High concentrations of insulin-like growth factor (IGF)-I and IGF-II have been demonstrated in human colonic adenocarcinomas and exert mitogenic effects through paracrine/autocrine interactions with the IGF-I receptor (IGF-IR). However, definitive studies of IGF-IR expression in these tissues have not been performed.
To study changes in the levels of the IGF-IR in colorectal carcinoma, we analyzed the expression of IGF-IR in 40 paired samples of normal and carcinomatous colonic tissue by quantitative reverse-transcription–polymerase chain reaction (RT-PCR), immunohistochemistry, and ligand binding.
As measured by RT-PCR, the IGF-IR mRNA ratio in paired tumor and adjacent normal mucosa was higher than 2.0 in 32 of 40 (80%) samples. The overall mean IGF-IR mRNA level was five-fold higher in tumor versus adjacent normal mucosa (P < 0.0001). Overexpression of IGF-IR in colon carcinomas was confirmed at the protein level by immunohistochemistry and receptor-binding studies. Colon carcinoma cells exhibited a positive staining for IGF-IR in 91% of all tumors (30 of 33) whereas the adjacent normal colonic epithelial cells showed only a very faint or no significant IGF-IR immunoreactivity. Radioligand assays and Scatchard analysis in both tissue types revealed a single class of high-affinity IGF-IR–binding sites with a similar dissociation constant (Kd; 0.14 ± 0.02 nmol/L, n = 18). However, specific 125IGF-I–binding and receptor concentrations were elevated in tumor membranes compared with normal mucosa (33.6 ± 5.6 vs. 22.7 ± 3.4 fmol/mg protein, P < 0.05). IGF-I affinity crosslinking and sodium dodecyl sulfate–polyacrylamide gel electrophoresis displayed specific bands corresponding to the size of the normal α-subunit of the IGF-IR that were more intense in carcinomatous samples. IGF-II mRNA levels were significantly elevated in colorectal carcinomas (P < 0.0001). The IGF-II mRNA ratio in tumor versus normal tissue was elevated more than twofold in 28 of 40 paired samples and a positive correlation was observed between the overexpression of IGF-II and IGF-IR in the tumors.
The insulin-like growth factor (IGF) system plays a critical role in the regulation of cell growth and transformation. IGF-I and IGF-II inhibit apoptosis, promote tumor growth, and induce transformation and metastasis in many types of malignancies.1–3 The mitogenic effects of IGF-I and IGF-II are mediated by the IGF-I receptor (IGF-IR), a transmembrane tyrosine kinase receptor structurally related to the insulin receptor, which exhibits a high affinity for both IGFs. The structurally distinct IGF-II/mannose-6-phosphate receptor, which preferentially binds IGF-II and has no tyrosine kinase activity, is believed to exert a growth suppressive function by internalization and degradation of extracellular IGF-II. IGFs bind to a variety of IGF-binding proteins (IGFBPs) present in most tissues and are involved in the regulation of IGF action.4
Colorectal tumors are one of the most common causes of cancer death in Western countries. Current therapeutic regimens for advanced stages of the disease are unsatisfactory with a low response rate and significant toxicity. The gastrointestinal system may be one of the major targets of autocrine and paracrine IGF action and there is increasing evidence that alterations in IGF signaling are involved in the neoplastic transformation and progression of colorectal carcinoma.5, 6 In addition, Ma et al.7 reported a positive correlation between high IGF-I serum levels and increased risk of colon carcinoma, suggesting a role of the IGF system in colon carcinoma pathogenesis.6
IGFs are mitogenic for intestinal epithelial cells in vitro. Human milk, which is a rich source of IGFs, enhances intestinal cell growth to a similar extent as IGF-I.8, 9 IGF-I prevents apoptosis, induces multidrug resistance, and stimulates chemotactic migration and angiogenesis in colon carcinoma cells.10, 11 While IGF-I mRNA levels are elevated only moderately in a minority of colon carcinoma patients,6 a significant overexpression of IGF-II mRNA and protein levels has been reported in 30–40% of colorectal carcinoma patients.5, 12–15 In addition, IGFBP-2, IGFBP-3, and IGF-I serum levels are elevated in patients with colorectal carcinoma and the secretion and degradation of IGFBPs are altered in human colon carcinoma cells.16, 17 Both IGF receptors are expressed in normal colonic epithelial cells and colorectal cell lines.18–20 However, there are sparse and conflicting data regarding the relative levels of IGF-IR in human colorectal carcinoma compared with normal mucosa. Some studies report no significant differences21–23 and others demonstrate overexpression.24, 25 To evaluate whether IGF-IR is overexpressed in colonic carcinomas as it is in other types of IGF-dependent malignant tissues,1, 2 we compared the expression of IGF-IR mRNA and protein in 40 paired samples of normal and carcinomatous colonic tissue.
MATERIALS AND METHODS
Materials and Sample Collection
Recombinant human IGF-I was purchased from Boehringer (Mannheim, Germany), (3-[125I]iodotyrosyl) IGF-I (human recombinant, specific activity 2000 Ci/mmol) was purchased from Amersham Buchler GmbH and CoKG (Braunschweig, Germany). Molecular biology reagents for reverse-transcription–polymerase chain reaction (RT-PCR) were obtained from Promega (San Diego, CA) and Gibco BRL (Eggenheim, Germany). Normal and carcinomatous human colon tissue samples were obtained from patients undergoing surgical treatment of colon carcinoma, according to the guidelines of the local ethics committee. Paired samples of primary human colon carcinoma and adjacent normal colon tissue were provided by the pathologist immediately after surgical removal. For the tumor samples, necrotic and ulcerative portions were removed and the presence of at least 90% of tumor cells was verified histologically. Normal colonic mucosal epithelium was dissociated from stroma, with macroscopic and histologic verification of normal mucosal appearance without contaminating nonepithelial cells. For RNA extraction and membrane preparation, tissue samples were snap frozen in liquid nitrogen and stored at −70 °C until further analysis. For histologic classifications and immunohistochemistry, tissue samples were formalin fixed and paraffin embedded.
The mean age of the patients at the time of surgery was 64 years and the female-to-male ratio was 3:5. Eight tumors were localized in the right colon, 13 tumors in the left colon, and 19 in the rectum. Histologically, all tumors were adenocarcinomas. According to the World Health Organization classification,26 the grade of differentiation was moderate in 31 tumors (Grade 2) and low in 9 tumors (Grade 3–4). Tumor stage was T1–2 in 5 patients, T3 in 29 patients, and T4 in 4 patients. At the time of surgery, lymph node infiltration was found in 16 patients and 8 patients had distant metastasis.
To measure IGF-IR levels in 40 paired normal and carcinomatous colonic tissue samples, a competitive RT-PCR method (Clontech Laboratories, Palo Alto, CA) was used, as previously described.27, 28 The MIMIC PCR technique utilizes an exogenous internal standard (MIMIC) that competes for the same primers as the target IGF-IR or IGF-II DNA. With knowledge of the amount of MIMIC DNA added in serial dilutions to the amplification reactions, the amount of the target template can be determined, and thus the initial IGF-IR or IGF-II mRNA concentration.29, 30 The competitive internal standard, which contains the identical primer binding sites used to amplify the IGF-IR or IGF-II DNA, was generated by amplifying a BamH1/EcoR1 fragment of v-erB with two composite primers. In these composite primers (40-mer), the first 20 nucleotides are complementary to IGF-II or IGF-IR and the next 20 nucleotides are complementary to v-erB. The internal standard was synthesized, purified, and quantified by spectrophotometry as described by the manufacturer (Clontech). Amplification of the competitive internal standard generated a 288-bp fragment for IGF-IR and a 242-bp fragment for IGF-II. The following primers were used to amplify the human IGF-IR and the internal standard, respectively:sense 5′ACAGAGAACCCCAAGACTGAGG3′, antisense: 5′TGATGTTGTAGGTGTCTGCGGC3′, corresponding to nucleotides 2095–2116 (exon 10) and 2341–2320 (exon 11) of the human IGF-IR cDNA sequence,31 and for human IGF-II: sense 5′CATCGTTGAGGAGTGTTT3′, antisense: 5′GGGGTATCTGGGGAAGTTGTC3′, corresponding to nucleotides 445–465 (exon 3) and 577–557 (exon 4) of the human IGF-II cDNA sequence.32 Amplification of the target DNA with these intron overlapping primers yielded one specific 247-bp fragment for IGF-IR and one 133-bp fragment for IGF-II, thereby excluding amplification of contaminating DNA. The PCR products obtained were confirmed by sequencing. In pilot experiments, the exponential phase of the amplification was determined for the target DNA and the internal standard. Subsequently, a cycle number in the middle of the linear amplification range (21–27 cycles) was chosen. In the system used, the efficiencies of amplification of target cDNA and competitive internal standard DNA were equal.
For RNA extraction, 50 mg of tissue sample was incubated with 1 mL of cell lysis buffer (Trizol, Gibco, Grand Island, NY) for 5 minutes. Total cellular RNA was then isolated using the acid-guanidinium isothiocyanate phenol chloroform extraction method as described.33 The concentration and purity of the RNA were determined by ultraviolet spectrophotometry and RNA was stored at −70°C until analyzed.
For RT of extracted RNA to cDNA, 1.0 μg of total RNA template was incubated for 60 minutes at 37 °C in a 20-μL reaction volume, containing 1 × first-strand buffer (50 mmol/L Tris/HCl,75 mmol/L KCl, 3 mmol/L MgCl2), 0.5 mmol/L of each deoxynucleotide, 1.8 μg random primer, Dithiotreitol (DTT, Promega, San Diego, CA)(10 mM), 20 U ribonuclease inhibitor RNasin (1.0 U/μL), and 240 U murine MM leukemia virus reverse transcriptase (12 U/μL). The reaction was stopped by incubating at 95 °C for 5 minutes and the samples were placed on ice or stored in −20 °C for further analysis. Subsequently, PCR reactions were performed in a thermal cycler (Gene Amp PCR System 2400, Perkin-Elmer, Weiterstadt, Germany): 2 μL RT product and internal mimic standard in serial dilution were amplified in a volume of 50 μL containing 1 × PCR buffer (10 mmol/L Tris-HCl, pH 9.0, 50 mmol/L KCl), 1.0 mmol/L MgCl2, 200 μM each deoxynucleotide, 0.6 pmol/μL each primer, and 0.06 U/μL Taq DNA polymerase. For amplification of the IGF-IR cDNA, the first denaturation step (95 °C for 6 minutes) was followed by 27 cycles with a 1-minute denaturing step at 95 °C, a 1-minute annealing step, starting at 70 °C and decreasing by 0.5 °C with each cycle to a minimum of 65 °C, and a 1.5-minute elongation step at 72 °C. For amplification of the IGF-II cDNA, the following protocol after the first denaturation step (95 °C for 5 minutes) was used: 26 cycles with a 1-minute denaturing step at 95 °C, a 50-second annealing step at 63.5 °C, and a 1.5-minute elongation step at 72 °C. As a final extension step in both protocols, the reaction was heated to 72 °C for 6 minutes and then cooled. PCR products were electrophoresed in an 8% polyacrylamide gel with a 1-Kb DNA ladder followed by silver staining. The stained gel was analyzed with a computerized scanner with image analyzing software (NIH Image, Version 1.61, National Institutes of Health, Bethesda, MD). Routinely, negative controls without input RNA or with ommitted RT steps were included.
For quantification of target mRNA levels, equal amounts of target cDNA were amplified with different dilutions of known amounts of mimic DNA. After RT-PCR, the ratios of mimic-to-target band intensity were determined and the concentration of a 1:1 mimic-to-target ratio was calculated as described.28 For each sample, an initial estimate of IGF-IR and IGF-II mRNA was performed with a single dose of internal standard DNA, followed by a narrow titration of internal standards around this estimated value, according to the method of Alms et al.30 To minimize methodologic variations in the assessment of IGF-IR and IGF-II mRNA levels, RNA extraction, RT, and amplification of paired normal and carcinomatous tissue from a single patient were performed simultaneously in the same reaction in serial dilutions and PCR products were separated and quantified in the same gel. The RNA of each sample was reverse transcribed and analyzed by RT-PCR in duplicate in two separate experiments. Results were expressed as the number of molecules per microgram total RNA or as the ratio between mRNA levels of normal and carcinomatous tissue from paired samples. Using this method, the intraassay coefficient of variation (CV) for IGF-IR and IGF-II mRNA quantification was less than 6% and the interassay CV was less than 12%.
Immunohistologic staining for human IGF-IR was performed on formalin-fixed and paraffin-embedded tissue from 33 primary colorectal carcinomas and adjacent normal mucosa. Sections (3–4 μm thick) were cut from the paraffin blocks and mounted on Super-Frost/Plus microscope slides (Menzel, Bielefeld, Germany). After deparaffinization and rehydration, the slides were heated three times for 10 minutes in a microwave oven at 750 W in an enhancer (E 8000 Camon) and allowed to cool for 1 hour. After three washings in phosphate-buffered saline (PBS), the slides were incubated with 75 μL of a polyclonal chicken anti–IGF-IR antibody from Biozol (Eching, Germany, No. 06429; dilution 1:20, incubation time 22 hours at 4 °C and 1 hour at 21 °C in a wet chamber). The reaction was visualized by a biotin-streptavidin standard procedure. After washings in PBS, the sections were incubated for 30 minutes with a biotinylated goat anti-chicken-IgG-antibody (Biozol, No. 6100 05; 21 °C in a humid chamber). After washing again in PBS, an incubation with peroxidase-conjugated streptavidin (Dako, Carpinteria, CA, No. K0377) was performed (30 minutes at 21 °C in a humid chamber) with AEC (Sigma, Deisenhofen, Germany) as chromogen. Finally, the slides were counterstained with hematoxylin and mounted in Aquatex (Merck, Darmstadt, Germany). Routine negative controls were performed by application of PBS instead of the primary antibody. An isotype control was done for the IGF-IR chicken antibody by using a chicken IgY control serum (Promega, Mannheim, Germany, No. G116A). All negative and isotype controls resulted in negative immunohistochemical reactions. The immunohistochemical results for IGF-IR were recorded as positive if the tumor cells showed an unequivocally strong cytoplasmatic and/or membranous reaction in greater than 50% of the carcinoma cells. Cases with faint staining or with a positive reaction in a minority of tumor cells were classified as negative.
IGF-I–Binding Studies and Affinity Crosslinking
125I-IGF-I–binding studies were performed with membrane preparations of 18 paired samples from normal and carcinomatous colonic tissue samples as previously described.34 Briefly, tissue samples were homogenized mechanically in homogenization buffer (0.25 M sucrose, 0.25 mg/L antipain, and 100 mg/L phenylmethyl sulfonyl fluoride [PMSF]) and centrifuged at 600 g for 10 minutes. The supernatant was centrifuged at 10,000 g for 30 minutes, adjusted to a final concentration of 0.1 mol/L NaCl and 0.2 × 10−3 mol/L Mg2SO4, centrifuged twice at 100,000 g for 90 minutes, and resuspended in membrane buffer (50 mM Tris-HCl, pH 7.4; 0.25 mg/L antipain, and 100 mg/L PMSF). Aliquots of 80 μg membrane protein were incubated for 3 hours at room temperature together with 125I-IGF-I (20,000 cpm) and increasing concentrations of unlabeled IGF-I in 400 μL binding buffer (Medium 199 containing 0.2% bovine serum albumin [BSA], 150 mM NaCl, and 1.2 mM MgSO4). Membrane-bound radioactivity was measured and receptor kinetics were calculated by Scatchard analysis35 using a standard software program (Ligand, National Institutes of Health, Bethesda, MD).
Affinity crosslinking experiments of 125I-IGF-I with membrane preparations from normal and carinomatous colonic tissue were performed as described.34 In brief, membranes (100 μg) were incubated with 125I-IGF-I (200,000 cpm) and unlabeled IGF-I (250 ng/mL) in 200 μL of binding buffer, washed, and bound ligand was cross-linked to the IGF-IR with 0.1 mol/L disuccinimidylsuberate in Dulbecco's minimum essential medium containing 0.2% BSA for 45 minutes. The samples were quenched with 10 mmol/L EDTA, dissolved in buffer containing 10 mM NaH2PO4, and subjected to sodium dodecyl sulfate–polyacrylamide gel electrophoreis (SDS-PAGE; 12% acrylamide) under reducing conditions. Gels were dried and analyzed by autoradiography.
All data are expressed as mean ± SEM. Comparative data were analyzed by multivariate analysis and Student t test for paired data with significance defined as P less than 0.05, unless otherwise mentioned.
IGF-IR mRNA Expression in Human Colon Carcinomas
The expression of IGF-IR mRNA in human colon carcinomas was compared with normal colon mucosa by quantitative RT-PCR in 40 paired tissue samples from patients with colon carcinoma. Amplification of cDNA with primers located in Exons 10 and 11 yielded one specific PCR product of 247 bp (Fig. 1). Sequence analysis showed that these products were identical with the cDNA sequence of the human IGF-IR (data not shown). No products were detected when the RT reaction or the input RNA was omitted. The same size and intensitiy of the amplification product were observed when the RNA was treated with DNAse before RT-PCR, excluding possible amplification of contaminating DNA. Expression of the human IGF-IR gene was detected in normal and carcinomatous colonic tissue of all patients. In normal mucosa, IGF-IR was expressed at 89 ± 12 × 103 molecules per microgram RNA. Significantly higher levels were observed in tumor samples, with a mean expression of 279 ± 46 × 103 molecules per microgram RNA (P < 0.0001). Figure 1 shows a representative result of the quantitative RT-PCR analysis from a paired sample of normal and carcinomatous human colonic tissue. When the IGF-IR mRNA levels of paired samples were compared (Fig. 2), a wide heterogeneity in the tumor/normal IGF-IR mRNA ratios was observed with a mean 4.9 ± 0.9-fold relative overexpression of the IGF-IR in the carcinomas (P < 0.0001). Because the accuracy of the competitive RT-PCR technique is sufficient to detect at least twofold differences in RNA levels,30 a significant overexpression of IGF-IR in the tumor was assumed when the ratio of carcinoma-to-normal mRNA levels was greater than 2.0. According to this criterion, an overexpression of IGF-IR was observed in 32 of 40 (80%) of the paired samples. Although there were tendencies for a greater degree of IGF-IR overexpression in larger tumors (ratios by tumor size stage: T1–2 = 3.0 ± 1.2, T3 = 4.6 ± 0.9, T4 = 8.4 ± 5.8) and in the presence of metastases at the time of surgery (ratios by metastatic stage: M0 = 4.7± 1.1, M1 = 5.9 ± 2.7), these observations did not reach statistical significance. IGF-IR expression ratios were unrelated to donor age, gender, lymph node infiltration, histologic grading, and localization of the tumor.
When the same tissue samples were subjected to quantitative RT-PCR for IGF-II, mean IGF-II gene expression was 57 times higher in malignant tissue than in adjacent healthy mucosa (P < 0.05). A relative overexpression of IGF-II was observed in 70% of all patients (28 of 40), with a moderate (2–35-fold) or a very strong (100–840-fold) increase in the IGF-II mRNA content of the tumor in 22 and 6 patients, respectively. Regression analysis revealed a positive correlation between the overexpression of IGF-II and IGF-IR in the tumors (r = 0.42, P < 0.05) with a 2.6, 4.1, and 9.6-fold increase in IGF-IR levels in the tumors with no, moderate, and strong IGF-II overexpression, respectively (P < 0.01). Although all carcinomas with a strong overexpression of IGF-II exhibited an advanced tumor stage, no significant correlation between IGF-II tumor levels and clinical parameters was detected.
In 30 of the 33 colon carcinomas (91%) analyzed, positive cytoplasmic and cell membrane staining for IGF-IR was detected. The three remaining samples exhibited limited focal positive immunostaining for IGF-IR. In all positive cases, greater than 50% of the carcinomatous cells were immunolabeled and a very strong reaction was observed in 18 of 33 cases (55%; Fig. 3). Staining within the immunoreactive colonic carcinomas was heterogeneous, i.e., cells or cell groups with strong IGF-IR reactivity were intermingled with weakly positive or negatively immunostained tumor cells. In contrast, adjacent normal colonic mucosa cells were essentially IGF-IR negative; faint immunostaining was observed in epithelial cells on the basolateral side of the crypts and occasionally in neuroendocrine cells at the base of the crypts.
IGF Binding to Normal and Carcinomatous Colonic Tissue
Binding kinetics of 125I-IGF-I to membranes from 18 paired normal and carcinomatous colon tissue samples was investigated. The mean specific binding of 125I-IGF-I to membranes from normal human colon mucosa was 3.1 ± 0.4 %. 125I-IGF-I binding was effectively displaced by unlabeled IGF-I with a 50% displacement (ED50) at 0.17 ± 0.03 nmol/L. In contrast, significantly higher concentrations of IGF-II were necessary for a 50% displacement and insulin was effective only at micromolar concentrations (data not shown). Scatchard analysis revealed a single class of high-affinity binding sites with a Kd of 0.14 ± 0.02 nmol/L and a receptor concentration of 22.7 ± 3.4 fmol/mg protein. In comparison to normal mucosa, membranes from the corresponding colon carcinomas showed a higher specific 125I-IGF-I binding of 4.3 ± 0.9 % as well as an elevated mean IGF-IR concentration of 33.6 ± 5.6 fmol/mg protein (P < 0.05). The mean tumor-to-mucosa IGF-IR concentration ratio in paired samples was 1.8 ± 0.3 and 6 of 18 human colon carcinomas showed a more than twofold overexpression of IGF-IR compared with the adjacent normal colon mucosa of the same donor. However, IGF-I was equally potent in displacing the labeled ligand from carcinomatous membranes (ED50 0.18 ± 0.04 nmol/L) and the Scatchard analysis showed a single class of high-affinity binding sites with normal binding kinetics in all examined carcinomas (Kd 0.14 ± 0.02 nmol/L), indicating overexpression of normal intact human IGF-IR in human colon carcinomas. A representative comparison of the Scatchard plots for a pair of normal and carcinomatous colon tissue samples is shown in Figure 4.
Affinity Crosslinking of IGF-IR
Affinity crosslinking of normal and carcinomatous human colon membranes with 125I-IGF-I followed by SDS-PAGE and autoradiography revealed bands at 135 kilodaltons (kD), corresponding to the size of the intact α-subunit of the IGF-IR (Fig. 5). In addition, larger bands with a molecular weight of 270 kD were observed, which probably represent cross-linked α-subunit dimers. The binding of 125I-IGF-I was displaced by competition with excess unlabeled IGF-I (Fig. 5). Affinity cross-linking of the corresponding carcinomatous membranes gave similar results but with greater intensity, consistent with a relative overexpression of intact IGF-IR in human colon carcinomas (Fig. 5). No specific lower molecular weight bands were detected after crosslinking membranes with 125I-IGF-I as would occur with contamination of the membrane preparations with IGFBPs.
This study was designed to address the controversial topic of whether IGF-IR is overexpressed in colon carcinoma. Zenilman and Graham21 and Mishra et al.23 found no difference in IGF-IR mRNA concentration between normal and carcinomatous human colonic tissue samples, whereas Freier et al.24 reported an increased expression of IGF-IR in colon carcinoma. However, because these studies investigated only a small number of samples (four,21 nine,24 and six23 tumors, respectively), we still do not know whether colorectal carcinomas overexpress IGF-IR.
To address this issue in a more definitive manner, we investigated IGF-IR expression in carcinomatous tissue and paired adjacent normal mucosa taken from 40 donors undergoing surgical treatment for colorectal carcinoma. By quantitative RT-PCR, IGF-IR was identified in all normal and malignant tissue samples. However, mean IGF-IR mRNA levels were elevated fivefold in primary colon carcinomas compared with normal mucosa (P <0.001) and a more than twofold overexpression of the IGF-IR was found in 80% of the tumor samples. Therefore, our results clearly support the findings of Freier et al.24 and provide strong evidence for overexpression of IGF-IR in the majority of human colorectal carcinomas.
These findings were strengthened by our studies demonstrating significantly stronger IGF-IR immunostaining in colorectal carcinoma cells compared withadjacent normal colonic epithelial cells. These results support the immunologic findings of Hakam et al.,25 who also reported stronger IGF-IR immunostaining in colon carcinomas. Normal colonic epithelial cells displayed faint IGF-IR immunostaining mostly on the basolateral side of the crypt cells, which is similar to a previous report in the differentiated human colonic cell line HT29.36
In colon carcinoma cells, IGF-IR immunoreactivity was identified not only on the membrane but also in the cytoplasm of these cells. This staining pattern is in accordance with IGF-IR immunostaining in liver metastasis from colorectal tumors.37 It is presumably due to the internalization of IGF-IR which, like the insulin receptor, undergoes internalization from the outer cell membrane to the cytoplasm after ligand binding.
The presence of elevated concentrations of IGF-IR protein in colorectal carcinoma was further confirmed by radioligand-binding assays and receptor affinity crosslinking studies. Although the IGF-IR–binding kinetics and the size of the IGF-IR α-subunit were similar for both normal and carcinomatous colonic epithelial cells, the specific binding and the IGF-IR density were significantly higher in colon carcinomas. Our finding of a single class of high-affinity IGF-I–binding sites in both normal and carcinomatous colonic membranes agrees with earlier reports of intact IGF-IRs in normal and malignant human colorectal tissue.18, 19, 22, 25 The observed Kd value of 0.14 ± 0.02 nmol/L is comparable to the Kd values previously reported for primary colorectal carcinoma and normal mucosa (Kd 0.12–0.17 nmol/L).18, 22 Therefore, it is unlikely that malignant transformation of colonic cells is associated with a significant change in the structure or ligand affinity of IGF-IR.
By competition experiments with 125I-IGF-I binding and subsequent Scatchard analysis, we found a higher IGF-IR concentration in colorectal tumors compared with normal mucosa. This is partially in contrast to the results of Adenis et al.,22 who performed 125I-IGF-I saturation experiments with membrane preparations from 20 human colorectal carcinomas and normal colonic mucosa. Although carcinoma samples exhibited a higher percentage of positve results for the presence of IGF-IR (70% vs. 46%) and a 1.3-fold higher IGF-IR concentration than normal mucosa, these differences did not reach significance. A possible explanation might be that the majority of samples were from well differentiated colon carcinomas (16 of 20), whereas we studied tumors of low to moderate differentiation.
In 125I-IGF-I crosslinking studies, we excluded the presence of contaminating IGFBPs. In addition, the apparent molecular weight of the IGF-IR α-subunit was the same size as the α-subunit found by other groups in normal human colonic mucosa and cell lines.18, 19, 38 Although the electrophoretic mobility of the α-subunit was the same in normal and tumorous colonic membrane preparations, the intensity of the bands was stronger in the carcinomatous samples, confirming a higher concentration of the intact IGF-IR in the tumors.
There was a tendency for a stronger expression of IGF-IR in larger colon tumors and in the presence of metastatic disease. However, we did not find a statistically significant correlation between IGF-IR status and tumor stage or grading. This might be due to the low number of samples in each subgroup. RT-PCR can be performed with very small amounts of tissue, including samples from colonic forceps biopsies. Therefore, future studies which assess IGF-IR status in a large number of colon adenomas and carcinomas will help to elucidate the clinical implications of IGF-IR expression in human colonic tumorigenesis.
In agreement with previous reports,12–14, 24 we found an average 57-fold overexpression of IGF-II mRNA in 70% (28 of 40) of colorectal carcinomas. Several previous studies have shown that the increase in IGF-II mRNA levels is associated with an elevated concentration of IGF-II protein in the tumor tissue13, 24 and that serum levels of IGF-II are elevated in patients with colorectal carcinoma.16 Although there was a tendency toward a more advanced tumor stage in carcinomas with a strong IGF-II overexpression, this trend did not reach statistical significance. However, IGF-II staining was correlated with tumor progression and patient survival in an immunohistologic study on tumor samples from 92 patients with colorectal carcinoma.39 We found a significant positive correlation between the overexpression of IGF-II and IGF-IR. This relationship, to our knowledge, has been described only in liver metastasis from primary colonic carcinomas.37
Although increasing evidence points to an important role of the IGF system in the pathogenesis of colorectal carcinoma,6 the mechanism and functional significance of the overexpression of IGF-II and IGF-IR in colonic tumors remain unknown. IGFs are mitogenic for gastrointestinal cells in vitro and in vivo.17 The mitogenic effect of IGF-II is dependent on the presence of the intact IGF-IR.3 The functional role of IGF-II in colonic tumor growth has been demonstrated in a transgenic mouse model.40
Overexpression of human IGF-IR promotes ligand-dependent neoplastic transformation.41 There is a quantitative relationship between tumorigenesis and IGF-IR levels,42 whereas the absence of IGF-IR prevents malignant growth and transformation in vitro and in vivo.2, 3 It is evident that high local levels of IGF-II in combination with elevated IGF-IR concentrations represent a significant growth advantage for colon carcinoma cells. This hypothesis is further supported by the fact that the liver, which is rich in IGFs but lacks IGF-IRs, is a primary organ of distant metastasis for colon carcinoma cells. It has been proposed that hepatocyte-derived IGFs and the relative expression of IGF-IR in tumor cells are important factors for the development of liver-specific metastasis.37, 43 Although the IGF system plays a key role in the tumorigenesis of colon carcinomas, it remains unclear whether the increase in IGF-IR and IGF-II peptide concentrations in colon carcinomas is merely an epiphenomenon or represents a specific step in tumorigenesis. In vitro data indicate that IGF-IR signaling is necessary for the proliferation and survival of several human colon carcinoma cell lines and that locally produced IGF-II represents an important autocrine growth factor for colon carcinoma cells.44, 45 Rearrangements of the IGF-II gene, aberarrant IGF-II promoter activity, and loss of imprinting of the IGF-II gene with biallelic expression of IGF-II due to aberrant hypermethylation have been implicated in the overexpression of IGF-II in colorectal carcinomas.46–48
Elevated IGF-II levels might contribute to the overexpression of IGF-IR in these tumors. In CaCo-2 human colon carcinoma cells, the stable overexpression of IGF-II resulted in an increased IGF-IR number with increased proliferation and anchorage-independent growth.44 This hypothesis is further supported by findings of positive correlations between the expression of IGF-II and IGF-IR in colorectal carcinomas and in liver metastases of colorectal tumors by our group and by Kawamoto et al.,39 respectively. However, the expression of the IGF-IR is regulated by a variety of factors including growth factors, oncogenes, and tumor suppressors. The exact mechanisms of the concomitant overexpression of IGF-II and the IGF-IR gene in many tumors remains unclear.49
In conclusion, our data show that both IGF-IR and IGF-II are overexpressed in the majority of human colon carcinomas. These findings support and extend previous investigations and suggest an important role of the IGF system in the pathogenesis of colon carcinoma.
The authors thank P.D.K. Lee, M.D., Ph.D., for editorial help and for a critical reading of the text.