosa, were lower than those found for the metastatic KM12SM tumors. In contrast, the expression of IFN-β was higher in the KM12C (nonmetastatic) than in the KM12SM (metastatic) tumors.

Figure 4.

Growth and expression of angiogenic properties in nonmetastatic human colon carcinoma KM12C cells implanted into the cecum of nude mice. Nude mice given intracecal injections of 1 × 10 6 viable KM12C cells were killed at the indicated times. The cecal tumors and normal mucosa (n = 5) were analyzed by immunohistochemistry for proliferative activities (PCNA and BrdU labeling), apoptosis (TUNEL), and expression of EGF-R, TGF-α, bFGF, mouse IFN-β, and VEGF. Vascular density was determined by immunostaining with CD31. PCNA, BrdU, and TUNEL results are expressed as the percentages calculated from 100 nuclei. Immunohistochemical expression intensity was quantitated by a computer program and standardized to that in control distant mucosa (value set at 100). CD31-positive vasculature was counted in five microscopic fields (original magnification, ×200). The value represents fold increase of that in the control distant mucosa. For all measurements, the SD from the mean did not exceed 10%.

Figure 5.

Histology and immunohistochemistry of colon carcinomas induced in the cecum of nude mice by the implantation of low-metastatic human colon cancer KM12C cells. Mice given intracecal injections of 1 × 10 6 KM12C cells were killed on day 28 (n = 5). The colon and tumors were prepared for histology and immunohistochemistry. The distant mucosa is from the descending colon. H&E staining demonstrates slightly elongated glands with mitotic figures. Labeling indexes (PCNA) in the distant mucosa, adjacent mucosa, and the tumor were 4, 12, and 60, respectively. The indexes for VEGF intensity were 100, 128, and 142, respectively. Vascular density was assessed by CD31 staining. There was no evidence for increased vascular formation at the junction between the tumor and the adjacent mucosa.

Figure 6.

Growth and expression of angiogenic properties in murine CT-26 colon cancer cells implanted into the cecum of BALB/c mice. BALB/c mice given intracecal injections of 2 × 10 5 viable syngeneic CT-26 cells were killed at the indicated times. The cecal tumors and mucosa (n = 5) were analyzed by histology and immunohistochemistry for proliferative activity (PCNA, BrdU), apoptosis (TUNEL), expression of EGF-R, TGF-α, bFGF mouse IFN-β, and VEGF. Vascular density was determined by immunostaining with CD31. PCNA, BrdU, and TUNEL results are given as the percentages calculated from 100 nuclei. Immunohistochemical expression intensity was quantitated by a computer program and standardized to that in control distant mucosa (value set at 100). CD31-positive vasculature was counted in five microscopic fields (original magnification, ×200). The value represents fold increase of that in the control distant mucosa. For all measurements, the SD from the mean did not exceed 10%.

In the final set of experiments, we injected the cecal wall of BALB/c mice with syngeneic CT-26 murine colon cancer cells (Figure 4) ▶ . The results obtained with this syngeneic model were very similar to those obtained with the KM12SM human colon cancer cells in nude mice. EGF-R and TGF-α were highly expressed in cecal tumors and adjacent mucosa. Cell proliferation (as determined using PCNA and BrdU labeling) in tumors was highest on day 7 after injection. In the adjacent mucosa, the highest labeling index followed 7 to 10 days later. This proliferation corresponded to increased vascularity in the adjacent mucosa, with maximal vessel density located at the junction between the tumor and the adjacent mucosa. The expression of IFN-β inversely correlated with cell proliferation, vascular density, and expression of bFGF (Figure 6) ▶ . High expression of VEGF in the tumor was independent of the other parameters.

Discussion

We examined the proliferative index, expression of angiogenic molecules, and vascular density in 74 surgical specimens of human colon carcinomas and in tumors induced in mice by the intracecal implantation of human (nude mice) or murine (syngeneic mice) colon cancer cells with different metastatic potentials. The transitional mucosa adjacent to growing human colon cancers produced high levels of pro-angiogenic molecules and, hence, can contribute to angiogenesis of human colon carcinomas. Neoplastic angiogenesis is known to be regulated by the balance between pro-angiogenic and anti-angiogenic molecules that are released by the tumor cells 38-41,67-69 and by infiltrating host leukocytes. 70,71 The hyperplasia in the mucosa adjacent to the colon cancers was likely a reaction to the neoplasms rather than a precursor lesion. We base this conclusion on the data showing that mucosal hyperplasia was induced by the intracecal implantation of human or murine colon cancer cells. The extent of the hyperplasia and the production of angiogenic molecules directly correlated with the metastatic potential of the cells, results that agreed with the findings using surgical specimens of human colon carcinomas, ie, Dukes’ stage B versus C neoplasms.

The increased expression of the pro-angiogenic molecules bFGF, VEGF, and IL-8 and the decreased expression of the anti-angiogenic molecule IFN-β in hyperplastic mucosa correlated with an increased vascular density, ie, number and size (diameter) of blood vessels, at the junction between the tumor and the mucosa (Figures 2 and 3) ▶ ▶ . The center of the tumors contained fewer blood vessels than at their periphery, ie, the tumor-mucosa junction, raising the possibility that the increased vascular density was because of pro-angiogenic molecules released by both tumor cells and proliferating mucosal cells.

IFN-β can down-regulate expression and protein production of bFGF 63,72,73 and matrix metalloproteinases. 72-75 This cytokine is expressed in differentiated epithelial cells that line tissues in diverse organs, such as the cornea, skin, gastrointestinal tract, and genitourinary tract, and in the airways. 64 The expression of IFN-β was shown to inversely correlate with the expression of bFGF and hyperplasia of human epidermis adjacent to proliferating hemangiomas. 43 In the murine model used here, the expression of bFGF inversely correlated with the expression of IFN-β in the mucosa adjacent to the developing tumors.

The present data show that colon cancers can induce hyperplasia in normal surrounding tissues. The induction of this mucosal hyperplasia could be mediated by EGF-R and its ligands, which are produced by colon cancer tumor cells 76 that, through an autocrine-paracrine mechanism, can increase tumor cell proliferation and production of pro-angiogenic molecules. 77-79 Indeed, we found high expression levels of EGF-R and TGF-α in metastatic human colon cancers and their adjacent mucosa. Moreover, the analyses of 74 surgical specimens of human colon cancers demonstrated a significant correlation between TGF-α production and Ki-67-labeling index in the tumors and in the adjacent mucosa P = 0.551, P < 0.001, and P = 0.582, and P < 0.0001, respectively (Spearman rank correlation).

IL-15 is a known growth factor for intestinal epithelial cells. The cytokine activates signal transducer and activator of transcription (stat)3 65 and can also activate natural killer cells. 66 IL-15 is known to be produced by monocytes-macrophages. 66 In addition, we found that, like macrophages in the lamina propria, colon cancer cells in advanced Dukes’ stage C lesions produced to IL-15. In fact, the production of IL-15 by tumor cells directly correlated with the Ki-67-labeling index of the adjacent mucosa (P = 0.582, P = 0.00016 by Spearman rank correlation), suggesting that IL-15 produced by cancer cells may also induce hyperplastic changes in the adjacent mucosa.

The present results suggest that immunohistochemical examination of the mucosa adjacent to colon cancer can be used to predict the malignant potential of the neoplasms. Many markers for metastasis of human cancers are located at the invasive edge of the tumors. 80,81 Tissue samples from the invasive edge of human colon cancers are difficult to obtain by colonoscopic examination. However, the adjacent mucosa may be more amenable to such routine screening.

In summary, our results show that human colon carcinomas can induce hyperplasia in the adjacent mucosa. Although the development of this so-called “transitional mucosa” has been recognized for many years, 1-4 our data clearly show that the hyperplastic tissue expresses a high level of pro-angiogenic molecules. This hyperplasia-induced angiogenesis is a perfect example of how tumor cells can usurp host homeostatic mechanisms 82 and explains why the junction between normal tissues and the invasive edge of the tumors is highly vascularized.