Molecular and Cellular Biochemistry 267: 83–89, 2004.c© 2004 Kluwer Academic Publishers. Printed in the Netherlands.

The effects of insulin-like growth factor bindingprotein-3 (IGFBP-3) on T47D breast cancer cellsenriched for IGFBP-3 binding sites

Suresh Mishra and Liam J. MurphyDepartments of Internal Medicine & Physiology, University of Manitoba, Winnipeg R3E 0W3 Canada

22 January 2004; accepted 26 May 2004

Abstract

To investigate insulin-like growth factor (IGF)-independent effects of IGF binding protein-3 (IGFBP-3), T47D cells wereenriched for a population of cells that expressed binding sites for biotinylated-IGFBP-3 by panning on streptavidin-coatedplate. Proliferation of cell enriched for IGFBP-3 binding sites was significantly inhibited by IGFBP-3, whereas IGFBP-3 hadno significant effect on the non-enriched cell population. Enriched and non-enriched cells were equally responsive to IGF-I,TGF-β and EGF. Conditioned medium from enriched cells had less IGFBP-3 than that from non-enriched cells. Cross-linkingof biotinylated IGFBP-3 to T47D cell membranes identified complexes with Mr of 32, 80 and 100 kDa. All of these complexeswere more abundant in enriched cells compared with the non-enriched cell population. These data demonstrate that despite theanti-proliferative effects of IGFBP-3 it is possible to selectively enriched for cell populations with more abundant IGFBP-3binding sites. These enriched cells are more responsive to IGFBP-3 and secrete less of this binding protein than non-enrichedcells, supporting the concept that IGFBP-3 secretion by human breast cancer cells may function as an autocrine or paracrinemodulator of cell proliferation. (Mol Cell Biochem 267: 83–89, 2004)

Key words: IGFBP-3, IGF-independent effects, growth inhibition, apoptosis

Introduction

Insulin-like growth factor binding protein-3 (IGFBP-3), themost abundant of the IGFBPs in the circulation, can bothpositively and negatively modulate the actions of insulin-likegrowth factors (IGF-I and -II) [1]. In addition, IGFBP-3 alsofunctions in an IGF-independent manner to stimulate apop-tosis and inhibit cellular proliferation [2–4] of variety of celllines including human breast cancer cells. Mutant IGFBP-3molecules and IGFBP-3 fragments which have minimal affin-ity for IGF-I [4–6] have pro-apoptotic and anti-proliferativeeffects, which are independent of the effects of IGF-I and-II and are demonstrable even in cell lines devoid of IGF-Ireceptors [2, 7].

∗Address for offprints: Liam J. Murphy, Room 843, John Buhler Research Centre, University of Manitoba 715 McDermot Avenue, Winnipeg MB R3E, 3P4Canada, (Email: ljmurph@cc.umanitoba.ca)

While IGF-I has been reported to inhibit the bindingof IGFBP-3 to cell membranes [8, 9], mutant IGFBP-3molecules that lack the ability to bind IGF-I, retain theirability to bind plasma membranes [5]. In a similar fashionIGFBP-3 is able to inhibit binding of IGF-I to the type 1 IGFreceptor and modulated IGF-I receptor signal transduction[10]. Simply stated, IGFBP-3 and IGF-I have opposing ef-fects on cell proliferation and survival and these actions areinhibited by the interaction with each other.

The IGF-independent effects of IGFBP-3 are thought tobe mediated by specific membrane binding sites [9] and mayalso involve transport to the nucleus and the interaction ofIGFBP-3 with nuclear proteins [11]. Both the C-terminal endof IGFBP-3 and the central domain appears to be involved

84

in binding of IGFBP-3 to plasma membrane binding sites[12, 13]. A variety of membrane proteins which interactwith IGFBP-3 have been demonstrated using cross-linkingtechniques, however, as yet the identity of most of thesemembrane proteins have not been established and it is un-clear which, if any, of these binding partners are important inthe IGF-independent effects of IGFBP-3.

IGFBP-3 is expressed and secreted by a wide varietyof cell lines including those that appear to be sensitive tothe IGF-independent growth inhibitory effects of IGFBP-3. While the functional importance of the growth inhibitoryIGF-independent effects of IGFBP-3 are unclear, the secre-tion by, and responsiveness to IGFBP-3 by the same cellpopulation suggests that IGFBP-3 may function as an au-tocrine or paracrine growth modulator. It is now appreciatedthat individual cells within an apparently homogeneous cellpopulation demonstrated different levels of expression of var-ious genes. Phenotypic drift, a phenomenon where the levelsof expression of various genes in individuals cells within acell population can vary widely, has been well described incultured breast cancer cell lines where it is possible to cloneout cell populations with different levels of expression ofestrogen receptor and estrogen responsive genes [14]. Wehypothesized that a similar situation may pertain to IGFBP-3expression in cultured breast cancer cells.

To examine the potential role of IGFBP-3 as a growthmodulator, and to provide some insight into which mem-brane binding sites are important in the growth inhibitor, pro-apoptotic effects of IGFBP-3, we developed a panning tech-nique to enrich cell populations for IGFBP-3 binding sites.Here we report the difference in IGFBP-3 responsiveness inT47D cells enriched for IGFBP-3 binding sites.

Methods

Materials and reagents

T47D cells were obtained from the American Type Tis-sue Culture (Manassas, Virginia). Cell culture reagents werefrom Gibco Life Sciences (Burlington ON). Glycosylatedand non-glycosylated IGFBP-3 were obtained from UpstateBiotechnology Inc., (Lake Placid, NY). All other reagentswere obtained from Sigma-Aldrich Canada (Oakville On.)unless otherwise stated.

Biotinylation of IGFBP-3

Non-glycosylated E. coli derived IGFBP-3 was biotinylatedusing the p-biotinoyl-aminocaproic acid-N -hydroxy- succi-namide ester (Roche Molecular Biochemicals, Mannheim,Germany) at room temperature as previously described [15].At the end of incubation free biotin ester were separated ona Sephadex G-25 column.

Cell sorting

Polystyrene bacteriological petri dishes (100 × 15 mm) wereused for cell sorting to reduce the non-specific cell adher-ence observed with tissue culture plasticware. Streptavidin(4 µg/ml) were diluted in 0.05M Tris, pH 9.5 and 10 ml waspoured onto the plates. After 1 h at room temperature, thebuffer was decanted, and the dishes were washed three timeswith PBS. Then 3 ml of blocking buffer (1 mg/ml BSA inPBS) was added to each dish and incubated for 90 min at roomtemperature. At the end of incubation the blocking buffer wasaspirated and plates were stored at −20 ◦C until used.

T47D cells, 2-3 ×107 were suspended in Dulbecco’s Mod-ified Eagle’s Medium (DMEM) with 10% fetal bovine serum(FBS). Biotinylated IGFBP-3 (400 ng/ml) were added to thecell suspension and incubated for 1 h on ice. After the incu-bation, cells were centrifuged to remove excess IGFBP-3 andthen re-suspended in fresh DMEM with 10% FBS and pouredonto streptavidin-coated dishes. The plates were incubated atroom temperature for 1h. After 40 min the non-adherent cellswere removed again by swirling the dish and decanting thesupernatant. Subsequently, adherent cells were washed threetimes with 5–10 ml of DMEM with 10% FBS. To recoverthe bound cells, the plates were filled with 25 ml DMEMwith 10% FBS and entire surface of the plate was flushed byusing a pasteur pipette. The recovered cells were washed andtransferred to tissue culture grade plastic dishes.

Receptor binding assay

Sorted and unsorted cells (0.6 × 105 cells/ml) were seededin 24 well culture plates and allowed to grow to near conflu-ence. The media was removed and cells were washed threetimes in PBS and incubated in the presence and absence of bi-otinylated IGFBP-3 (200 ng/ml) in PBS for 1 h at 4 ◦C. Afterincubation, unbound IGFBP-3 was removed by washing and125I-streptavdin (50,000 cpm) was added to each well and theincubation was continued for 90 min at 20 ◦C. After washingto remove unbound 125I-streptavdin, cells were solubilized in0.5M NaOH containing 0.5% Triton×100 and the radioactiv-ity was counted in a Wallac 1470 automatic gamma counter.

Cross-linking of biotinylated IGFBP-3

Solubilized cell membranes were obtained from confluentcultures using membrane preparation kit and protocol ac-cording to the manufacturer’s instructions (Pierce, Rockford.Il). Cross-linking was performed essentially as previouslydescribed [15]. 2 µg of biotinylated-IGFBP-3 was added to200 µl of solubilized cell membrane and incubated at 4 ◦C for2 h. At the end of incubation, disucinimidyl suberate (DSS,100 µg/ml) was added and incubated for another 15–20 min.Finally the IGFBP-3 cross-linked membrane proteins were

85

precipitated with streptavidin-agarose, boiled in SDS sam-ple buffer and analyzed on an 11% SDS-PAGE gel. Proteinswere transferred to the nitrocellulose membrane. Membraneswere blocked in 5% milk washed in TBST (0.5% T-20) andincubated with streptavidin-horseradish peroxidase conju-gate diluted 1:3000 for 1 h at room temperature. Membraneswere washed in TBST and analyzed using enhanced chemi-luminescence detection system (ECL, Amersham Pharma-cia Biotech, Baie d’Urfe, Quebec) and exposed to KodakBioMax film (Eastman Kodak Co., Rochester, NY).

For whole cell cross-linking studies, cells were grown to90% confluence on 35 mm culture dishes. Cells were washedthree times with ice cold PBS and incubated with biotinylatedIGFBP-3 (200 ng/ml) in PBS for 1 h at 4 ◦C. At the end ofincubation, DSS (100 µg/ml) was added and the incubationcontinued for another 15–20 min. Cells were washed threetimes with ice cold PBS and recovered with a cell scraper inice cold PBS containing aprotinin (2 µg/ml), 1 mM PMSFand 0.1 mM EDTA. IGFBP-3 cross-linked proteins were an-alyzed by SDS-PAGE and detected either using streptavidin-HRP as described above or using anti-IGFBP-3 antibody.Antibody AF675 directed against the whole molecule wasobtained from R&D Systems (Minneapolis MN).

Western blotting of conditioned medium and cell pellets

Sorted and unsorted cells were allowed to grow to near con-fluence in DMEM with 10% FBS. Before collection of con-

Fig. 1. Enrichment of T47D cells for IGFBP-3 binding. In panel A, sorted cells were enriched by incubating cells with biotinylated IGFBP-3 and panningon streptavidin-coated plates. Adherent cells were transferred to regular tissue culture plates and grown to confluence. These cultures were incubated withbiotinylated IGFBP-3 and after washing with radiolabeled streptavidin. In panel B, enriched cells were subjected to a second and third round of panning. Ineach case the data represent the mean ± SEM for four experiments. In panel B the 100% is equivalent to 599.7 cpm.

ditioned medium (CM), cells were washed with PBS andserum free media added. After 6 h media were replaced againwith fresh serum free media and collected 48 h later. Subse-quently, media were centrifuged at 3500 rpm for 15 min toremove cellular debris, concentrated five times (5×) withCentricon 10 (Millipore Corporation, MA, USA) and storedat −20 ◦C until used. CM was analyzed by Western blottingto determine the presence of IGFBP-3. 30 µg of CM pro-tein per well was loaded in gel loading buffer (62.5 mM Tris,2% SDS, 0.02% bromophenol blue, 10% glycerol, pH 6.8)and electrophoresed through 12% separating gel. Cell pel-lets were analyzed in a similar fashion. Separated proteinswere electrophoretically transferred to a nitrocellulose mem-brane in transfer buffer (25 mM Tris, 192 mM glycine in 15%methanol). After transfer, membranes were blocked in 5%milk, washed in TBST (0.5% T-20), and incubated with bi-otinylated anti-hIGFBP-3 rabbit polyclonal antiserum (1:200dilution, Diagnostic System Laboratories Inc.) for 1 h at roomtemperature. After washing in TBST, membranes were incu-bated with streptavidin-horseradish peroxidase (HRP) conju-gate diluted 1:3000 (Life Technologies, Inc.) for 1 h at roomtemperature. Membranes were washed in TBST and analyzedusing ECL as described above.

MTT assay

The effects of growth factors were enumerated by colori-metric 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium

86

bromide (MTT) assay and by hemocytometer as well. Cellswere seeded at a plating density of 3 × 103/well and culturedfor 24 h to allow them to adhere to the plate. The culturemedium was then changed to serum-free medium (phenol-red free DMEM containing 0.1% BSA) supplemented withgrowth factors and the cells were allowed to grow for sixdays. Fresh medium was added every 48 h. The concentra-tion of growth factors used were; TGF-β 1ng/ml, EGF 10ng/ml, IGF-I 100 ng/ml and IGFBP-3 100 ng/ml). On day sixMTT (10 µg/100 µl) was added and incubation continuedfor 3 h. At the end of incubation the tetrazolium productswere solubilized in acidic isopropanol and OD was read at570 nm. Data are presented as percentage of controls.

Growth curves

Cells were seeded in 24 well culture plates, 104 cells/well,and cultured in DMEM with 10% FBS for 24 h to allow themto adhere to the plate. After 24 h, media were replaced withserum-free media supplemented with IGF-I or IGFBP-3 (100ng/ml). Cells were trypsinized on day 2, 4 and 6 and countedby hemocytometer.

Results

Enrichment of IGFBP-3 responsive cells by panning

The panning technique was used to enrich populations ofT47D cells that expressed binding sites for IGFBP-3. Cellswere incubated with biotinylated IGFBP-3 and subsequentlyallowed to attach to streptavidin-coated plates. After sub-culturing, binding of IGFBP-3 to the monolayers was deter-mined. A representative experiment is shown in Fig. 1a. Asingle round of panning resulted in an approximately four-fold increase in IGFBP-3 binding capacity. A further smallincrease in IGFBP-3 binding to the monolayer was observedwhen panned cells were grown and then subjected to a sec-ond and third round of panning (Fig. 1b). For all subsequentexperiments at least two rounds of panning was used.

Biotinylated IGFBP-3 was incubated with sorted and un-sorted cells and subsequently cross-linked using DSS. Aspreviously described three major bands were identified [15].These had a Mr of ∼32, ∼80 and ∼100 kDa with the 100 kDaspecies being the least abundant and the 32 kDa species be-ing the most abundant (Fig. 2). All three major bands wereapparent in both unsorted and sorted cells and in the pres-ence or absence of DSS. There was no selective enrichmentof any one the IGFBP-3 binding complexes in the pannedcells. The intensity of all major bands was greater in sortedcells and was increased further by cross-linking with DSSin both unsorted and sorted cells. Interestingly, the 32 kDaband, which was similar in size to the IGFBP-3 standard, was

Fig. 2. Cross-linking with DSS of biotinylated IGFBP-3 to T47D cell mono-layers. Sorted and unsorted cell monolayers were incubated with biotinylatedIGFBP-3 and cross-linked as described in the Material and methods section.The upper panel shows a representative gel. The position of the 100, 80 and32 kDa complexes are indicated by the arrowheads. In the lower panel, theintensity of the 32 and 80 kDa in sort and unsorted cells, in the absence andpresence of cross-linking have been quantified by densitometry. The datarepresent the mean ± SEM for 3 separate experiments.

also increased by cross-linking. Immunoblotting with severaldifferent anti-IGFBP-3 antisera indicated that all three ma-jor bands identified by cross-linking biotinylated-IGFBP-3to T47D cell membranes not only contained biotin but alsocontained immunoreactive IGFBP-3 (data not shown).

Intact IGFBP-3 was detected by Western blot analysis ofconditioned media from unsorted and sorted T47D cells.There was consistently more IGFBP-3 present in media from

87

Fig. 3. Immunoblot of conditioned medium and cell pellet from unsorted and sorted T47D cells cultures. Conditioned medium and the cell pellets from sortedand non-sorted cells were analyzed. The molecular mass of the bands were determined by comparison with molecular weight markers. The experiment wasreplicated on three separate occasions with similar results.

Fig. 4. Growth curves for unsorted and sorted cells in serum-free medium together supplement with IGF-I or IGFBP-3 at a final concentration of 100 ng/ml.The statistical difference between the growth curves for treated cultures compared to control cultures is indicated.

unsorted cells than sorted cells (Fig. 3). In contrast, sortedcells had less cell pellet associated IGFBP-3 than unsortedcells.

We compared the effect of IGF-I and IGFBP-3 on cellgrowth of unsorted cells and sorted cells enriched for IGFBP-3 binding sites. IGFBP-3 had no significant effects on cellgrowth in cultures that had not been enriched by panning. Incontrast, in sorted T47D cells, IGFBP-3 reduced cell num-

ber by ∼20% on day 6, p < 0.05 (Fig. 4). In unsorted andsorted cells, IGF-I caused a small increase in cell number.There was no significant difference in the magnitude of re-sponse of sorted and unsorted cells to IGF-I suggesting thatthe anti-proliferative effect of IGFBP-3 on enriched cells wasnot due to enhanced IGF-I sensitivity in this cell populationand was likely to be due to an IGF-independent effect ofIGFBP-3.

88

Fig. 5. The effects of IGFBP-3, TGF-β and EGF on cell proliferation as measured in the MTT assay. The closed histograms show data for sorted cells whilethe open histograms show data for the unsorted cells. The data represent the mean ±SEM expressed as a percentage of the control wells, for three or moreseparate experiments. The statistically significant differences between the treated and untreated cells are shown. ∗ and ∗ ∗ ∗ indicate p < 0.05 and p < 0.005respectively.

The effects of IGF-I and IGFBP-3 on T47D cell prolifer-ation were confirmed using the MTT assay. In sorted cells,but not unsorted T47D cells, IGFBP-3 resulted in a signifi-cant reduction in cell proliferation, 70.9 ± 7.3% of control,(Fig. 5). Sorted and non-sorted cells responded similarly toEGF and TGF-β.

Discussion

Since the panning technique required not only adhesion toIGFBP-3, but also subsequent proliferation of the cells, weinitially hypothesized that the enriched cell may be devoidof the particular membrane receptors that mediate the anti-proliferative, pro-apoptotic effects of IGFBP-3 and would beunresponsive to IGFBP-3. Thus by comparison of IGFBP-3membrane binding sites on unsorted and enriched T47D cells,it would be possible to infer which membrane binding siteswere responsible for the IGFBP-3 growth inhibitory effects.In contrast, we demonstrated using the panning techniquethat it is possible to enrich T47D cells which not only bindingmore IGFBP-3 but, surprisingly are also more responsive tothis binding protein in terms of growth inhibition. Although

the enriched cells express more IGFBP-3 binding capacityand their proliferation is significantly inhibited by IGFBP-3,the short duration of exposure to IGFBP-3 during the panningprocedure does not result in sufficient cell growth inhibitionto preclude successful enrichment of the cells.

We were able to demonstrate by cross-linking that IGFBP-3 bound to multiple proteins on the T47D cell membrane.This is consistent with other reports using different cell lineswhere multiple IGFBP-3 binding sites were also identified[9, 16] and our recent report using T47D cells [15]. Someof these binding sites were of approximately similar sizeto that reported by Oh and colleagues who also used non-glycosylated E. Coli-derived IGFBP-3 [16]. They reportedspecies with apparent size of 20, 26 and 50 kDa after sub-traction of the mass of the IGFBP-3. With the additional massof the cross-linked IGFBP-3, the larger band could representthe 80 kDa bands that we observed. The most abundant bi-otinylated species we observed was the 32 kDa band. Sincethis band was enhanced by cross-linking, we speculate that itrepresents IGFBP-3 which had been tightly associated witha membrane protein and was subsequently dissociated ei-ther by proteolytic activity or by the process of boiling underthe reducing conditions prior to analysis on SDS-PAGE. The

89

relationship between the 100, 80 and 32 kDa band is unclear.However, all the cross-linked bands were equally enhancedby cell sorting suggesting that these three species were coor-dinately regulated and/or interrelated.

Our data does not provide any insight into which of thecross-linked bands is likely to be important in the IGF-independent effects of IGFBP-3, however, since the anti-proliferative effect of IGFBP-3 were more marked in theenriched cells these IGFBP-3 binding sites are likely to me-diate the antiproliferative effects. Furthermore the antiprolif-erative effect of IGFBP-3 demonstrated here is most likelyIGF-independent effects rather than IGF-dependent effects.In support of this argument is the observation that enrichedcells and non-enriched cells were equally responsive to IGF-I.

This breast cancer cell line is known to express IGFBP-3and cells enriched for IGFBP-3 binding sites secreted lessIGFBP-3 into the medium than the unsorted cells. In ad-dition there was less cell pellet associated IGFBP-3 in theenriched cells despite the enhanced abundance of IGFBP-3 binding sites. The inverse relationship between secretionof IGFBP-3 and abundance of membrane binding sites andIGFBP-3 sensitivity would be consistent with lower levels ofsecreted IGFBP-3 having similar paracrine/autocrine growthinhibitory functional activity in these enriched cells as in theunsorted cells.

In the unsorted cell population, there are cells that aremore sensitive to the growth inhibitory effects of IGFBP-3 by virtue of the increased abundance of IGFBP-3 bindingsites. These cells are not lost from the cell population despitethe presence of IGFBP-3 in the conditioned medium possiblybecause of reduced endogenous expression of IGFBP-3 bythe cells that express the highest levels of IGFBP-3 bindingsites and lower IGFBP-3 in the peri-cellular environment.

In summary, we have demonstrated that it is possible toutilize the panning technique to generate an enriched cellpopulation, which are growth inhibited by the agent usedfor selection. To our knowledge the use of this technique inthis manner has not been previously reported. The abilityto identify cells enriched for IGFBP-3 may provide a use-ful tool to further examine the IGF-independent effects ofIGFBP-3.

Acknowledgements

This research was supported by a grant from the CanadianInstitutes for Health Research. L.J.M is a recipient of a CIHRSenior Scientist award and the Henry G Friesen Chair inEndocrine and Metabolic Research.

References

1. Jones JI, Clemmons DR: Insulin-like growth factors and their bindingproteins: Biological actions. Endocr Rev 16: 3–34, 1995

2. Valentinis B, Bhala A, DeAngelis T, Baserga R, Cohen P: The humaninsulin-like growth factor (IGF) binding protein-3 inhibits the growth offibroblasts with a targeted disruption of the IGF-I receptor gene. MoleEndocrinol 9: 361–367, 1996

3. Oh Y, Gucev Z, Ng L, Muller HL, Rosenfeld RG: Antiproliferative ac-tions of insulin-like growth factor binding protein (IGFBP)-3 in humanbreast cancer cells. Prog Growth Factor Res 6: 205–212, 1995

4. Lalou C, Lassarre C, Binoux MA: A proteolytic fragment of insulin-likegrowth factor (IGF) binding protein-3 that fails to bind IGFs inhibits themitogenic effects of IGF-I and insulin. Endocrinology 137: 3206–3212,1996

5. Hong J, Zhang G, Dong F, Rechler MM: Insulin-like growth factor (IGF)binding protein-3 mutants that do not bind IGF-I or IGF-II stimulateapoptosis in human prostate cancer cells. J Biol Chem 277: 10489–10497, 2002

6. Lalou C, Sawamura S, Segovia B, Ogawa Y, Binoux M: Proteolyticfragments of insulin-like growth factor binding protein-3: N-terminalsequences and relationships between structure and biological activity.C R Acad Sci III 320: 621–628, 1997

7. Zadeh SM, Binoux M: The 16-kDa proteolytic fragment of insulin-likegrowth factor (IGF) binding protein-3 inhibits the mitogenic action offibroblast growth factor on mouse fibroblasts with a targeted disrup-tion of the type 1 IGF receptor gene. Endocrinology 138: 3069–3072,1997

8. Karas M, Danilenko M, Fishman D, LeRoith D, Levy J, Sharoni, Y:Membrane-associated insulin-like growth factor-binding protein-3 in-hibits insulin-like growth factor-I-induced insulin-like growth factor-Ireceptor signaling in Ishikawa endometrial cancer cells. J Biol Chem272: 16514–16520, 1997

9. Oh Y, Muller HL, Lamson G, Rosenfeld RG: Insulin-like growth factor(IGF)-independent action of IGF-binding protein-3 in Hs578T humanbreast cancer cells. Cell surface binding and growth inhibition. J BiolChem 268: 14964–14971, 1993

10. Ricort JM, Binoux M: Insulin-like growth factor (IGF) binding protein-3inhibits type 1 IGF receptor activation independently of its IGF bindingaffinity. Endocrinology 142: 108–113, 2001

11. Lee KW, Cohen P: Nuclear effects: unexpected intracellular actions ofinsulin-like growth factor binding protein-3. J Endocrinol 175: 33–40,2002

12. Clemmons, DR: Use of mutagenesis to probe IGF-binding protein struc-ture/function relationships. Endocr Rev 22: 800–817, 2001

13. Hollowood AD, Stewart CE, Perks CM, Pell JM, Lai T, Alderson D,Holly JM: Evidence implicating a mid-region sequence of IGFBP-3 inits specific IGF-independent actions. J Cell Biochem 86: 583–589, 2002

14. Mangel WF, Toledo DL, Nardulli AM, Reiner GC, Norman MJKatzenellenbogen BS: Plasminogen activators in human breast cancercell lines: Hormonal regulation and properties. J Steroid Biochem 30:79–88, 1988

15. Mishra S, Murphy LJ: Phosphorylation of insulin-like growth factor(IGF) binding protein-3 by breast cancer cell membranes enhances IGF-I binding. Endocrinology. 144: 4042–4050, 2003

16. Oh Y, Muller HL, Pham H, Rosenfeld RG: Demonstration of receptorsfor insulin-like growth factor binding protein-3 on Hs578T human breastcancer cells. J Biol Chem 268: 26045–26048, 1993