ELSEVIER Molecular and Cellular End~rinolo~ lOl(1994) 141-150

Human colon carcinoma cells (CaCo-2) synthesize IGF-II and express IGF-I receptors and IGF-II/M6P receptors

Andreas Hoeflich a, Yi Yang a,‘, Ulrike Kessler a, Peter Heinz-Erian ‘, Helmut Kolb b, Wieland Kiess *7a

’ Department of Puediatric Endocrj~~o~, Cell Biolosy Laborntorp: Children’s Hospital Unive~i~ of Munch, Lindwu~tr. 4, D 80337 Munich, Germany

b Institute of Clinical Chemistry, Harlaching City Hospital, D 81545 Munich, Germany ‘Department of Paediatric and Biochemical Gastroenterology, Children’s Hospital, University of Munich, Lindwurmstr. 4, D 80337 Munich, Germany

(Received 1 November 1993; accepted 13 December 1993)

Abstract

The IGFs have been implicated in the development of the intestinal tract. We have studied the human colon carcinoma cell line CaCo-2 to gain more insight into the function of the IGFs in the gut. [1251]IGF-I and -11 bound specifically to CaCo-2 cells as measured in competitive binding experiments. The existence of IGF-I receptors was further demonstrated by affinity crosslinking studies using DSS as the crosslinking agent. Western blotting of CaCo-2 cell extracts using an anti IGF-II/M6P receptor antiserum provided additional evidence for the expression of the IGF-II/MBP receptor. In addition, Northern blotting experiments showed specific IGF-I receptor and IGF-II/M6P receptor gene expression in CaCo-2 cells. An 11 kb band was visualized with a 614 bp Psf I IGF-I receptor probe on autoradiographs. Hybridization with a 663 bp IGF-II/M6P receptor probe yielded a 9 kb RNA species. Analysis of CaCo-2 cell RNA using solution hybridization/RNase protection assays yielded two protected fragments, approximately 379 bases in length, with a 394 base IGF-I receptor riboprobe and a 250 base protected fragment with a 260 base IGF-II/M6P receptor riboprobe. In a subset of experiments a PstI 700 base fragment of the IGF-I cDNA and a 554 base Sal1 fragment of the IGF-II cDNA were used for hybridization: no hybridization was detected with the IGF-I probe. However, using the [32P]IGF-II probe bands at 6.0 and 5.0 kb were labeled in Northern blotting experiments. Analysis of CaCo-2 cell RNA using solution hybridization/RNase protection assays yielded a 289 base protected fragment and a faint 534 base species with a 556 base human IGF-II riboprobe. In addition, IGF-II immunoreactivity was measured in CaCo-2 cell-conditioned medium using an IGF-binding protein blocked radioimmunoassay. CaCo-2 cell-conditioned medium contained 5-15 ng/ml IGF-II immunoreactivi~. In conclusion, (1) CaCo-2 cells express both IGF-I receptor mRNA and IGF-II/M6P receptor mRNA and contain functional IGF-I receptor and IGF-II/M6P receptor protein. (2) CaCo-2 cells express IGF-II mRNA and secrete IGF-II immunoreactivity. We hypothesize that in human colon carcinoma cells IGF-II could act as an autocrine growth factor or alternatively could serve as a regulatory factor during differentiation.

K?y words: Colon carcinoma cell; Insulin-growth factor receptor

1. Introduction

The insulin-like growth factors are polypeptides with extensive homology with insulin in regard to their amino acid sequence, structure and biological activity

* Corresponding author. ’ On leave from the Children’s Hospital, Shanghai Medical Univer- sity, Feng Lin Rd., Shanghai, People’s Republic of China.

(Rechler and Nissley, 1990; Humbel, 1990; Rotwein, 1991). IGF-I mediates the anabolic action of pituitary growth hormone. The biological function of IGF-II is less clear: apart from its possible role as a growth factor during fetal life and embryonal development, IGF-II might be important as a differentiation factor for skeletal muscle cells and other cell types. In addi- tion, the IGFs might play an important role during carcinogenesis (Daughaday, 1990; Rechler and Nissley, 1990).

0303-7207/94/$07.# 0 1994 Elsevier Science Ireland Ltd. All rights reserved

SSDI 0303-7207(93)E0334-Q

142 A. Hoeflich et al. /Molecular and Cellular Endocrinology 101 (1994) 141-150

In the circulation and in extracellular fluids the IGFs are complexed to IGF binding proteins (IGFBPs). Until now six classes of IGFBPs have been described and characterized in regard to their amino acid se- quence and molecular structure. The IGFBPs are thought to act as IGF reservoirs ahd as modulators of IGF action (Humbel, 1990; Rechler and Nissley, 1990).

The IGFs bind to two classes of receptors: the IGF-I receptor is homologous to the insulin receptor. It binds IGF-I with high affinity and also recognizes IGF-II. Insulin binds to the IGF-I receptor with low affinity. The IGF-I receptor is composed of two alpha- binding subunits and two P-subunits which contain tyrosine kinase activity and the membrane spanning domains. The IGF-I receptor is thought to mediate most if not all of the biological functions of the IGFs (Nissley and Lopaczynski, 1991; Kiess et al., 1993). The IGF-II receptor is identical to the cation-independent mannose-6-phosphate receptor. This receptor func- tions to target lysosomal enzymes to lysosomes. A signaling role for the IGF-II/M6P receptor has also been postulated. The IGF-II/M6P receptor does not contain tyrosine kinase activity but is linked to the activation of a class of small G proteins (Gialpha) which could be involved in receptor signaling or intra- cellular trafficking of lysosomal enzymes (Kornfeld, 1992; Kiess et al., 1993).

Recently, it has become apparent that the IGF-sys- tern is involved in the development of gastrointestinal cells (Read et al., 1990; Termanini et al., 1990; Labur- the et al., 1988; Rouyer-Fessard et al., 1990). Enhanced levels of IGF-I and -11 mRNA were shown in human colon carcinomas in contrast to adult colon mucosa (Tricoli et al., 1986). In apical membranes of proximal colon epithelial cells receptors for IGF-I but not for IGF-II were detected (Pillion et al., 1989). A differen- tial distribution of IGF-I and IGF-II/M6P receptors throughout the rat gastrointestinal tract was demon- strated more recently (Heinz-Erian et al., 1991). IGF-I seems to be a major growth factor for the intestine in vivo (Read et al., 1990) and in vitro (Vanderhoof et al., 1992). A potential autocrine loop involving IGF-II in HT-29-D4 human colon-carcinoma cells has been sug- gested (Remacle-Bonnet et al., 1992). Another immor- talized cell line which is derived from intestinal epithe- lial cells, IEC-6, has also been shown to synthesize and secrete IGF-II (Park et al., 1992). We have now studied the colon carcinoma cell line CaCo-2 in order to gain more insight into the expression and function of the IGFs and their respective receptors in human gastroin- testinal epithelial cells. We report for the first time that CaCo-2 cells express high levels of IGF-II and possess IGF-I receptors and IGF-II/M6P receptors. We postulate that IGF-II could play an important role in the autonomous growth or spontaneous differentia- tion of colon carcinoma (CaCo-2) cells.

2. Materials and methods

2.1. Peptides, antibodies and cDNA probes

Human recombinant IGF-I and IGF-II were kindly donated by Dr. Klaus Mtiller, Ciba-Geigy, Basel, Switzerland and Dr. Anna Skottner, Kabi-Pharmacia, Stockholm, Sweden. Human insulin was obtained from Eli Lilly, Indianapolis, USA. The monoclonal anti hu- man IGF-I receptor antibody aIR3 was purchased from Oncogene Science, Vienna, Austria. A polyclonal antibody (No. 66416) which recognizes the human and bovine IGF-II/M6P receptor was a kind gift from Dr. W. Sly, St. Louis, Missouri. The cDNA probes used for Northern blotting were as follows: a hIGF-I, 700 bp PstI fragment kindly donated by Dr. J. Schwander, Basel, a 5.54 bp hIGF-II Sal1 fragment given by Dr. R. Voutilainen, Helsinki, a 614 bp PstI fragment of the hIGF-I receptor obtained from Dr. A. Ullrich, Munich, and a 663 bp hIGF-II/M6P receptor PstI fragment kindly also donated by Dr. W. Sly. The following probes were used in solution hybridisation/RNase protection assays and were a kind gift from Dr. C.T. Roberts, Bethesda: a 5’ 556 base (283 base coding region) hIGF- II probe, a 442 base hIGF-I probe, and a 379 base (394 base full-length probe) hIGF-I receptor probe. The hIGF-II/M6P receptor probe was 260 nucleotides in length with 250 of the bases complementary to the IGF-II/M6P receptor mRNA.

2.2. Cell culture

CaCo-2 cells were obtained from A’ITC, Rockville, USA, and were grown in Dulbecco’s modified Eagle medium (DMEM) (Seromed, Berlin, Germany) con- taining 10% FCS until confluency at 37°C in 5% CO,. For long term cultures, cells were placed serum-free when confluency was reached and fresh medium was added every three days.

2.3. Binding studies

For the binding studies, lo5 cells were seeded serum-free into 24-well plates. After 48 h cells were incubated with 2 x lo4 cpm of ‘251-labeled IGF-I or -11 (specific activity 2000 Ci/mmol, Amersham, Braun- schweig, Germany) and unlabeled ligands (human re- combinant IGF-I, human recombinant IGF-II, insulin, anti IGF-I receptor antibody alphaIR-3, or non-im- mune mouse IgG as indicated) for 2.5 h at room temperature in buffered medium. Cells were washed twice and lysed in 0.2 N NaOH. Radioactivity was measured in a gamma-counter (Berthold, Munich, Germany). Binding isotherms were analyzed after the method of Scatchard (1949).

A. Hoeflich et al. /Molecular and Cellular Endocrinology 101 (1994) 141-150 143

2.4. Crosslinking studies

lo6 cells were seeded out serum-free and after 48 h incubated with unlabeled ligands and ‘*‘I-labeled IGF-I or -11 (2 x lo5 cpm) for 2.5 h at room temperature as indicated. Disuccinimidylsuberate (DSS) (2 x lop4 M) was used as the crosslinking reagent. The monolayers were lysed in SDS and analysed by SDS-PAGE (6% acrylamide bis, 50 V for 6 h) (Laemmli, 1970) under reducing conditions (100 mM dithiothreitol (DTT)) as described before (Kiess et al., 1987).

alleled by a corresponding blot of 5 pg total RNA of the same samples which was methylene blue stained to monitor the amount and quality of the RNA applied. The probes were labeled using a random prime label- ing kit (Promega, Serva, Heidelberg, Germany) (Bandy et al., 1990). Prehybridization was carried out in 5 X

Denhardts, 6 x SSPE, 0.5% SDS, 100 pg/ml SSS (single-stranded sonified salmon sperm DNA). Hy- bridization was performed in 6 X SSPE, 0.5% SDS, 100 pg/ml SDS.

2.7. Solution hybridization /RNase protection assay 2.5. Western blotting

Triton X-100 lysates of CaCo-2 monolayers were separated by SDS-PAGE. After electroelution to cellu- lose membrane (Schleicher and Schuell, Dassel, Ger- many) the IGF-II/M6P receptor was visualized using an immunoperoxidase technique and the polyclonal antibody No. 66416 which recognizes the human and bovine IGF-II/M6P receptor (Funk et al., 1992).

Solution hybridization was performed with antisense RNA probes at 45°C overnight in a buffer containing 75% formamide, Tris 20 mM, pH 7.6, EDTA 1 mM, pH 8.0, NaCl 0.4 M, 0.1% SDS. After hybridization samples were digested with RNase (RNase A, and RNase Tl, Boehringer Mannheim, Mannheim, Ger- many), and the protected fragments were elec- trophoresed on an 8% polyacrylamide-8 M urea gel (Hernandez et al., 1992).

2.6. Northern blotting 2.8. Radioimmunoassay

RNA was extracted from cell monolayers by lysis in 4 M guanidiniumisocyanate and pelleted in 5.7 M CsCl by ultracentrifugation (33000 rpm, SW-411 for 20 h. The quality and quantity of the extracted RNA was estimated by measuring the optical density at 260 nm and 280 nm, and ethidium bromide staining of TBE gels. Total RNA (15 pg aliquots) was separated on a 0.8% phosphate-agarose gel after denaturation in DMSO/glyoxal. After capillary transfer and fixing onto Nytran membrane (Schleicher and Schuell, Dassel, Germany) the RNA was hybridized (16 h, 68°C) with 32P-labeled DNA-probes. Each Northern blot was par-

q IGF-I

l IFGIl

insulin

alpha R-3

0 n.i. IgG

An IGF-BP blocked IGF-II radioimmunoassay was used to measure IGF-II immunoreactivity as described before (Blum et al., 1988; Daughaday et al., 1980; Yang et al., 1993). 250 ng/ml IGF-I was added to each sample. After incubation with anti-hIGF-II antiserum (kindly provided by Dr. Werner Blum, Tubingen, Ger- many) and [ ‘251]IGF-II, immunoprecipitation with goat anti-rabbit IgG (IBL, Hamburg, Germany) was carried out. The radioactivity in the precipitate was collected by centrifugation and counted in a gamma-counter (Berthold, Munich, Germany).

10 loo --. 1ooa 5 25

IGFsdgGs @g/ml) insulin (ughnl)

am bound (np) ‘1 OE-3

Fig. 1. (A) Competitive binding of [1251]IGF-I to human colon carcinoma cell monolayer cultures. Binding of [t25111GF-I was measured in the absence or presence of unlabeled ligands or IgGs as described in Materials and methods. The data are means of duplicates of one representative of three independent experiments. Half-maximal displacement of radioligand binding occurred at a concentration below 1 ng/ml. Half-maximal displacement of the radioligand occurred at lo-fold higher concentrations of IGF-II compared to IGF-I. (B) Scatchard plot analysis of IGF-I binding (see panel A). The curvilinear graph suggests the presence of two different binding sites for IGF-I. The left part of the panel indicates a K, = 5 x 1O-‘3 M the right part’s K, is 1OW” M.

A. Hoeflich et al. /Molecular and Cellular Endocrinology IO1 (1994) 141-150

o- 00 1 10 100 1000 5 25

IGFsllgGs @s/ml) insulin @g/ml)

+ WFII

e IGF.1

insulin

alpha R-3

0 n.i. IgG

bound (ng) ‘lOE-3

* IFG-II

Fig. 2. (A) Competitive binding of [‘251]IGF-II to human colon carcinoma cell monolayer cultures. Binding of [‘251]IGF-II was measured in the presence or absence of unlabeled ligands as described in Materials and methods. The data are means of duplicates of one representative out of four independent experiments. Half-m~mal displacement of radioligand binding by unlabeled IGF-I occurred at a ~ncentration below 1 ng/ml. Half-maximal displacement of the radioligand by unlabeled IGF-II occurred at 1%fold higher concentrations compared to IGF-I. The upper part of the panel shows characteristics of the IGF-I receptor (displacement by arIR3, insulin and higher affinity for IGF-I than for IGF-II). The lower part (higher affinity of IGF-II) indicates binding of [‘2sI]IGF-II either to IGF-II/M6P receptors or binding proteins. (B) Scatchard plot analysis of IGF-II binding (see panel A). The graph is linear and indicates a K, = 2 x lo-l2 M.

3. Results

3.1. Identification and characterization of IGF receptors on CaCo-2 cells

Competitive binding experiments using confluent CaCo-2 monolayer cultures were carried out with [12JI]IGF-I (Fig. 1) and [‘251]IGF-II (Fig. 2). Binding of [“251]IGF-I was inhibited by unlabeled IGF-I (half-max- imal displacement less than 1 ng/ml), less so by IGF-II (half-maximal displacement, 6 rig/ml) and by insulin only at supraphysiological concentrations (S-25

rc - 115

00 - 80

49.5 - - 4q.5

50 IGF-I Mew IfIs Ab nlAb

~g/ml). A monoclon~ antibody against the human IGF-I receptor (aIR3) blocked the [‘2sIlIGF-I binding to C&o-2 cells at a concentration of 1 lug/ml almost completely, while non-immune mouse IgG (1 pg/ml) had no effect (Fig. 1A). In four independent experi- ments [‘251]IGF-II binding to CaCo-2 monolayers was competed for by nanomolar concentrations of IGF-II (half-maximal displacement, 15 ng/ml> and surpris- ingly even at lower concentrations by IGF-I (half-maxi- mal displacement, less than 1 ng/ml). The anti-IGF-I receptor antibody aIR3 and insulin only partially com- peted for [‘25111GF-II binding, whereas control IgG did

Fig. 3. (A) Affinity crosslinking of [‘251]IGF-I to human colon carcinoma cell monolayer cultures. Crosslinking was carried out as described in Materials and methods (Bo: no competitive ligand added, IGF-I: 250 ng/ml, IGF-II: 500 ng/ml, ins: 25 Kg/ml insulin, Ab: 1 pg/ml aIR3, ni Ab: 1 &g/ml non-immune mouse antibody). The upper arrow points to a 240 kDa radioligand receptor complex, which is inhibited by IGF-I/-II, insulin and by aIR3 presumably representing uncompletely reduced a-subunits of the IGF-I receptor. The lower arrow points to a complex of 130 kDa which represents the cu-subunits of the IGF-I receptor. (B) Affinity crosslinking of [‘251]IGF-II to human colon carcinoma cell monolayer cultures. Crosslinking using [‘“IIIGF-II and DSS was carried out as described in Materials and methods (abbreviations and concentrations see panel A). The upper arrow points to a 240 kDa radioligand receptor complex, which is inhibited by IGF-I/-II, insulin and by ruIR3. This band shows the characteristics of uncompletely reduced a-subunits of the IGF-I receptor. The lower arrow points to a complex of 130 kDa. Binding of labeled IGF-II can be prevented by IGF-I, -11, insulin and to some extent by the antibody against the IGF-I receptor &IRS).

A. Hoejlich et al. /Molecular and Cellular Endocrinology 101 (1994) 141-150 145

Fig. 4. Western blot analysis of IGF-II/M6P receptors in CaCo-2 lysates. Immunoblotting was carried out as described in Materials and methods using an immunoperoxidase technique. As positive controls two bovine tissues (lung and heart, midgestation bovine embryo) were analyzed in parallel. In lane ‘stand’ molecular weight markers are shown. The arrow heads point to a protein band at approximately 215 kDa representing imnmnostained IGF-11/6P re- ceptors.

Id, I-*

not influence the binding of IGF-II radioligand at all (Fig. 2A). Scatchard analysis of the binding isotherms of IGF-I revealed curvilinear graphs suggesting two different binding sites for IGF-I. One binding site was characterized by a K, of 5 X lo-l3 M, the other by a K, of 1 X 10-r’ M (Fig. 1B). Surprisingly, Scatchard analysis of the binding isotherms of IGF-II revealed a linear graph suggesting a single class of binding sites for IGF-II with a K, of 2 X lo-l2 M (Fig. 2B). How- ever, the fact that insulin and the anti-IGF-I receptor antibody (~1R3 partially inhibit binding of [1251]IGF-II to CaCo-2 cells points to the possibility that under the circumstances of these experiments IGF-II mainly binds to the IGF-I receptor and/or possibly to an insulin receptor type. In addition, [‘251]IGF-II binding that is not inhibitable by insulin and the anti IGF-I receptor antibody but can be inhibited by very low concentra- tions of IGF-I could represent binding of the radioli- gand to cell-associated IGF binding proteins. To fur- ther characterize the IGF-I and IGF-II/M6P receptors on CaCo-2 colon carcinoma cells, [1251]IGF-I and [1251]IGF-II were affinity crosslinked to CaCo-2 mono- layer cultures, using DSS as the crosslinking agent. Affinity crosslinking experiments were performed by incubating CaCo-2 cells with either [‘251]IGF ligand alone or with high concentrations of unlabeled IGF-I, IGF-II, insulin, (~1R3 or control IgG as indicated.

Fig. 5. (A) CaCo-2 cells express IGF-I receptor RNA as shown by Northern blot hybridization. Total RNA (15 pg) was hybridized with a IGF-I receptor cDNA probe as described in Materials and methods. Rat liver total RNA was used as the negative control. Analysis of total RNA from MCF-7 human mammary carcinoma cells, used as the positive control, showed bands at 11 and 6-7 kb and a signal in low molecular RNA (arrowheads). RNA samples from CaCo-2 cells kept 0, 1 and 2 days serum free showed a pattern of expression of a 11 kb RNA species. (B) CaCo-2 cells express IGF-II/M6P receptor RNA as shown by Northern blot hybridization. Total RNA (15 pg1 was hybridized with a IGF-II/M6P receptor cDNA probe as described in Materials and methods. Total RNA from MCF-7 cells and U2-OS cells, used as the positive controls, showed a band at 9 kb. In CaCo-2 cells strong expression of the 9 kb RNA species was detected.

146 A. Hoeflich et al. /Molecular and Cellular Endocrinology 101 (1994) 141-150

Affinity labeling experiments using [ ‘251]IGF-I (Fig. 3A) and [1251]IGF-II (Fig. 3B) as radioligands and analysis of the radioligand-receptor complexes by SDS-PAGE under reducing conditions showed two la- beled bands, one at 130 kDa and one at 240 kDa. Surprisingly, both [ ‘251]IGF-I and [ ‘2SI]IGF-II bound predominantly to the 130 kDa species and were dis- placed by unlabeled IGF-I, IGF-II, insulin and ruIR3, while control IgG showed no effect (Fig. 3A and 3B). The 130 kDa species has the characteristics of the a-subunit of the IGF-I receptor, whereas the 240 kDa species shows the properties of crosslinked alpha dimers of the IGF-I receptor. Preliminary evidence that [‘251]IGF-II could bind to cell-associated IGF binding proteins on CaCo-2 cells is shown in Fig. 3B: a faint radiolabeled species of approximately 43 kDa is detected at the bottom of the autoradiograph. High concentrations of unlabeled IGF-II completely inhib- ited the labeling of this band.

Since the identification of the IGF-II/M6P receptor in the affinity crosslinking study is more of an indirect nature, the IGF-II/M6P receptor in CaCo-2 cells was also examined by Western blotting. After SDS-PAGE

and electroelution of proteins onto nitrocellulose pa- per a specific rabbit antiserum against the human IGF-II/M6P receptor (No. 66416) was used to visual- ize the 220 kDa IGF-II/M6P receptor band in protein extracts from homogenized CaCo-2 cells and for com- parison in extracts from bovine fetal lung and heart tissue (Fig. 4). No comparable band was detected when non-i~une rabbit serum was used instead of the anti-receptor antiserum (data not shown).

3.2. Identification of IGF-I receptor and IGF-II/h&P receptor RNA in CaCo-2 cells

Total RNA was extracted from human CaCo-2, hu- man MCF-7 breast carcinoma, human U2-OS osteosar- coma cells and adult rat liver as described in Material and methods. Northern blotting was then carried out as described in Material and methods. With the Pst I 614 bp human IGF-I receptor probe a 11 kb RNA species was detected in CaCo-2 and MCF-7 cell RNA samples. No detectable 11 kb band was identified on the autoradiogram with adult rat liver RNA (Fig. SAL With the Pst I 663 bp human IGF-II/M6P receptor

Fig. 6. (A) IGF-I receptor RNA is expressed by CaCo-2 cells as shown by solution h~ridization/RNase protection. The assay was performed with 5 pg total RNA from different cell lines and an antisense IGF-I receptor r&probe as described in Materials and methods. Cl: riboprobe with RNases and without total RNA, C2: riboprobe without RNases and without total RNA, rat liver: total RNA which was used as the negative control, MCF-7: total RNA from MCF-7 cells which were used as the positive control, U: total RNA from U2-OS cells, Ca: total RNA from undifferentiated CaCo-2 cells, C14: total RNA from CaCo-2 cells that had been maintained serum-free for 14 days. The existence of two signals at about 379 bp is possibly due to alternative splicing of the IGF-I receptor. (B) IGF-II/M6P receptor RNA is expressed by CaCo-2 cells as shown by solution hybridizationfRNase protection. The assay was performed with total RNA (5 pg per lane) from different cell lines and antisense IGF-II/M6P receptor riboprobe as described in Materials and methods. Abbreviations are explained in panel A. CaCo-2 cehs showed a signal at 250 bp. The signal at 260 bp is possibly due to uncomplete DNase digestion. A faint signal at about 534 bp is possibly due to 5’ untranslated sequences.

A. Hoeflich et al. /Molecular and Cellular Endocrinology 101 (1994) 141-150 147

probe a major 9 kb RNA species and minor bands at 7.6, 5.2 and 1.4 kb were labeled with both CaCo-2, U2-OS and MCF 7 cell RNA samples (Fig. 5B). When solution hybrid~ation assays were performed with a 394 base hIGF-I receptor, and a 260 base hIGF- II/M6P receptor antisense RNA probe protected frag- ments of 379 base and 250 base length, respectively, were detected with CaCo-2 RNA (Fig. 6A and 6B). Protected fragments of 379 base and of 250 and/or 260 bases, respectively, were also seen when RNA from MCF-7, and U2-OS cells were analyzed as positive controls (Fig, 6A and 6B). The existence of two signals at about 379 base is thought to be due to alternative splicing of IGF-I receptor RNA. The presence of a 260 base protected fragment after hybridization with the IGF-II/M6P receptor riboprobe in MCF-7 cell and rat liver RNA is also possibly explained by alternative splicing.

3.3. Identification of IGF-II RNA in CaCo-2 cells

Total RNA was extracted from CaCo-2 cells, MCF-7 cells and adult rat liver as mentioned before and Northern blotting was carried out as described in Ma-

terials and methods. With the PstI 700 bp IGF-I probe no signal was observed when blots of CaCo-2 cell RNA were analysed by autoradiography (data not shown). In contrast, when the blots were hybridized with the radi- olabeled Sal1 554 bp human IGF-II probe, a 6.0 kb and a 5 kb radiolabeled species were seen on the radiograms (Fig. 7A). This band was present both in lanes where CaCo-2 cell RNA had been analyzed and in the lane where MCF 7 RNA had been run on the agarose gel. MCF 7 cells are known to express IGF-II RNA and synthesize IGF-II peptides (Mathieu et al., 1990, Yang et al., 1993). No IGF-II specific signal was detected on the blots where RNA from adult rat liver or C6 glial cells had been added. It is well known that IGF-II RNA is not at all or to a very low degree expressed in adult rat liver and C6 glial cells. In addi- tion, solution hybridization experiments with antisense RNA probes (a human IGF-II 5’ probe of 556 bases (283 base coding region), and a hIGF-I 442 base probe) revealed that no or extremely little IGF-I but rather high levels of IGF-II RNAs are expressed by CaCo-2 cells: a 289 base fragment was protected by the IGF-II probe. A similar, albeit only faintly stained, fragment was protected when control RNA from MCF 7 cells was analyzed in parallel (Fig. 7B).

Fig. 7. (A) CaCo-2 cells express IGF-II RNA as shown by Northern blot hybridization. Total RNA was hybridized with IGF-II cDNA probe as described in Materials and methods. Total RNAs from rat liver and C6 rat glial cells were used as the negative controls. Total RNA from MCF-7 cells, used as the positive control, showed bands at 5 kb and at 6 kb. In CaCo2 cells a major 6 kb RNA species was visualized. (B) IGF-II RNA is expressed by CaCo-2 cells as shown by solution hybridiiation/RNase protection. The assay was performed with 5 fig total RNA from different cell lines and antisense IGF-II riboprobe as described in Materials and methods. Abbreviations are explained in panel A. MCF-7 cell RNA was used as the positive control (weak signal at 289 bp). CaCo-2 cells show a strong signal at 289 bp.

148 A. Hoeflich et al. / Molecular and Cellular Endocrinology I01 (1994) 141-150

” control serum standard

-+- Caco-2 l total binding

OJ (01) medium 6 30 60 125 250 500

I I I I I I I (nq) IGF-II 0.06 0.13 0.25 0.50 1.00 2.00 4.00

Fig. 8. IGF-II immunoreactivity was measured using an IGFBP blocked ICE11 RIA in cell conditioned medium from CaCo-2 cells as described in Materials and methods. The plotted data points (mean of duplicate determinations) show the result of one represen- tative experiment from three independent experiments (control serum: diluted, acid ethanol extracted serum from an adult volun- teer, standard: recombinant IGF-II standard, CaCo-2: cell-condi- tioned medium from CaCo-2 cells).

3.4. measurement of IGF-II ~mrnu~ore~ctiui~ in CaCo-2 cell-conditioned medium

Cell-conditioned medium was collected from conflu- ent CaCo-2 cell monolayer cultures as described in Materials and methods. In brief, cells were cultured until confluency, changed serum-free and the medium was discarded after 24 h. Cell-conditioned medium was collected after additional 24, 48 and 72 h, concen- trated, acid-ethanol extracted and analyzed by an IGFBP-blocked radioimmunoassay employing [ 12’ II- IGF-II. Cell-conditioned medium from CaCo-2 cell cultures contained approximately 5-15 ng/ml IGF-II immunoreactivity after 24 h of culture. Dilution curves of cell-conditioned media were linear and paralleled those of standards of recombinant human IGF-II (Fig. 8).

4. Discussion

We have studied the human colon carcinoma cell line CaCo-2 in order to (1) gain more insight into the expression of the IGF system in gastrointestinal cells and to (2) establish an in vitro tissue culture model of IGF action in human colon cells. To assess whether CaCo-2 cells were capable of responding to IGFs we first set out to identify IGF receptors on these cells. The demonstration of the existence of characteristic IGF-I receptors and IGF-II/M6P receptors in CaCo-2 cells is based upon several experimental findings: first, competitive binding studies using [ 1251]IGF-I and [1251]IGF-II showed the typical binding profiles for the

IGF-I receptor: binding of [iZS]IGF-I to its receptor was inhibited by nanomolar concentrations of unla- beled IGF-I and a specific IGF-I receptor antibody (aIR-3). IGF-II inhibited [‘25111GF-I binding at slightly higher concentrations, whereas insulin did so only at about lOOO-fold higher concentrations (Fig. 1A). The binding sites for [‘25]IGF-II preferred IGF-I over IGF- II and partially recognized insulin or aIR-3 (Fig. 2A). Analysis of the binding isotherms for IGF-I suggest that IGF-I binds to two classes of binding sites, one with a I(, of 5 x lo-l3 M, the other with a K, of 10-i’ M (Fig. 1B). The binding site with the lower K, for IGF-I possibly represents an insulin type receptor or an alternatively spliced IGF-I receptor subtype, the existence of which is suggested by the results of solu- tion hybridization studies (Fig. 6A), whereas the high affinity class of binding sites is characteristic of the IGF-I receptor. Analysis of the binding isotherms for IGF-II suggests that IGF-II binds to a single class of binding sites with a K, of 2 X lo- l2 M (Fig. 2B). This binding site might represent an IGF-I or insulin recep- tor type. Second, affinity labeling experiments using [‘251]IGF-I (Fig. 3A) and [‘251]IGF-II (Fig. 3B) and analysis of crosslinked radioIigand-receptor complexes by SDS-PAGE and autoradiography showed two dif- ferent receptor bands: a 130 kDa band with typical binding properties of the IGF-I receptor cu-subunit and a 240 kDa band which presumably represents crosslinked alpha dimers of the IGF-I receptor, the insulin receptor or insulin receptor/IGF-I receptor hybrids. [‘251]IGF-II under the conditions of the crosslinking experiments clearly binds mainly to the IGF-I receptor in CaCo-2 cells (Fig. 3B). Third, the IGF-II/M6P receptor was identified by Western blot analysis of CaCo-2 cell extracts using an anti-human IGF-II/M6P receptor antiserum (No. 66416, Fig. 4). In addition, both IGF-I receptor mRNA and IGF-II/M6P receptor mRNA was demonstrated in CaCo-2 cells by Northern blotting and solution hybridization tech- niques (Fig. 5A, B, Fig. 6A, B). Differential splicing of IGF-I receptor RNA is suggested by the solution hy- bridization data. These results are in agreement with previous results demonstrating the existence of both IGF-I receptors and IGF-II/M6P receptors in human colonic tumors and tumor cell lines (Lambert et al.. 1990, 1991; Laburthe et al., 1988). These data suggest that CaCo-2 cells might be responsive to IGFs. Since IGF-II has been implicated in a possible autocrine loop in gastrointestinal cells, we have studied the ex- pression of the IGF genes and the putative secretion of IGFs from CaCo-2 cells in vitro. CaCo-2 cells express IGF-II mRNA but no or only very small amounts of IGF-I mRNA as shown by Northern blotting and solu- tion hybridization techniques and secrete IGF-II im- munoreactivity. It is conceivable that IGF-II plays an autocrine role in the multiplication of CaCo-2 cells in

A. Hoeflich et al. /Molecular and Cellular Endocrinology I01 (1994) 141-150 149

vitro. This putative autocrine loop involving IGF-II would be mediated via an IGF-I receptor: IGF-II has been shown to act via the IGF-I receptor in a number of other tumor cells such as neuroblastoma (El Badry et al., 1989, 1990), rhabdomyosarcoma (Minniti et al., 1992) and MCF-7 breast carcinoma cells (Osborne et al., 1989, 1990). In addition, a potential role of the IGFs as autocrine growth stimulants for lung cancer cells (Havemann et al., 1990; Nakanishi et al., 1988), smooth muscle tumors and sarcomas (Hoppener et al., 1988; Roholl et al., 1990) pheochromocytomas and Wilm’s tumors (Scott et al., 1985; Hasselbacher et al., 1987; Gansler et al., 1989) has been suggested. Re- cently, a putative role of the IGFs in general and of IGF-II in particular for the development and growth of gastrointestinal tumors has been suggested by a num- ber of different laboratories (Tricoli et al., 1986; Win- kler et al., 1993; Schneid et al., 1992; Remacle-Bonnet et al., 1992, Lambert et al., 1991, 1990). The IGFs are thought to act through the IGF-I receptor in many cells and tissues (Kiess et al., 198’7, 1993; Nissley and Lopaczynski, 1991; Rechler and Nissley, 1990). The IGF-II/M6P receptor on the other hand might be important for targeting lysosomal enzymes to lyso- somes (Kornfeld, 1992; Kiess et al., 1993). LysosomaI degradation could then play a role in tumor metastasis in vivo. Since IGF-II is capable of modulating lysoso- ma1 enzyme binding to the IGF-II/M6P receptor (Kiess et al., 1993), the hormone IGF-II might also play an important role in tumor metastasis. Alternatively, IGF- II could act as a differentiation factor in CaCo-2 celIs since these tumor cells are known to spontaneously differentiate during long term culture into gastroin- testinal epithelial cells. Indeed, a possible role for IGF-II as a promoter of differentiation of HT-29-D4 human colon carcinoma cells has been put forward recently (Remacle-Bonnet et al., 1992). It will be im- portant to further explore the putative functions and signaling mechanisms of IGF-II synthesis in CaCo-2 cells in order to explore the role of the IGFs in the gut and to possibly find new therapeutic potentials in colon carcinomas in vivo.

5. Acknowledgements

The generous support from Deutsche Forschungsge- meinschaft, Bonn, Germany [DFG, Ki 365/1.1.-1.3.1, and Deutscher Akademischer Austauschdienst, Bonn, Germany, is kindly acknowledged. We are grateful to Drs. J. Schwander (Basel, Switzerland), R. Voutilainen (Helsinki, Finland), W. Sly (St. Louis, USA) and A. Ullrich (Munich, Germany) for their kind gifts of cDNA probes. We wish to thank Dr. A. Skottner (Stockholm, Sweden) and Dr. K. Mueller (Basel, Switzerland) for recombinant human IGF-I and IGF-II. We also wish to

express our gratitude to Dr. Charles Roberts (Be- thesda, USA) for his kind gift of riboprobes and invalu- able advice and support. This study was supported in part by a grant from the Friedrich Baur Stiftung of the University of Munich, Germany.

This study was presented in part at the Fourth Joint Meeting of the LWPES/ESPE, June 1993, San Fran- cisco, USA.

6. References

Blum, W.F., Ranke, M.B. and Bierich, J.R. (1988) Acta Endoerinol. Copenhagen 118374-380.

Bondy, C.A., Werner, H., Roberts, C.T.J. and LeRoith, D. (1990) Mol. Endocrinol. 4, 1386-1380.

Daughaday, W.H. (1990) Endocrinology 127, 1-4. Daughaday, W.H., Mariz, I.K. and Blethen, S.L. (1980) J. Clin.

Endocrinol. Metab. 51, 781-788. El Badry, O.M., Romanus, J.A., Heirnan, L.J., Cooper, M.J., Rech-

ler, M.M. and Israel, M.A. (1989) J. Clin. Invest. 84, 829-839. El Badry, O.M., Minniti, C., Kohn, E.C., Houghton, P.J., Daugha-

day, W.H. and Helman, L.J. (1990) Cell Growth Differ. 1, 325 331.

Funk, B., Kessler, U., Eisenmenger, W., Hansmann, A., Kolb, H.J. and Kiess, W. (1992) J. Clin. Endocrinol. Metab. 75, 424-431.

Gansler, T., Furlanetto, R., Gramling, T.S., Robinson, K.A., Blocher, N., Buse, M.G., Seus, D.A. and Garvin, A.J. (1989) Am. J. Pathoi. 135, 961-966.

Haselbacher, G.K., Irminger, J.C., Zapf, J., Ziegler, W.H. and Hum- bel, R.E. (1987) Proc. Natl. Acad. Sci. USA 84, 1104-1106.

Havemann, K., Rotsch, M., Schoneberger, H.J., Erbil, C., Hennig, C. and Jaques, G. (1990) J. Steroid Biochem. Mol. Biol. 37, 877-882.

Heinz-Erian, P., Kessler, U., Funk, B., Gais, P. and Kiess, W. (1991) End~rinoIo~ 129, 1769-1778.

Hernandez, E.R., Hurwitz, A., Vera, A., Pellicer, A., Adashi, E.Y., LeRoith, D. and Roberts, C.T.J. (1992) J. Clin. Endocrinol. Metab. 74, 419-425.

Hoppener, J.W.M., Mosselman, S., Roholl, P.J.M., Lambrechts, C., Glebos, R.J., de Pagter-Holthuizen, P., Lips, C.J., Jansen, H.S. and Sussenbach, J.S. (1988) EMBO J. 7, 1379-1385.

Humbel, R.E. (1990) Eur. J. Biochem. 190, 445-462. Kiess, W., Haskell, J.F., Lee, L., Greenstein, L.A., Miller, B.E.,

Aarons, A.L., Rechler, M.M. and Nissley, S.P. (1987) J. Biol. Chem. 262, 12745-12751.

Kiess, W., Hoeflich, A., Yang, Y., Flyvbjerg, A. and Barenton, B. (1993) in IGFs: Molecular and Cellular Aspects II (LeRoith, D. and Raiszada, M.H., eds.), Plenum Press, New York (in press).

Kornfeld, S. (1992) Annu. Rev. Biochem. 61,307-330. Laburthe, M., Rouyer-Fessard, C. and Gammeltoft, S. (1988) Am. J.

Physiol. 254, G457-G462. Laemmli, U.K. (1970) Nature 227, 680-685. Lambert, S., Vivario, J., Boniver, J. and Gol-Winkler, R. (1990) Int.

J. Cancer 46, 405-410. Lambert, S., Collette, J., Gillis, J., Franchimont, P., Desaive, C. and

Gol-Winkler, R. (1991) Int. J. Cancer 48, 826-830. Mathieu, M., Rochefort, H., Barenton, B., Prebois, C. and Vignon,

F. (1990) Mol. Endocrinol. 4, 1327-1335. Minniti, C.T., Kohn, E.C., Grubb, J.H., Sly, W.S., Oh, Y., Muller,

H.L., Rosenfeld, R.G. and Helman, L.J. (1992) J. Biol. Chem. 267, 9000-9004.

N~anishi, Y., Mulshine, J.L., Kasprzyk, P.G., Natale, R.B. and Cuttitta, F. (1988) J. Clin. Invest. X2, 354-359.

Nisstey, S.P. and Lopaczynski, W. (1991) Growth Factors 5, 29-43.

150 A. Hoeflich et al. /Molecular and Cellular Endocrinology 101 (1994) 141-150

Osborne, C.K., Coronado, E.B., Kitten, L.J., Arteaga, CL, Fuqua, Rotwein, P. (19911 Growth Factors 5, 3-18. S.A., Ramasharma, K., Marshall, M. and Li, C.H. (1989) Mol. Rouyer-Fessard, C., Gammeltoft, S. and Laburthe, M. (1990) Gas- Endocrinol. 3, 1701-1709. troenterology 98, 703-707.

Osborne, C.K., Clemmons, D.R. and Arteaga, C.L. (19901 J. Steroid.

Biochem. Mol. Biol. 37, 805-809.

Park, J.H.Y., McCusker, R.H., Vanderhoof, J.A., Mohammadpour,

H., Harty, R.F. and MacDonald, R.G. (1992) Endocrinology 131,

1359-1368.

Scatchard, D.G. (19491 Ann. NY Acad. Sci. 51, 660-674.

Scott, J., Cowell, J., Robertson, M.E., Priestley, L.M., Wadey, R. and

Hopkins, B. (1985) Nature 317, 260-262.

Schneid, H., Seurin, D., Noguiez, P. and Le Bout, Y. (1992) Growth

Regul. 2, 45-54.

Pillion, D.J., Haskell, J.F., Atchinson, J.A., Ganapathy, V. and

Leibach, F.H. (19891 Am. J. Physiol. 257, E26-E34.

Read, L., Lemmey, A.B., Howarth, G.S., Martin, A.A., Tomas, F.M.

and Ballard, F.J. (1990) in: Modern Concepts of Insulin-like

Growth Factors (Spencer, E.M., ed.), pp. 225-234, Elsevier,

Holland.

Rechler, M.M. and Nissley, S.P. (1990) in Handbook of Pharmacol-

ogy, Peptide Growth Factors and Their Receptors I (Sporn, M.S.

and Roberts, A.B., eds.), pp. 263-367, Springer-Verlag, Heidel-

berg/New York.

Termanini, B., Nardi, R.V., Finan, T.M., Parikh, I. and Korman,

L.Y. (19901 Gastroenterology 99, 51-60.

Tricoli, J.V., Rail, L.B., Karakousis, C.P. Herrera, L., Petrelli, N.J.,

Bell, G.I. and Shows, T.B. (1986) Cancer Res. 46, 6169-6173.

Vanderhoof, J.A., McCusker, R.H., Clark, R., Mohammadpour, H.,

Blackwood, D.J., Harty, R.F. and Park, J.H.Y. (1992) Gastroen-

terology 102, 1949-1956.

Remacle-Bonnet, M., Garrouste, F., El Atio, F., Roccabianca, M.,

Marvaldi, J. and Pommier, G. (1992) Int. J. Cancer 52, 910-917.

Roholl, P.J., Skottner, A., Prinsen, I., Lips, C.J., Den Otter, W. and

Van Unnik: J.A. (19901 Histopathology 16, 455-460.

Winkler, R., Lambert, S., Collete, J., Carlisi, A., Hodzik, D. and

Bensbaho, K. (1993) Abstract No. 1137, 75th Annual Meeting,

Endocrine Society, Las Vegas, USA.

Yang, Y., Hoeflich A., Kessler, U. and Kiess, W. (1993) Regul. Peptides 48, 41-53.