Int. J. Cancer: 65,812-820 (1996) 0 1996 Wiley-Liss, Inc.

Publication of the International Union Against Cancer Publication de I'Union Internationale Contre le Cancer

VARYING PATTERNS OF EXPRESSION OF INSULIN-LIKE GROWTH FACTORS I AND I1 AND THEIR RECEPTORS IN MURINE MAMMARY ADENOCARCINOMAS OF DIFFERENT METASTASIZING ABILITY Fabiana K. GUERRA', h a Maria EIJAN~, Lydia PURICELL13, Daniel F. ALONSO~, Elisa BAL DE KIER J0FFk3, Albert0 R. KORNBLIHTT~, Eduardo H. CHARREAU~ and Patricia V. ELIZALDE1,4

lInstituto de Biologia y Medicina Experimental, 21nstituto de Investigaciones en Ingeniena Genktica y Biologia Molecular, and 31nstituto de Oncologi A.H. Roffo, Buenos Aires, Argentina.

We studied the expression of insulin-like growth factors I (IGF-I) and II (IGF-II) and their receptors (IGF-R) in 2 related murine mammary adenocarcinoma in vivo lines, M3 and MM3, with different metastasizing ability. We further investigated the effects of IGFs on the secretion of a key enzyme in the metastatic cascade, the urokinase-type plasminogen activator @PA) in M3 and MM3 cells. M3 is a spontaneous mammary tumor originated in BALB/c mice, with a 40% incidence of lung metastases. MM3 variant, obtained by successive S.C. implants of M3 lung metastases into syngeneic mice, shows a 95% incidence of lung metastases. Similar levels of expression of IGF-l protein were found in M3 and MM3 tumors, whereas IGF-II expression was 4-fold higher in MM3. RNAse protection assays showed similar levels of IGF-l mRNA in M3 and MM3 tumors and revealed a 4-fold increase in IGF-ll transcripts in MM3 tumors compared with M3. Authentic IGF-I and II mes- sages were also found in primary cultures of M3 and MM3 cells. IGF-l mRNA levels were similar in both cultures and, as described for solid tumors a 5-fold increase in IGF-II message was detected in MM3 cells. The presence of type I and mannose-6-phosphate (Man-6P)ltype II IGF-R was demon- strated in both M3 and MM3 tumors. A 2-fold increase of type I IGF-Rwas detected in MM3 tumors compared with M3. Man-6P/ type II IGF-R levels were 2-fold lower in MM3 tumors than in M3. As observed in tumor membranes, type I IGF-R concentra- tions were higher and Man-6Pltype II IGF-R lower in cultures of MM3 epithelial cells compared with MM3 cells. In addition, we found that IGF-I enhanced secreted uPA activity in both M3 and MM3 cells while IGF-ll only stimulated uPA secretion in MM3 cells. o 1996 Wiley-Liss, Inc.

Tumor invasion and the development of metastasis is a complex and multistep process in which unique properties are acquired by tumor cells. These properties include limitless growth, alterations in cell communication, enhanced motility and the capacity to degrade basement membrane components, invade tissues and grow autonomously at secondary sites (Nicolson, 1993). Autonomous tumor cell proliferation at the primary and secondary sites may be associated with the production of growth factors (GFs) acting by an autocrine mechanism (Kerbel, 1990). These constitutively produced GFs could also stimulate the growth of surrounding cells in a paracrine way, which in turn may contribute to tumor progres- sion.

As a way of linking the expression of GFs and their receptors to the metastatic process, we examined our previously de- scribed in vivo model of metastasis for mammary cancer. M3 is a mammary adenocarcinoma of spontaneous origin in BALB/c mice with a 40% incidence of lung metastases. MM3 variant, obtained by successive S.C. implants of M3 lung metastases into syngeneic mice, shows a 95% incidence of lung metastases (Bal de Kier Joffe et al., 1986).

We have focused our study on the insulin-like growth factor (IGF) family of polypeptides. The complex IGF system is composed of 2 ligands (IGF-I, IGF-11), 3 receptors [type I IGF receptor (IGF-R), mannose-6-phosphate (Man-6P)/type I1 IGF-R and an insulin/type I IGF hybrid receptor] and 6 IGF binding proteins (IGF-BPsl4) (for review, see Yee, 1994). IGFs are strongly mitogenic for many breast cancer cell lines (Os- borne et al., 1989; Cullen et al., 1990), and several breast cancer

lines express IGF-I1 (Yee et al., 1988; Osborne et al., 1989); IGF-I mRNA has been detected in stromal cells of breast cancer tissue (Paik, 1992), and IGF-Rs have been found in breast cancer cells (Osborne et al., 1989; Cullen et aL, 1990; Mathieu et al., 1990,1991; Peyrat and Bonneterre, 1992). Little is known about the involvement of IGFs in breast cancer metastasis; however, the role of IGF-I and IGF-I1 as chemotac- tic agents in melanoma and rhabdomyosarcoma cells (Stracke et al., 1989; El-Badry et al., 1990) suggests that IGFs may also be involved in tumor invasion and metastasis of other malig-, nant cell types.

The present study was undertaken to examine IGFs and IGF-R expression in M3 and MM3 tumors and to investigate a possible association between the pattern of expression of IGFs and IGF-R in M3 and MM3 tumors and their different metastatic abilities. To elucidate the cellular origin of the IGFs present in the M3 and MM3 tumor tissue and the cell type localization of the IGF-R, we also studied primary cultures of malignant M3 and MM3 epithelial cells, which are devoid of host contaminating cells (Bal de Kier Joffe et al., 1986).

MATERIAL AND METHODS Tumors

We used 2 previously characterized transplantable mam- mary adenocarcinomas with different lung metastasizing poten- tial (Bal de Kier Joffe et al., 1986). M3 tumor, of spontaneous origin in a BALB/c female mouse, is maintained by S.C. transplantation of tumor fragments and has a 40% incidence of lung metastases. The MM3 variant (Bal de Kier Joffe et al., 1986) was obtained by successive S.C. transplants of M3 lung metastases into the flanks of syngeneic mice. Once MM3 achieved a stable behavior, it was further maintained by S.C. grafts of tumor fragments. This variant showed a 95% inci- dence and a higher number of lung metastases. Female BALB/c mice aged 10 weeks received S.C. trocar transplants of M3 and MM3 tumor fragments, 2 mm in diameter. Tumor- bearing animals were used at two-thirds of the survival time (30 days for M3 or 40 days for MM3).

Primaiy cultures M3 or MM3 tumors from three animals were surgically

removed under sterile conditions. Tumors were minced into small fragments and disaggregated using 0.01% Pronase (Sigma, St. Louis, MO) and 0.0035% DNAse (Sigma). Cells were resuspended in MEM (GIBCO, Grand Island, NY) supple- mented with 10% FCS (Gen, Buenos Aires, Argentina) and 2 mM L-glutamine (Sigma). Viability, as assayed by the Trypan blue exclusion test, was higher than 80%. The cells (1.5 x lo5

4To whom correspondence and reprints requests should be ad- dressed, at the Instituto de Biologia y Medicina Experimental, Obligado 2490, Buenos Aires 1428, Argentina.

Received: July 14,1995 and in revised form November 21,1995.

IGFS IN METASTASIZING MURINE MAMMARY CANCER 813

cells/well) were plated in 6-multiwell plastic dishes for binding assays or in flasks (5 x lo6 cells) for RNA extraction. Both assays were performed when the monolayers reached 90% confluence. In a previous work we demonstrated that primary M3 and MM3 cultures are composed of 98% epithelial malignant cells and 2% contaminating host cells (Bal de Kier Joffe et al., 1986).

Acid extraction and IGFpurification by C18 chromatography Tumor samples (600-1,000 mg) were homogenized 1:4 (w/v)

in 1 M acetic acid with 0.5 mM phenylmethylsulfonyl fluoride (PMSF), 0.025 mM N-tosyl-L-phenylalanine chloromethyl ke- tone (TPCK), 0.025 mM N-a-p-tosyl-L-lysine chloromethyl ketone (TLCK) and 0.025 mM N-CBL-L-phenylalanine chlo- rometyl ketone (ZPCK). Samples were extracted overnight at 4°C and then centrifuged at 4°C for 30 rnin at 1,500g. The resulting precipitates were reextracted for 2 hr with 1 M acetic acid, with proteolysis inhibitors, and the extracts were centri- fuged as described above. Both supernatants were mixed (final volume of 2-3 ml), and 500 p1 of the mixture was loaded on to a Sep-Pack C-18 cartridge (Waters, Millipore, Bedford, MA) for quantitative separation of IGF-I and I1 from IGF-BPs, as described by Hsu and Hammond (1987). The separation procedure was performed twice in sequence on each sample. Recovery of lZ5I-IGF-I and -11 by this method was about 60-70%. To test that IGF-BPs had been adequately removed, a charcoal separation binding assay was performed. Briefly, 1251-IGF-I or -11 (40,000 dpm) were incubated with aliquots from each sample in buffer (Tris 25 mM, MgCI2 10 mM, pH 7.4). After an overnight incubation at 4"C, activated charcoal was used to separate free from charcoal-bound 1251-IGF-I or -11 as described by Moses et al. (1979), and radioactivity in the supernatant was counted. No residual IGF-BPs were detected with this assay.

Iodination andpurification of IGF-I and -11 Recombinant human IGF-I and IGF-I1 were kindly pro-

vided by Dr. J. Merryweather (Chiron, Emeryville, CA) and carrier-free 1251 was purchased from NEN (Boston, MA). 1251 labeling of IGF-I and -11 was performed by the chloramine T method as previously described for epidermal growth factor (Molinolo et al., 1987), and purification was done in Sephadex G-50 columns equilibrated and eluted with buffer (phosphate 0.05 M, NaCl 0.1 M with 0.05% Tween 20). Specific activity was about 150-175 pCi/pg for both peptides. Maximal binding capacity was determined using a full-term human placenta membrane preparation as a source of IGF-I receptors and M3 or MM3 membrane preparations as sources of IGF-I1 recep- tors.

IGF-I radioimmunoassay The incubationvolume of 300 pI consisted of 100 p1 of IGF-I

standard or samples, 100 p1 of anti-rabbit IGF-I (National Hormone and Pituitary Program, NIADDK, NIH, Bethesda, MD) (1:5,000 final dilution) and 100 p1 of lZ5I-IGF-I (20,000 dpm). The assay buffer was phosphate 0.05 M, NaCl 0.1 M, with sodium azide 0.005% and Tween-20 0.05%, pH 7.4). The tubes were incubated for 2 days at 4°C and then 100 pl of normal rabbit serum (1:3 v/v) and 100 p1 of second antibody (anti-rabbit IgG from the National Hormone and Pituitary Program, NIADDK), were added. After 12 hr of incubation, the tubes were centrifuged at 3,OOOg for 20 min, the superna- tant was removed by aspiration and the radioactivity in the pellet was counted. Standards and samples were determined in triplicate. The detection limit of this assay was 19.5 pg.

IGF-II radioreceptor assay (RRA) The amount of IGF-I1 present in the M3 and MM3 tumor

samples or Bio Gel (BioRad, Hercules, CA) P-60 chromatogra- phy fractions was quantified using a radioreceptor assay. M3 and MM3 microsomal membranes without IGF-BP contamina- tion were obtained using the method described by Barenton et

al. (1988). Aliquots of 100 p1 of microsomal fraction (30-50 pg protein) were incubated in 0.3 ml of buffer (Tris 25 mM, MgC12 10 mM, pH 7.4) containing 1251-IGF-II (0.6 nM) and increasing amounts of unlabeled IGF-II(O.01-100 nM) or samples. After an overnight incubation at 4"C, 2 ml of cold buffer was added to each tube, the particulate membranes were separated by centrifugation at 3,OOOg for 15 rnin and the radioactivity was counted. RRA experiments were repeated for at least 3 times with each individual tumor extract. The RRA sensitivity was 0.01 ng/assay. The intra- and interassay CVs were below 10% and 15%, respectively.

IGF-I and II receptor binding assay Binding to tumor membranes. Aliquots of 100 ~1 of M3 or

MM3 microsomal membranes (30-50 pg protein), obtained as described by Barenton et al. (1988), were used. For IGF-I binding studies microsomal pellets were incubated with 1251- IGF-I (33-1,000 pM) in a final volume of 250 p1 of incubation buffer. Nonspecific binding was determined by adding an excess of 500 ng of IGF-I. For IGF-I1 binding studies micro- soma1 pellets were incubated with a constant amount of 1251-IGF-II tracer (600 pM) and with increasing amounts of unlabeled IGF-II(O.01-150 nM) in the same final volume and buffer as above. Nonspecific binding was determined by adding an excess of 500 ng of IGF-11. After an overnight incubation at 4"C, bound IGF was separated by filtration through HAWP filters (Millipore), and the retained radioactivity was counted. Assays were done in triplicate, and the results were analyzed by the Scatchard method (1949).

Binding to intact cell monolayers. To prepare cultures for binding studies, the cells were seeded in 6-multiwell plastic dishes and grown until monolayers reached 90% confluence (approximately 1 x lo6 cells/well). The medium was discarded and the cells were washed 3 times with PBS. IGF binding studies were performed using the same concentration of ligands as described above in 1 ml of binding buffer (NaCI 125 mM, KCI 5 mM, CaC12 1.2 mM, MgS04 1.2 mM, HEPES 50 mM, pH 7.5) containing 2 mg/ml of BSA for 2 hr at 4°C. Then the radioactive medium was aspirated and the monolayers were washed 3 times with ice-cold PBS. The cells were dissolved with 1% Triton X-100-PBS buffer, the solution transferred to disposable tubes and the radioactivity counted. Assays were performed in triplicate and results were analyzed as above.

Cross-linking, electrophoresis, and autoradiography Aliquots from 200 to 500 pg of membrane protein were

incubated with 2 nM 1251-IGF-I or -11 in the presence or absence of an excess of 500 ng of non-radioactive IGF-I or -11 in buffer (Tris 25 mM, MgCI2 10 mM, pH 7.4). Incubation proceeded overnight at 4°C and then the excess unbound IGF-I or -11 was washed out by dilution with PBS buffer and centrifugation at 12,OOOg for 2 min. The resulting pellet was resuspended in 195 pl of PBS Dulbecco's buffer, and disuccin- imidyl suberate (DSS) (Pierce, Rockford, IL), freshly dis- solved in DMSO, was added to a final concentration of 0.2 nM. The cross-linking reaction was allowed to proceed for 15 rnin at 4"C, and then it was quenched with Tris 1 M, EDTA 1 mM, pH 7.4 (100 pl). The incubation mixture was centrifuged at 12,OOOg for 2 min and the pellet was resuspended in 50 p1 of buffer (Tris 10 mM, EDTA 1 mM, pH 6.8) with 2% SDS and 0.005% bromophenol blue with 5% mercaptoethanol. Samples were boiled for 3 rnin and centrifuged at 12,OOOg for 4 rnin to eliminate insoluble material. Electrophoresis on 7.5% polyacryl- amide gels was performed according to the method of Laemmli (1970). After electrophoresis gels were stained overnight with Coomassie brilliant blue R-250, destained and dried. Gels were then exposed to Kodak X-Omat AR films with the use of Cronex intensifying screens (Dupont, Wilmington, DE). Mo- lecular weight markers were: myosin (200,000 kDa), phosphory- lase B (97,400 kDa), BSA (68,000 kDa), ovalbumin (43,000

GUERRA ETAL 814

kDa) and chymotrypsinogen (25,700 kDa) (BRL, Gaithers- burg, MD).

Probes IGF-I and -11 mouse cDNA clones were kindly provided by

Dr. P. Rotwein (Washington University School of Medicine, St. Louis, MO). The IGF-I clone, constructed in a pGEM 4 vector (Promega, Madison, WI), was linearized with BamHI providing a template for antisense riboprobe protecting a 185 bp mRNA fragment. The IGF-I1 clone, constructed in the vector pBluescript I1 (Stratagene, La Jolla, CA) was linearized with BamHI providing a template for antisense riboprobe protecting a 152 bp mRNA fragment. The rpL32 human cDNA clone (encoding ribosomal protein L32) (Young and Trowsdale, 1985) was provided by Dr. R. Rochford (The Scripps Research Institute, La Jolla, CA). This recombinant constructed in a pGem 4 vector (Promega), was linearized with EcoRI providing a template for antisense riboprobe protecting a 76 bp mRNA fragment.

RNAse protection assay Total cellular RNA from M3 and MM3 tumor specimens

and from primary epithelial cell cultures was isolated using the guanidinium thiocyanate-cesium chloride method (Chirgwin et al., 1979). All RNA concentrations were determined spectro- photometrically. The concentration and integrity of the RNAs were confirmed on denaturing 1% agarose minigels containing 2.2 M formaldehyde prior to use. Antisense 32P-labeled RNA probes were transcribed (IGF-I and rpL32 with T7 RNA polymerase, IGF-I1 with T3 RNA polymerase) following the manufacturer’s protocol (Promega). Fifty micrograms of total RNA were hybridized overnight at 50°C with 2 X lo5 cpm of each probe in 30 p1 of buffer containing 80% (vol/vol) formamide, 40 mM piperazine-N,N’-bis (2-ethanesulfonic acid), 0.4 M NaCl and 0.1 M EDTA. All RNA samples were hybridized with IGF-I or -11 probes and rpL32 probe (used to correct for small variations in the amount of RNA loaded) simultaneously. Samples were then digested with 40 pg/ml RNase A (Sigma) for 30 min at 25°C. Digestion was terminated with proteinase K and SDS. The samples were extracted with phenol/chloroform/isoamyl alcohol (20:20: 1) and precipitated with 20 pg of tRNA (Sigma) and 2 vol of absolute ethanol. The pellets were resuspended in 5 pl of an 80% formamide loading buffer and run on a 6% polyacrylamide sequencing gel with 8 M urea. Size markers were prepared by end-labeling HinfI- digested fragments of PFH3 plasmid. The gels were dried and exposed to an X-ray film in the presence of an intensifying screen at -70°C for 1 to 7 days. Band intensities were quantitated by scanning the autoradiograms using an LKB (Bromma, Sweden) Ultroscan XL densitometer.

In vitro proliferation studies The effect of IGF-I and -11 on log-phase proliferating cells in

the presence or absence of FCS was studied as follows: 24-multiwell trays (Falcon, Oxnard, CA) were seeded with 5 x lo4 M3 or MM3 tumor cells in growth medium (MEM plus 10% FCS and 2 mM L-glutamine). After 18 hr the wells were washed with PBS and supplied with MEM with or without 2% FCS. IGFs were added at different concentrations (1.25-80 ng/ml) per triplicate. As a positive control some cells were cultured in the presence of 10% FCS. Cells were refed and samples added every 2 days. Cell growth was evaluated by cell count and by the measurement of DNA content (Labarca and Paiyen, 1980). Each experiment was repeated 3 times. The effect of the IGFs on DNA synthesis in quiescent M3 or MM3 cells was studied according to Furlanetto et al. (1987).

Urokinase-type plasminogen activator (uPA) activity assay and zymography

The effect of IGF-I and -11 on the production of uPA by M3 and MM3 cells was assayed in conditioned medium. Semi-

confluent cells cultured in 24-well plates were extensively washed with PBS and then incubated in serum-free MEM with or without IGF-I (2.5 ng/ml) or IGF/II (5 ng/ml) for 24 hr. Conditioned media were individually harvested, clarified, ali- quoted and stored at -40°C. The remaining monolayers were trypsinized and cell number was evaluated by hemocytometer counting. Secreted uPA activity was measured with a radial fibrinolytic assay using plasminogen-rich fibrin plates as previ- ously reported (Alonso et al., 1993). Purified human uPA (Serono, Buenos Aires, Argentina) was used as standard and the results were expressed in uPA IU/106 cells. Plasminogen- free fibrin-plates were used to test plasminogen-independent activity.

Zymographic analysis of uPA was performed according to the method of Granelli-Piperno and Reich (1978). Condi- tioned media were electrophoresed on SDS polyacrylamide 10% slab gels. Gels were washed in 2.5% Triton X-100 and applied to a plasminogen-rich fibrin-agarose underlay. After incubation for 16 hr at 37°C in a moist atmosphere, lysis zones were visible. Molecular weights were determined using stan- dard prestained m.w. markers (Sigma). Some samples were preincubated with anti-catalytic rabbit polyclonal antibody to murine uPA, kindly provided by Dr. G. Hoyer Hansen (Righos- pitalet, Copenhagen, Denmark) at 4°C for 45 min before the zymographic assay. Control experiments demonstrated com- plete anti-catalytic specificity for murine uPA from mouse urine.

RESULTS Expression of IGF-I and IGF-II at theprotein and mRNA levels

Expression of IGF-I and -11 at the protein level was detected in both M3 and MM3 tumors. There were no significant differences in the levels of IGF-I between M3 and MM3 tumors (Table I), but a 4-fold increase in IGF-It levels was detected in MM3 compared with M3 tumors (Table I).

Expression of IGF-I and IGF-I1 at the mRNA level was studied using an RNAse protection assay. Authentic IGF-I and IGF-I1 messages were detected in M3 and MM3 tumors. No differences were found in the levels of IGF-I mRNA between M3 and MM3 tumors (Fig. la), but IGF-I1 mRNA levels were approximately 4-fold higher in MM3 tumors compared with M3 (Fig. 16). Since we had previously shown that 98% of the cells in primary M3 and MM3 cultures are epithelial cells (Bal de Kier Joffe et al., 1986), we decided to assess the epithelial origin of IGF-I and I1 mRNA from M3 and MM3 tumor pecimens using primary cultures of these tumors. Both IGF-I and I1 messages were detected in these cultures. IGF-I mRNA levels were similar in both cultures, and a 5-fold increase in IGF-I1 message was detected in MM3 cells, as described for solid tumors (Fig. 2u,b). IGFs receptors

Binding studies of IGFs were performed on washed micro- soma1 membranes from M3 and MM3 tumors, to avoid interference due to the possible presence of IGF-BP, and in intact M3 and MM3 cell monolayers. In both M3 and MM3

TABLE I - QUANTIFICATION OF IGF-I AND IGF-I1 IN M3 AND MM3 TUMORS’

IGF-I IGF-I1 (ngig tissue) (ngig tissue) Tumor

M3 50.8 2 9.1 180.1 2 40.3 MM3 48.2 5.7 741.9 2 148.9* ‘The amounts of IGF-I by RIA and IGF-I1 by RRA were

calculated after acid extraction and Sep-Pack C18 chromatogra- phy as described in Material and Methods. Each value represents mean k SD of 10 tumor samples.-$ < 0.001 compared with M3 tumors by Student’s t test.

815 IGFS IN METASTASIZING MURINE MAMMARY CANCER

FIGURE 1 - IGF-I and -11 expression in M3 and MM3 tumors. IGF-I (a) and IGF-I1 (b) RNAse protection assays. Fifty micrograms of total RNA from M3 and MM3 tumors were hybridized with IGF-I or IGF-I1 probes and rpL32 probe as described in Material and Methods. Total RNA from mouse liver was used as the positive control for IGF-I expression and fetal mouse liver for IGF-I1 expression. The IGF-I-protected fragment is 182 bp, the IGF-I1 protected fragment 152 bp and the rpL32 protected fragment is 76 bp. MWM: m.w. markers.

membranes specific IGF-I binding was detected; higher levels of type I IGF-R were observed in MM3 tumors (Table 11). Scatchard plots of the binding data in both M3 and MM3 tumors were curvilinear, suggesting the existence of multiple classes of binding sites, with intrinsically different affinities. When data were analyzed using the Ligand Program (Munson and Rodbard, 1980) for a 2-site model, the presence of high-affinity (M3: Kd = 2.1 X M, Q = 18.2 fmol/mg protein; MM3: Kd = 3.1 X 10-l2 M, Q = 33.8 fmol/mg pro- tein) and low-affinity sites (M3: Kd = 1 x M, Q = 45.2 fmol/mg protein; MM3: Kd = 1.4 x M, Q = 75.8 fmol/mg protein) was determined. Aflinity labeling studies were performed to characterize the IGF-I receptors further. When analyzed by SDS-PAGE under reducing conditions, membrane preparations from M3 and MM3 tumors showed 2 labeled bands on autoradiograms (Fig. 3). The most intensely labeled band appeared in the Mr 130,000 region correspond- ing to the a-subunit of type I IGF-R. The less intensely labeled band had an apparent Mr of 260,000. The presence of an excess of 500 ng of unlabeled IGF-I during incubation abol- ished the affinity labeling of both bands. The absence of the 260,000 minor band suggests that it may represent a cross- linked dimer of type I IGF-R a-subunit rather than type I1 receptor. The relative potencies of unlabeled IGF-I, IGF-11, and insulin to inhibit 1251-IGF-I binding competitively in M3 and MM3 tumor membranes were determined (Fig. 4). Half- maximal inhibition was obtained with 0.7 nM of IGF-I in M3 membranes and 0.67 nM IGF-I in MM3. However, 1251-IGF-l

binding was only partially inhibited by IGF-I1 or insulin (40% and 30%, respectively) in both M3 and MM3 membranes.

Specific binding of IGF-I1 to tumor membranes was also found. A significantly lower number of MandP/type I1 IGF-R was detected in MM3 tumors compared with M3 (Table 11). Scatchard analysis of IGF-I1 binding data showed one class of high-affinity IGF-I1 binding sites in both M3 (Kd = 5.8 x 10-lO M, Q = 23,900.94 fmol/mg protein) and MM3 tumors (Kd = 5 x 10-lo M, Q = 13,173.55 fmol/mg protein). Chemi- cal cross-linkings of 1251-IGF-II to M3 and MM3 tumors are shown in Figure 5. When analyzed by SDS-PAGE under reducing conditions, membrane preparations from M3 and MM3 tumors showed only one labeled band (apparent Mr 260,000) corresponding to the Man-6P/type I1 TGF-R. The specificity of these receptors was studied by performing lZI- IGF-I1 competitive binding studies in M3 and MM3 tumor membranes (Fig. 6). Unlabeled IGF-I1 was able to inhibit 1251-IGF-II binding completely in a dose-dependent manner, with a half-maximal displacement at a concentration of 2 nM in M3 membranes and 1 nM in MM3. IGF-I 80 nM inhibited 50% of the 1251-IGF-II binding in both M3 and MM3 mem- branes, whereas the same concentration of insulin was not effective .

IGF binding was also studied in monolayer cultures of M3 and MM3 cells. Higher levels of type I IGF-R were found in primary cultures of MM3 than in M3 cells (Table HI), as observed in tumor membranes. Representative Scatchard plots are shown in Figure 7a. Plots from M3 and MM3 cells in

GUERRAETAL 816

FIGURE 2-IGF-I and -11 expression in primary cultures of M3 and MM3 epithelial cells. (a and b) IGF-I and IGF-I1 RNAse protection assays. Fifty micrograms of total RNA from M3 and MM3 epithelial cell cultures were hybridized with IGF-I or IGF-I1 probes and rpL32 probe as described in Material and Methods. Positive controls are the same as in Figure 1. The IGF-I protected fragment is 182 bp, the IGF-I1 protected fragment is 152 bp and the rpL32 protected fragment is 76 bp. MWM: m.w. markers.

TABLE I1 - IGF-R LEVELS IN M3 AND MM3 TUMOR MEMBRANES'

Type 1 IGF-R (fmolimg orotl (fmolimz mot)

Man-6Pitype I1 IGF-R Tumor

M3 42.2 * 4.9 24,710.1 t 2,110.5

MM3 76.9 t 6.62 14,129.5 t 1,973.3* (n = 6)

(n = 6)

(n = 5 )

(n = 5 ) 'Binding assays to 30-50 pg membrane protein were performed

in the conditions described in Material and Methods. The number of IGF-Rs was determined using a single-point assay at a saturat- ing dose of 1.58 nM for IGF-I and 20 nM for IGF-11. Each value is the mean f SD. n, number of tumor samples in each of which the determinations were done in triplicate.-2p < 0.001 compared with M3 tumors by Student's t test.

culture were curvilinear, suggesting the existence of multiple classes of binding sites. Data analysis using the Ligand Pro- gram for a 2-site model showed the presence of high- and low-affinity sites (Fig. 7a) . IGF-I1 binding studies in monolayer cultures detected a lower number of Man-6P/type I1 IGF-R in MM3 cells compared with M3 cells (Table 111). Scatchard analysis of IGF-I1 binding data showed the presence of one class of high-affinity IGF-I1 binding sites in both M3 (Kd = 4.3 x M. 0 = 1.9 x lo6 sites/cell) and MM3 cells (Kd = 1.2 x lO-*O M; Q = 7.4 x lo5 sites/cellj (Fig. 7b).

FIGURE 3 - Chemical cross-linking of 1251-IGF-I to membrane oreDarations of M3 and MM3 tumors. lZ5I-IGF-I was affinity

Effect of exogenous IGFs on cellularproliferation and secreted uPA activity

cross-linked to its receptor in the absence (-) or presence (+) of 500 n unlabeled IGF-I as described in Material and Methods. Sampks of the cross-linked membranes (200 pg of membrane

Neither IGF-I nor IGF-I1? Or at different protein) were electrophoresed on 7.5% polyacrylamide gels accord- ing to Laemmli (1970) in the presence of 5% mercaptoethanol. The autoradiogram from the fixed, dried gel is shown, Arrow: position of the 130 kDa radioligand receptor complexes. MWM: m.w. markers.

in the presence Or absence Of 2% FCS, stimulated the proliferation Of either M3 Or MM3 log-phase growing cells. The same results were obtained by DNA synthesis assays with 3H-thymidine (data not shown).

IGFS IN METASTASIZING MURINE MAMMARY CANCER 817

0

A

I G F - I I G F - I I Insulin

I I

ai 1 10 loo 1000 GROWTH FACTOR CONCENTRATION InH) a

a I G F - I o I G F -1

- 0

\ m

!? cl

LL 0 n

H ln N

b

b Insul in 0- o--o--o -0-oA4c; -

-0; A,

'\ '\ ' 'A .* 0.6 1 .

' GROWTH FACTOR CONCENTRATION LnMl

FIGURE 4 - Competitive inhibition of lZI-IGF-I binding to M3 (a) and MM3 (b) tumor membranes. Fifty micrograms of protein from M3 or MM3 microsomal membranes were incubated with 40,000 c m of I25I-IGF-I and different concentrations of unlabeled IGF-I &), IGF-I1 (0) and insulin (A). After an overnight incubation at 4"C, bound 1251-IGF-I was separated by filtration through HAWP filters as described in Material and Methods. Data are expressed as B/Bo, i.e., ligand bound in the presence of competing unlabeled ligand divided by the quantity of ligand bound in the absence of competition. Non-specific binding, deter- mined using an excess of 500 ng of IGF-I, has been subtracted. Mean of 3 separate experiments.

Zymographic assay of conditioned media from both M3 and MM3 cells showed that PA activity corresponded to a main band of 48 kDa. This PA activity was completely abolished by preincubation with anti-catalytic antibodies to murine uPA (10 Fg/ml or more). Plasminogen-independent activity was not detected in the present assay conditions. IGF-I at a concentra- tion of 2.5 ng/ml significantly enhanced secreted uPA activity in both M3 and MM3 monolayers. When MM3 cells were treated with IGF-I1 (5 ng/ml), a significant increase in se- creted uPA activity was also noted. In contrast, the same concentration of IGF-I1 had no effects on uPA secretion in M3 cells (Table IV, Fig. 8).

DISCUSSION

In this study we demonstrated IGF-I expression at the protein and mRNA level in 2 murine mammary adenocarcino- mas with different metastasizing abilities. We found no differ- ences in IGF-I expression between the highly metastatic variant MM3 and the moderately metastatic M3. A role as a paracrine regulator of breast cancer cell growth has been postulated for IGF-I since it is expressed by stromal but not by epithelial cells in breast tumors (Cullen et al., 1991; Paik, 1992), and it is strongly mitogenic for breast cancer cell lines (Cullen et al., 1990). However, no information is available about the association between IGF-I expression in the primary tumor and the metastatic ability of the tumor. In this report we demonstrated in vitro expression of the authentic IGF-I mes- sage in murine malignant epithelial mammary cells. There is a striking contrast between this finding and the lack of expres- sion of IGF-I mRNA seen in breast cancer cell lines, possibly

FIGURE 5 - Chemical cross-linking of 1251-IGF-II to membrane reparations of M3 and MM3 tumors. IGF-I1 was affinity cross-

inked to its receptor in the absence (-) or the presence (+) of an excess of 500 ng unlabeled IGF-11. Methods are the same as those in Figure 3. Arrow: position of the 260 kDa 1251-IGF-II receptor complexes. MWM: m.w. markers.

because primary cultures retain properties that cells possess in vivo and that are lost by established cell lines. The only other model in which an alternate IGF-I mRNA transcript was found in the malignant epithelial cells is the human breast cancer xenograft T61 grown in nude mice (Tobin et al., 1990).

A significantly higher number of type I IGF-R was detected in tumors and in primary cultures of the highly metastatic variant MM3 compared with M3. The central role of this receptor in controlling the cell cycle has been shown by gene disruption experiments (Liu et al., 1993), in which mice with a disrupted type I IGF-R gene showed underdevelopment of critical organs, including the respiratory muscles, and die within minutes of birth. Activation of type I IGF-R can lead to mitogenesis and serum independence in cultured cells (Gai et aL, 1988; Kaleko et aL, 1990), and physiological levels of type I IGF-R are an obligatory requirement for the establishment and maintenance of the transformed phenotype for several cell types, both in vitro and in intact animals (reviewed in Baserga, 1995). Furthermore, antisense strategies against type I IGF-R RNA show that a 60-70% decrease in the number of type I IGF-Rs causes a reversal of the transformed phenotype in a variety of cell types of 3 different species (reviewed in Basega, 1995). In this work we show that a decrease in the number of type I IGF-Rs is associated with a lower in vivo metastatic ability of breast cancer cells. The presence of type I IGF-R has been reported in human breast cancer specimens and in breast cancer cell lines (Arteaga and Osborne, 1989; Cullen et aZ., 1990; Peyrat and Bonneterre, 1992). Typical type I IGF-R has a roughly equal affinity for both IGF-I and IGF-I1 and mediates most of the mitogenic effects of IGF-I and -11 in breast cancer cell lines (Arteaga and Osborne, 1989; Cullen et al., 1992). The different affinities for IGF-I and IGF-I1

GUERRA ETAL. 818

1.0-

0.8 - 0.6-

0.4 - 0.2 -

- 0

\ m

m ? LL CJ H

U ln N

o I G F - l l I G F - I

b Insulin

I I I 1 1 I

0.1 1 10 100 1000

GROWTH FACTOR CONCENTRATIONlnHI 0 IGF-.II e I G F - I

a

0.

-0-0- '0.

0.2 I M3

Bound IGF-I kites I ld'/cell 1

0.2

0.1

I I I I I I

01 1 10 100 1000

b I GROWTH FACTOR CONCENTRATION (nM1

FIGURE 6 - Competitive inhibition of 1251-IGF-II binding to M3 (a) and MM3 (b) tumor membranes. Thirty micrograms of protein from M3 or MM3 microsomal membranes were incubated with 0.6 nM '25I-IGF-II and increasing concentrations of unlabeled IGF-I (a), IGF-II(0) and insulin (A). The data are plotted as B/Bo, i.e., ligand bound in the presence of competing unlabeled ligand divided by the quantity of ligand bound in the absence of competition, after substracting non-specific binding. Mean of 3 separate experiments.

Band IGF-II kites I 10-6/cell 1

TABLE 111 - IGF-R LEVELS IN M3 AND MM3 PRIMARY CULTURES]

Cell culture Type I IGF-R (sitesicell)

Man-6Pltype I1 IGF-R (sitesicell)

M3 1,431 k 80 (19 t 0.1) x 105

MM3 3,300 & 32S2 (7 2 0.1) x 1 0 5 ~ (n = 4)

(n = 4)

(n = 4)

(n = 8) 'Binding assays to M3 and MM3 cell monolayers (1 x lo6

cells/well) were performed as described in Material and Methods. The number of IGF-Rs was determined using a single-point assay at a saturating dose of 1.58 nM for IGF-I and 20 nM for IGF-11. Each value is the mean t SE of n experiments in each of which the determinations were done in triplicate.-$ < 0.01.-3p < 0.001 compared with M3 tumors by Student's t test.

exhibited by the type I IGF-R found in both M3 and MM3 membranes suggest the presence of an atypical type I IGF-R. Blockade of type I IGF-R with the monoclonal anti-receptor antibody aIR3 in human tumor xenografts grown in athymic mice inhibited tumor formation by the breast cancer cell line MDA-MB-231 (Arteaga et al., 1989). In another tumor model, Koenuma et al. (1989) found no differences in the number or affinity of type I IGF-Rs between a highly metastatic variant of the mouse colon adenocarcinoma 26 (NL-17) and a low metastatic variant (NL-44). However, when they examined the IGF-I-dependent tyrosine phosphorylation of cellular compo- nents, they found a remarkably higher degree of phosphoryla- tion of several proteins in the highly metastatic variant (Yamori et aL, 1991). Furthermore, IGF-I stimulated to a greater extent proliferation of the highly metastatic NL-17 variant (Koenuma et al., 1989). Long et al. (1995) demon- strated that transfection of the highly metastatic murine lung carcinoma H-59 cells with a plasmid vector expressing anti- sense type I IGF-R mRNA results in the loss of their metastatic ability.

U m

b Bound IGF-R (sites x 10-6/cell I

FIGURE 7 - 1251GF binding to M3 and MM3 epithelial cell cultures. (a) 'ZIGF-I binding to 1 x lo6 M3 and MM3 cells was performed as described in Material and Methods. Shown are the binding data from a representative experiment of a total of 3 plotted by the Scatchard method. Each point represents the mean of duplicate determinations. Data were analyzed using the Ligand Program (see text). (b) 1251GF-II binding to 1 x lo6 M3 and MM3 cells was performed as described in Material and Methods. Data correspond to a representative experiment of a total of 3, plotted by the method of Scatchard (1949).

TABLE lV - uPA ACTIVITY (IU/106 CELLS) SECRETED BY M3 AND MM3 TUMOR CELLS: EFFECT OF EXOGENOUS IGF-I AND 11'

M3 MM3

61.1 2 6.6 Control 10.5 t 1.3 IGF-I (2.5 ngiml) 17.9 t 4.02 92.4 2 7.g2 IGF-II(5 ndmf) 9.2 * 1.7 100.5 2 11.52

lEach value represents mean 2 SD of 4 samples.-% < 0.005, ANOVA test.

A significantly higher expression of IGF-I1 at the protein and mRNA levels was found in MM3 tumors compared with M3. RNAse protection assays performed on primary cultures revealed S-fold higher IGF-I1 mRNA levels in MM3 than in M3 epithelial cells. The IGF-I1 message has been detected in several breast cancer cell lines (Yee et al., 1988; Osborne et al.,

819 IGFS IN METASTASIZING MURINE MAMMARY CANCER

FIGURE 8 - Zymographic analysis of plasminogen activatois secreted in vitro by MM3 cells in the presence or absence of IGFs. Conditioned media were electrophoresed and gels incubated at the surface of a plasminogen-rich fibrin-agarose underlay as described in Material and Methods. Proteolytic enzymes were detected as transparent bands on the background of Amid0 black-stained underlays. Lane A: Mouse urine as a source of murine uPA. Lane B: MM3 conditioned medium obtained in the absence of IGFs. Lane C: MM3 conditioned medium obtained in the presence of 2.5 ng/ml of IGF-I. Lane D: In the presence of 5 ng/ml of IGF-11. Lane E: MM3 conditioned medium preincubated with 10 kg/ml anti-catalytic rabbit polyclonal antibody to murine uPA.

1989), in a high number of breast tumors specimens (Yee et aL, 1988), and in the T61 human breast cancer xenograft (Brunner et al., 1992). In in situ hybridization studies on human breast cancer tissues, IGF-I1 mRNA was found in both the malignant epithelial cells and their adjacent stroma (Paik, 1992). Further- more, antisense RNA to IGF-I1 inhibits tumorigenesis in vivo (Christofori et al., 1994).

The levels of MandPitype I1 IGF receptor were lower in the IGF-I1 overproducing MM3 tumor than in M3. Competitive inhibition and affinity labeling studies showed an almost exclusive binding of IGF-I1 to Man-6Phype I1 IGF receptor on M3 and MM3 tumor membranes. The MandP/type I1 IGF receptor is a multifunctional receptor with high affinity on distinct sites for 2 classes of ligands: IGF-I1 and Man-6P- bearing molecules such as lysosomal enzymes (Mathieu et al., 1990). The presence of Man-6P/type I1 IGF receptor in breast cancer cell lines and tumor specimens is well known (Osborne et al., 1989; Cullen et al., 1990; Mathieu et al., 1990, 1991). A correlation between high IGF-I1 content and low MandP/ type I1 IGF receptor concentration has been observed in breast cancer cells (Mathieu et al., 1991). Although the function of this type of receptor remains to be elucidated, it may play a role in mitogenesis (Hari et al., 1987; Tally et aL, 1987; Mathieu et al., 1990). Particularly in breast cancer the mitogenic effects of IGF-I1 in MCF-7 cells in culture and growing as xenografts in athymic mice, and in the human breast cancer xenograft T61D, cannot be completely blocked by aIR3, suggesting that the Man-6P/type I1 IGF-R may be involved in IGF-11-mediated growth (Arteaga et al., 1989; Mathieu et al., 1990; Brunner et al., 1993). Gene disruption experiments suggest the existence of another yet unidentified receptor for IGF-I1 that could be involved in IGF-11-mediated growth (Baker et al., 1993). A role as a potential autocrine/ paracrine regulator of breast cancer cell growth has been postulated for IGF-11. However, as mentioned for IGF-I, no information is available about IGF-I1 involvement in breast cancer metastasis.

Our findings on IGF-I and type I IGF-R expression in M3 and MM3 tumors and primary cultures allows us to hypoth-

esize that the presence of a higher number of type I receptors in the highly metastatic MM3 tumor could favor the response of MM3 cells to endogenously produced IGF-I. Similarly, the higher production of IGF-I1 by MM3 tumor cells could confer an advantage in their autonomous growth. A correlation between autonomous growth and metastatic ability of breast cancer cells was found in both cell cultures and in a mouse model (Kerbel, 1990). In this study, we were unable to detect any mitogenic effect of IGF-I or IGF-I1 on M3 and MM3 cells. However, this was not an unexpected finding taking into account the constitutive production of IGF-I and 11 by M3 and MM3 cells. The endogenously produced IGFs could either occupy the surface receptors or induce an intracellular mito- genic signal (Korc-Grodzicki et al., 1992). Another explanation could be the presence in M3 and MM3 cultures of IGF-BPs, which could bind the exogenous IGFs, inhibiting their mito- genic effects (Koscielniak et al., 1994). The failure of M3 and MM3 cells to respond to IGF-I may be related to the low level of type I IGF-R expression. We are aware that studies using anti-IGFs and anti-IGFs-R are required for the assess- ment of the probable effects of IGF-I and I1 in M3 and MM3 cells.

IGFs could play roles other than modulating M3 and MM3 tumor growth. We show here that IGF-I stimulates M3 and MM3 cells secretion of uPA, a key enzyme that initiates an extracellular proteolytic cascade that facilitates the motility of invasive cells. However, IGF-I1 only stimulated uPA synthesis in MM3 cells, leading to the possibility that selective produc- tion of uPA by MM3 cells in response to IGF-I1 could be one of the factors associated with MM3 increased metastatic capacity. Several mutual interactions among growth factors, pericellular proteins and proteases should be considered (Laiho and Keski-Oja, 1989). Interactions between the IGF system and proteolytic activities in M3 and MM3 tumors could also involve uPA modulation of the bioavailability of IGFs in the tumor microenvironment through the proteolytic degrada- tion of IGF-BPs. Thus, stimulation of M3 and MM3 cells growth and invasiveness could be mediated by the proteolytic activation of GFs, which in turn have the ability to up-regulate the expression of proteolytic enzymes. IGF-I1 binding to type I1 IGF/Man-6-P receptor in breast cancer membranes inhibits the binding and cellular uptake of lysosomal enzymes (Kiess et al., 1989). This is particularly important in the process of metastasis, since it would result in a higher extracellular concentration of acid hydrolases and a consequent more effective digestion of the extracellular matrix favoring tumor spread. The increased production of IGF-I1 by the highly metastatic MM3 tumor cells and the preferential binding of this factor to type I1 IGFs receptor might strongly alter the targeting and routing of Iysosomal enzymes. This would result in an oversecretion of acid hydrolases associated with MM3 tumor increased metastasizing ability.

In summary, our results show a varying pattern of expression of IGFs and their receptors as well as different IGF regulation of uPA secretion in M3 and MM3 murine mammary adenocar- cinomas of different metastatic ability.

ACKNOWLEDGMENTS

The authors thank Dr. A.A. Molinolo for critical discussion and review of the manuscript, Dr. J.C. Calvo for discussion of the binding data and Drs. R. Lupu, K.J. Cullen and M. Saceda (Washington, DC) for advice with the IGF RNAse protection assays.

REFERENCES

ALONSO, D.F., FARIAS, E.F. and BAL DE KIER JOFFE, E., Impairment of fibrinolysis during the growth of two murine mammary adenocarci- nomas. CancerLett., 70,181-187 (1993). ARTEAGA, C.L., KITTEN, L.J., CORONADO, E.B., JACOBS, S., KULL, F.C. JR., ALLRED, D.C. and OSBORNE, C.K., Blockade of the type I

somatomedin receptor inhibits growth of human breast cancer cells in athymic mice.]. clin. Invest., 84,1418-1423 (1989). ARTEAGA, C.L. and OSBORNE, C.K., Growth inhibition of human breast cancer cells in vitro with an antibody a ainst the type I somatomedin receptor. CuncerRes., 49,6237-6241 ($989).

GUERRA ETAL 820

BAL DE KIER JOFFE, E., PURICELLI, L. and LUSTIG, E.S., Modified adhesion behaviour after in vitro gassage of two related murine mammary adenocarcinomas with di erent metastasizing ability. Inva- sion Metastasis, 6,302-312 (1986). BAKER, J., LIU, J.P., ROBERTSON, E.J. and EFSTRATIADIS, A., Role of insulin-like rowth factors in embryonic and postnatal growth. Cell, 75, 73-82 (1993f. BARENTON, B., PATE, B., KHAN, M.N., GUYDA, H.J. and POSNER, B.I., Insulin-like growth factor-binding proteins in hypophysectomized rat liver: characterization and subcellular localization. Endocrinology, 122, 2499-2507 (1988). BASERGA, R., The insulin-like growth factor I receptor: a key to tumor growth? Cancer Res., 55,249-252 (1995). BRUNNER, N., MOSER, C., CLARKE, R. and CULLEN, K.J., IGF-I and IGF-I1 expression in human breast cancer xenografts: relationship to hormone independence. Breasr Cancer Res. Treat., 22,3945 (1992). BRUNNER, N., YEE, D., KERN, F.G., SPANG-THOMPSEN, M., LIPPMAN, M.E. and CULLEN, K.J., Effect of endocrine therapy on growth of T61 human breast cancer xenografts is directly correlated to specific down-regulation of insulin-like growth factor 11. Europ. J. Cancer, 29A, 562-569 (1993). CHIRGWIN, J., PRYZBYLA, A., MCDONALD, R. and RUTTER, W., Isolation of biologically active ribonucleic acid from sources enriched for ribonuclease. Biochemistry, 18,5294-5298 (1979). CHRISTOFORI, G., NAIK, P. and HANAHAN, D., A second signal supplied by insulin-like growth factor I1 in oncogene-induced tumori- genesis. Nature (Lond.), 369, 414-417 (1994). CULLEN, K.J., LIPPMAN, M.E., CHOW, D., HILL, S., ROSEN, N. and ZWIEBEL, J.A., Insulin-like growth factor I1 overexpression in MCF-7 cells induces phenotypic changes associated with malignant progres- sion. Mol. Endocrinol., 6,91-100 (1992). CULLEN, K.J., SMITH, H.S., HILL, S., ROSEN, N. and LIPPMAN, M.E., Growth factor messenger RNA expression by human breast fibroblasts from benign and malignant lesions. Cancer Res., 51,4978-4985 (1991). CULLEN, K.J., YEE, D., SLY, W.S., PERDUE, J., HAMPTON, B., LIPPMAN, M.E. and ROSEN, N., Insulin-like growth factor receptor expression and function in human breast cancer. Cancer Res., 50,48-53 (1990). EL-BADRY, O.M., MINNITI, C., KOHN, E.C., HOUGHTON, P.J., DAUGHADAY, W.H. and HELMAN, L.J., Insulin-like growth factor I1 acts as an autocrine growth and motility factor in human rhabdomyo- sarcoma tumors. Cell Growth Diff, 1,325-331 (1990). FURLANETTO, R.W., DI CARLO, S.N. and WISETTAR, T.C., The type I1 insulin-like growth factor receptor does not mediate deoxyribonucleic acid synthesis in human fibroblasts. J. clin. Endocrinol., 64, 1142-1149 (1987). GAI, X., RIZZO, M.G., LEE, J., ULLRICH, A. and BASERGA, R., Abrogation of the requirements for added growth factors in 3T3 cells constitutively expressing the p53 and IGF-I genes. Oncogene Res., 3,

GRANELLI-PIPERNO, A. and REICH, E., A study of protease and proteinase inhibitor complexes in biological fluids. J. exp. Med., 146,

HARI, J., PIERCE, S.B., MORGAN, D.O., SARA, V., SMITH, M.C. and ROTH, R.A., The receptor for insulin-like growth factor I1 mediates an insulin-like response. EMBOJ., 6,3367-3371 (1987). Hsu, C.J. and HAMMOND, J.M., Gonadotropins and estradiol stimulate immunoreactive insulin-like growth factor-I production by porcine granulosa cells in vitro. Endocrinology, 120,198-207 (1987). KALEKO, M., RUTTER, W.J. and MILLER, A.D., Overexpression of the human insulin-like growth factor I receptor promotes ligand- dependent neoplastic transformation. Mol. cell. Biol., 10,464-473 (1990). KERBEL, R.S., Growth dominance of the metastatic cancer cekcellular and molecular aspects.Adv. Cancer Res., 55,87-132 (1990). KIESS, W., THOMAS, C.L., GREENSTEIN, L.A., LEE, L., SKLAR, M.M., RECHLER, M.M., SAHAGIAN, G.G. and NISSLEY, S.P., Insulin-like growth factor I1 (IGF-11) inhibits both the cellular uptake of p-galacto- sidase and the binding of p- alactosidase to purified IGF-II/mannose 6-phosphate receptor. J. biof Chem., 264,4710-4714 (1989). KOENUMA, M., YAMORI, T. and TSURUO, T., Insulin and insulin-like growth factor I stimulate proliferation of metastatic variants of colon carcinoma 26.Jpn. J. Cancer Rex, 80,51-58 (1989). KORC-GRODZICKI, B., FURLANETTO, R.W. and HILF, R., Characteriza- tion of the insulin-like growth factor receptors in the insulin- responsive R3230 sc mammary adenocarcinoma. Oncol. Res., 4, 109- 119 (1992).

377-386 (1988).

223-234 (1978).

KOSCIELNIAK, E.B., GUTBROD, R., BLUM, W., HENNE, T. and TREUNER, J., Varying patterns of secretion of insulin like growth factors (IGFs) and their binding proteins (BPs) by sarcoma cell lines. Proc. Amer. Assoc. CuncerRes., 35,44 (1994). LABARCA, C. and PAIYEN, K., A simple, ra id and sensitive DNA assay procedure. Anal. Biochem., 102,344-352 8980). LAEMMLI, U.K., Cleavage of structural proteins during the assembly of the head of bacteriophage T4. Nature (Lond.), 227,68&685 (1970). LAIHO, M. and KESKI-OJA, J., Growth factors in the regulation of pericellular proteolysis: a review. Cancer Rex, 49,2533-2553 (1989). LIU, J.P., BAKER, J., PERKINS, A.S., ROBERTSON, E.J. and EFSTRATIA- DIS, A., Mice carrying null mutations of the genes encoding insulin-like growth factor I (Igf-1) and type I IGF receptor (Igflr). Cell, 75,59-72 (1993). LONG, L., RUBIN, R., BASERGA, R. and BRODT, P., Loss of the metastatic phenotype in murine carcinoma cells expressing an anti- sense RNA to the insulin-like growth factor receptor. Cancer Res., 55, 1006-1009 (1995). MATHIEU, M., ROCHEFORT, H., BARENTON, B., PREBOIS, C. and VIGNON, F., Interactions of cathepsin D and IGF-I1 on the IGF-II/ M6P receptor in human breast cancer cells and ossible consequences on mitogenic activity of IGF-11. Mol. Endocrinolfj 4,1327-1335 (1990). MATHIEU, M., VIGNON, F., CAPONY, F. and ROCHEFORT, H., Estradiol down-regulates the mannose-6-phosphate/IGF-II receptor gene and induces cathepsin D in breast cancer cells: a receptor saturation mechanism to increase the secretion of lysosomal proenzymes. Mol. Endocrinol., 5,815-822 (1991). MOLINOLO, A.A., LANARI, C., CHARREAU, E.H., SAN JUAN, N. and DOSNE PASQUALINI, C., Mouse mammary tumors induced by medroxy- progesterone acetate: immunohistochemistry and hormonal receptors. J. nut. Cancer Inst., 79,1341-1350 (1987). MOSES, A.C., NISSLEY, S.P., PASSAMANI, J. and WHITE, R.M., Further characterization of growth hormone-dependent somatomedin-binding proteins roduced by rat liver cells in culture. Endocrinology, 104, 536-546 8979). MUNSON, P.J. and RODBARD, D., Ligand: a versatile computerized a proach for characterization of ligand-binding systems. Anal. Bio- ciem., 107,220-239 (1980). NICOLSON, G.L., Cancer progression and growth: relationship of paracrine and autocrine growth mechanisms to organ preference of metastasis. Exp. Cell Res., 204,171-180 (1993). OSBORNE, C.K., CORONADO, E.B., KITTEN, L.J., ARTEAGA, C.L., FUQUA, S.A., RAMASHARMA, K., MARSHALL, M. and LI, C.H., Insulin like growth factor I1 (IGF-11): a potential autocrineiparacrine growth factor for human breast cancer acting via the IGF-I receptor. Mol. Endocrinol., 3,1701-1709 (1989). PAIK, S., Expression of IGF-I and I1 mRNA in breast tissue. Breast CancerRes. Treat., 22,31-38 (1992). PEYRAT, J.P. and BONNETERRE, J., Type 1 IGF receptor in human breast diseases. Breast Cancer Res. Treat., 22,5947 (1992). SCATCHARD, G., The attraction of proteins for small molecules and ions. Ann. N.Y. Acad. Sci., 51,660-672 (1949). STRACKE, M.L., ENGEL, J.D., WILSON, L.W., RECHLER, M.M., LIOITA, L.A. and SCHIFFMANN, E., The type I insulin-like growth factor receptor is a motility receptor in human melanoma cells. J. biol. Chem., 264,21544-21549 (1989). TALLY, M., LI, C.H. and HALL, K., IGF-2 stimulated growth mediated by the somatomedin type 2 receptor. Biochem. biophys. Res. Comm.,

TOBIN, G., YEE, D., BRUNNER, N. and ROTWEIN, P., A novel insulin- like growth factor I messenger RNA is expressed in normal and tumor cells. Mol. Endocrinol., 4,1914-1920 (1990). YAMORI, T., IIZUKA, Y., TAKAYAMA, Y., NISHIYA, S., IWASHITA, S., YAMAZAKI, A., TAKATORI, T. and TSURUO, T., Insulin-like growth factor I rapidly induces tyrosine phosphorylation of Mr 150,000 and a Mr 160,000 protein in higkly metastatic mouse colon carcinoma 26 NL-17 cells. Cancer Res., 51,5859-5865 (1991). YEE, D., The insulin-like growth factor system as a target in breast cancer. Breast CancerRes. Treat., 32,85-95 (1994). YEE, D., CULLEN, K.J., PAIK, S., PERDUE, J.F., LIPPMAN, M.E. and ROSEN, N., Insulin-like growth factor I1 expression in human breast cancer. Cancer Res., 48,6691-6696 (1988). YOUNG, J.A.T. and TROWSDALE, J., A processed pseudogene in an intron of the HLA-DPpl chain gene is a member of the ribosomal protein L32 gene family. Nucleic Acids Res., 13,8883-8891 (1985).

148,811-816 (1987).