(cid:1) 2 -Macroglobulin: a New Component in the Insulin-like Growth Factor/Insulin-like Growth Factor Binding Protein-1 Axis*

Insulin-like growth factors (IGFs) are crucial for many aspects of development, growth, and metabolism yet control of their activity by IGF-binding proteins (IGFBPs) remains controversial. The effect of IGFBP-1 depends on its phosphorylation status; phosphorylated IGFBP-1 inhibits IGF actions whereas the nonphosphorylated isoform is stimulatory. In order to understand this phenomenon, we purified phosphorylated IGFBP-1 from normal human plasma by immunoaffinity chromatography. Un-expectedly, the resulting preparation enhanced IGF-stimulated 3T3-L1 fibroblast proliferation, due to the presence of a co-purified protein of (cid:1) 700 kDa. Matrix-assisted laser desorption ionization-mass spectrometry and Western immunoblotting analysis identified this co-purified protein as (cid:1) 2 -macroglobulin ( (cid:1) 2 M). Anti- (cid:1) 2 M an- tibodies co-immunoprecipitated IGFBP-1 from human plasma and from 125 I-IGFBP-1 (cid:1) (cid:1) 2 M complexes formed in vitro . The 125 I-IGFBP-1/ (cid:1) 2 M association could be inhibited with excess unlabeled IGFBP-1. Surface plasmon resonance analysis indicated that (cid:1) 2 M preferentially associ- ates with the phosphorylated isoform of IGFBP-1 and that when complexed to (cid:1) 2 M, IGFBP-1 can still bind IGF-I. These findings have functional significance since (cid:1) 2 M protects IGFBP-1 from proteolysis and abrogates the inhibitory effect of phosphorylated IGFBP-1 on IGF-I stimulated 3T3-L1 cell proliferation. We conclude that (cid:1) 2 M is a binding protein of IGFBP-1 which modifies IGF-I/IG-FBP-1 actions resulting in enhanced IGF effects. In line its the and of we predict that (cid:1) 2 M has a novel and important role in controlling the transport and biological activity of IGFs. was used to isolate Peptides— 2 -Macroglobulin Phosphorylated and nonphosphorylated preparations and IGF-I were iodinated to a specific activity of 70 and 100 (cid:3) Ci/ (cid:3) g, respectively. Protein Digestion and Analysis by MALDI-MS— The IGFBP-1 preparation purified by immunoaffinity chromatography (10 (cid:3) l) was ana- lyzed by SDS-PAGE and visualized with a silver stain. Bands of interest were excised from the gel and subjected to digestion with trypsin in accordance with the method of Shevchenko et al . (18). Digests were analyzed by matrix-assisted laser desorption ionization (MALDI)-MS with a VG Tofspec E mass spectrometer. The resulting peptide masses were mapped with the ProFound Internet peptide data base search site. Immunoprecipitation and Western Immunoblotting— Samples were incubated with polyclonal anti- (cid:1) 2 M (Sigma) or monoclonal anti-IG- FBP-1 (6303) antibodies overnight at 4 °C and then 25 (cid:3) l of protein-A-Sepharose CL-4B (Zymed Laboratories Inc.) for 2 h atroom tempera- ture. The immune complexes were pelleted by centrifugation, washed (4 times) Triton X-100, 0.25% bovine 35 (cid:3) nonreducing M Tris-Cl,

IGFBP-1 inhibits IGF actions by competing with the type 1 IGF receptor for IGF binding, however, the mechanism by which IGFBP-1 enhances IGF activity is less certain. Early work to address this phenomenon resulted in the isolation of two IGFBP-1 isoforms from amniotic fluid, which had similar physicochemical properties but markedly different effects on IGF activity (5). IGFBP-1 association with the cell surface was suggested as an explanation of these findings since only the stimulatory isoform was found to bind to cell membranes. It is now known that IGFBP-1 binds to ␣ 5 ␤ 1 integrin via its RGD site and disruption of this interaction leads to inhibition of the subsequent cellular response (6). Polymerization of IGFBP-1 was also postulated as a mechanism for enhancing IGF action (7) and this has been confirmed recently (8).
Many studies have also focused on the influence of phosphorylation in relation to IGFBP-1 effects on IGF activity. The inhibitory isoform purified by Busby et al. (5) was subsequently shown to be phosphorylated (9) whereas the stimulatory preparation contained nonphosphorylated IGFBP-1. Highly phosphorylated IGFBP-1, which is the only form found in plasma (10) has a high affinity for IGF-I (9,11) and can therefore inhibit IGF-I actions by sequestering it from cell surface receptors. Nonphosphorylated IGFBP-1, which has a relatively low affinity for IGF (9,11) is thought to allow more IGF/IGF receptor interactions and this hypothesis has been supported by numerous in vitro studies (9,(12)(13)(14). However, it is unclear whether in vivo alteration of IGFBP-1 phosphorylation status represents an important mechanism for regulating IGF bioavailability in the non-pregnant adult, since non-and lesser phosphorylated isoforms of IGFBP-1 are only present at high concentrations during pregnancy (10,15).
In the light of the above findings, we were surprised to observe that phosphorylated IGFBP-1 purified from plasma could enhance IGF-I stimulated cell proliferation. Further biochemical analysis led to the discovery that IGFBP-1 in plasma is associated with the homotetrameric glycoprotein ␣ 2 -macroglobulin (␣ 2 M). This paper describes our characterization of the ␣ 2 M/IGFBP-1 association and its functional impact on the IGF axis.

MATERIALS AND METHODS
Purification of IGFBP-1 from Plasma-Immunoaffinity chromatography was used to isolate IGFBP-1 from normal human plasma. Monoclonal antibody 6303 (a kind gift of Medix Biochemica, Kauniainen, Finland) was coupled to Sephacryl S-300 (16) at 1 mg/ml to form the immunoaffinity matrix. A 10-ml column was equilibrated for 24 h at 4°C by the application of PBS, 0.25% bovine serum albumin, 0.1% Tween 20 at a flow rate of 5 ml/h. 250 ml of plasma was recirculated through the column for 72 h at a flow rate of 3.75 ml/h, the column was washed with 100 ml of Tris buffer, pH 8.0 (50 mM Tris, 0.5 M NaCl, 0.1% Tween 20), and then the bound peptide was eluted by application of 0.1 M hydrochloric acid. 10 ϫ 1-ml fractions were collected into tubes containing 200 l of 1 M Tris, pH 9.0, and analyzed for IGFBP-1 by radioimmunoassay (10). Fractions containing Ͼ100 g/liter IGFBP-1 were pooled and concentrated by centrifugation through Centricon 10 filters (Amicon, Stonehouse, Gloucestershire, UK).
Biochemical Characterization of IGFBP-1 Phosphorylation Status-IGFBP-1 phosphorylation status was determined using our previously described method of immunoprecipitation followed by n-octylglucoside electrophoresis and Western ligand blotting (10). Samples were incubated with anti-IGFBP-1 (6303) antibody at 4°C overnight and then anti-mouse IgG antibody (Sac Cel; IDS, Tyne & Wear, United Kingdom) was added for 1 h at room temperature. Bound antibody was separated by centrifugation for 10 min and the precipitated proteins were washed in PBS, 0.25% bovine serum albumin, 0.1% Tween 20 prior to resuspension in gel loading buffer. All samples were then boiled for 5 min. Electrophoresis was performed using stacking (4%) and resolving (12%) gels containing 20 mM non-ionic detergent n-octylglucoside. Following overnight transfer onto nitrocellulose membranes, proteins were revealed by incubation with 150,000 cpm/ml 125 I IGF-I (4 h at 25°C) and autoradiography.
Peptides-␣ 2 -Macroglobulin was obtained as a gift from Dr. Claus Oxvig, University of Aarhus, Denmark (17), and also purchased from Sigma. Recombinant human IGFBP-1 and recombinant IGF-I were the kind gift of Dr. V. Quarmby (Genentech Inc, San Francisco, CA). Phosphorylated and nonphosphorylated IGFBP-1 preparations were also purchased from Sigma (Dorset, UK). IGFBP-1 and IGF-I were iodinated to a specific activity of 70 and 100 Ci/g, respectively.
Protein Digestion and Analysis by MALDI-MS-The IGFBP-1 preparation purified by immunoaffinity chromatography (10 l) was analyzed by SDS-PAGE and visualized with a silver stain. Bands of interest were excised from the gel and subjected to digestion with trypsin in accordance with the method of Shevchenko et al. (18). Digests were analyzed by matrix-assisted laser desorption ionization (MALDI)-MS with a VG Tofspec E mass spectrometer. The resulting peptide masses were mapped with the ProFound Internet peptide data base search site.
Affinity Labeling and Cross-linking-Complex formation was carried out in a reaction mixture containing 10 g of ␣ 2 M or 10 l of plasma and 1 nM 125 I-IGFBP-1 in 200 l of PBS, 0.2% Triton X-100 for 4 h at 37°C. The associated proteins were cross-linked according to the protocol of Vaughan and Vale (19); BS 3 (Pierce; final concentration 0.5 mM) was added for 30 min at room temperature followed by 10 l of 2.5 M glycine to stop the reaction. In some experiments, these reactions were performed in the presence of excess unlabeled IGFBP-1. A 20-l aliquot of each sample was added to 2 ϫ nonreducing SDS loading buffer and the remainder was immunoprecipitated with an antibody to h␣ 2 M or hIGFBP-1.
Immunoprecipitation and Analysis of Affinity Labeled Complexes-Cross-linked affinity labeled complexes were incubated overnight at 4°C with either 10 l of rabbit anti-h␣ 2 M or 10 l of mouse anti-IGFBP-1. Antibody-bound complexes were precipitated by the addition of 25 l of protein-A-Sepharose CL-4B and analyzed on 4 -12% SDS gradient gels followed by autoradiography.
All experiments were performed at 25°C and a constant flow rate of 15 l/min. Immediately before injection of ligand, the surface of the IGFBP-1 sensor chip was preconditioned using 100 mM HCl, ensuring equal conditions for analysis of all samples. A 10-l injection of ligate (200 ng/ml IGF-I or 200 g/ml ␣ 2 M diluted in HBS buffer) was passed across the immobilized IGFBP-1 and the binding profile recorded. Dissociation of bound ligate from immobilized ligand, initiated by flowing buffer across the sensor surface, was monitored for 500 s. This was followed by a regeneration phase (35 l of 100 mM HCl) to dissociate the remaining ligand from the binding protein and provide a ligate free surface for subsequent interaction analyses.
Proteolysis of IGFBP-1-125 I-IGFBP-1 was incubated with 1 g of chymotrypsin in the absence or presence of 1-10 g of ␣ 2 M in a final volume of 20 l of PBS, 0.5 mM CaCl 2 for 16 h at 37°C. Following addition of 2 ϫ SDS loading buffer, samples were subjected to 10% SDS-PAGE followed by autoradiography.
Activation of ␣ 2 M-␣ 2 M was activated by reaction with 300 mM methylamine as previously reported (21). Tris borate native electrophoresis (22) demonstrated differing mobility between the activated and native isoforms indicating that conformational change and thus activation had occurred. This was confirmed by comparing uptake of 125 I-␣ 2 M and 125 I-activated ␣ 2 M (iodinated by chloramine T to a specific activity of 0.4 Ci/g) in mouse embryo fibroblast cells expressing the LRP receptor which only recognizes ␣ 2 M in the activated form. Briefly, cells plated at 2 ϫ 10 5 were serum starved for 3 h before the addition of 3 nM 125 I-␣ 2 M or 125 I-activated ␣ 2 M. After 10 -360 min incubation at 37°C, cells were washed 2 times with PBS and then incubated for 4 min at 25°C with EDTA, trypsin, 0.2 mg/ml proteinase K. The resulting cell suspension was centrifuged at 13,000 rpm; membrane bound versus incorporated 125 I-␣ 2 M was determined by counting the supernatant and solubilized cell pellet, respectively.
[ 3 H]Thymidine Uptake Assay-Cells were serum starved for 24 h before the addition of IGF-I (10 ng/ml) Ϯ IGFBP-1 (40 ng/ml) and/or 100 g/ml activated ␣ 2 M. 20 h later, [methyl-3 H]thymidine was added to a final concentration of 0.25 Ci/ml and after a further 4 h, cells were washed twice with PBS and once with 10% trichloroacetic acid. The cells were incubated with 10% trichloroacetic acid for 2 h at 4°C and solubilized with 0.1 M NaOH and counted on a ␤-counter using Optiphase HiSafe liquid scintillant.

Effect of IGFBP-1 Purified from Plasma on IGF-I Stimulated [ 3 H]Thymidine Uptake by 3T3-L1
Fibroblasts-IGF-I (10 ng/ ml) caused a 1.5-fold increase (p Ͻ 0.005) in [ 3 H]thymidine uptake by 3T3-L1 fibroblasts in mid-log growth (Fig. 1). This was unaffected by nonphosphorylated IGFBP-1 added at a 1:1 molar ratio (40 ng/ml; Fig. 1). Contrary to our expectations, however, 40 ng/ml of a preparation of phosphorylated IGFBP-1 isolated from normal human plasma, enhanced the effect of IGF-I by 5-fold (p Ͻ 0.0005; Fig. 1). Furthermore, the IGFBP-1 preparation also increased [ 3 H]thymidine uptake independently of IGF-I (130% increase over control; p Ͻ 0.001) whereas npIGFBP-1 alone had no effect on cell proliferation. Enhancement of IGF action by the presence of phosphorylated IGFBP-1 was unexpected since this high affinity isoform was previously shown to be inhibitory (9) (11). This was not due to the presence of co-purified IGF-I since measurement of IGF-I in the purified IGFBP-1 preparation indicated that levels were below the detection limit of our radioimmunoassay (0.8 ng/ml; data not shown). We therefore questioned if the 3T3-L1 cells had dephosphorylated the IGFBP-1 to produce the nonphosphorylated isoform that is thought to enhance IGF activity. The phosphorylation status of IGFBP-1 was monitored by n-octylglucoside electrophoresis/Western ligand blotting before and after exposure to 3T3-L1 cells. Fig. 1B shows that 3T3-L1 cells do not dephosphorylate IGFBP-1 because the nonphosphorylated isoform could not be detected in the medium harvested from these cells after 24 h. The control in this experiment was the BeWo choriocarcinoma cell line, which has placental alka-line phosphatase and converts phosphorylated IGFBP-1 to the fully dephosphorylated form.
Protein Co-purified with IGFBP-1 Is Identified as ␣ 2 -Macroglobulin-These results suggested that a protein which had co-purified with IGFBP-1 might be enhancing IGF-I action on target cells. Indeed, increased IGF-I stimulated [ 3 H]thymidine uptake was observed in response to a plasma-derived preparation that had been depleted of IGFBP-1 by immunoprecipitation (data not shown). Proteins in the IGFBP-1 preparation were therefore isolated from silver-stained SDS-polyacrylamide gels, treated with trypsin, and subjected to MALDI-MS analysis. The predominant contaminating component was identified using the Pro-Found data base as ␣ 2 -macroglobulin.
IGFBP-1/␣ 2 M Are Associated in Plasma-Western immunoblotting with an anti-human ␣ 2 M antibody confirmed the presence of ␣ 2 M in the IGFBP-1 preparation purified from plasma ( Fig. 2A). Immunoprecipitates of human plasma with an antihuman IGFBP-1 antibody, contained a high molecular weight protein that co-migrated with human ␣ 2 M, providing further proof of the association between IGFBP-1 and ␣ 2 M (Fig. 2B).
Characterization of IGFBP-1/␣ 2 M Complexes Formed in Vitro-IGFBP-1/␣ 2 M binding in solution was assessed by incubating 125 I-IGFBP-1 with human plasma (10 l) for 4 h, fixing the resulting complexes with the cross-linking agent BS 3 , and SDS-PAGE analysis both before and after immunoprecipitation with an anti-␣ 2 M antibody. Fig. 3A demonstrates that 125 I-IGFBP-1 can associate with a high molecular weight species in plasma which can be immunoprecipitated with an anti-body to ␣ 2 M. This complex co-migrated with the labeled species seen as the result of 125 I-IGFBP-1 incubation with ␣ 2 M (10 g). The high molecular weight 125 I-IGFBP-1/human plasma-binding protein and the 125 I-IGFBP-1⅐␣ 2 M complexes could also be precipitated by an antibody to IGFBP-1 (Fig. 3B) although not by protein A-Sepharose CL-4B alone. The radioactive species migrating to ϳ30 kDa represents uncomplexed 125 I-IGFBP-1. 125 I-IGFBP-1 also formed high molecular weight complexes with activated ␣ 2 M (data not shown).
Surface Plasmon Resonance Analysis of IGFBP-1/␣ 2 M Association-Surface plasmon resonance was used to investigate further the association of IGFBP-1 and ␣ 2 M; using standard amine coupling procedures, phosphorylated or nonphosphorylated IGFBP-1 was immobilized on a BIAcore sensor chip CM5 and then exposed to ␣ 2 M. The sensorgram depicted in Fig. 4A shows binding of ␣ 2 M to phosphorylated IGFBP-1. ␣ 2 M was applied at a concentration sufficient to saturate the immobilized IGFBP-1, based on a 1:1 binding ratio. After binding ␣ 2 M, the chip was further exposed to IGF-I and Fig. 4A shows that IGFBP-1 can still bind IGF-I in the presence of ␣ 2 M. Nonphosphorylated IGFBP-1, however, does not bind to ␣ 2 M (Fig. 4B) although IGF-I binding of this isoform is evident.
␣ 2 M Protects IGFBP-1 from Proteolysis-Since ␣ 2 M is a recognized protease inhibitor, one physiological consequence of the IGFBP-1/␣ 2 M association could be protection of IGFBP-1 from proteolysis. 125 I-IGFBP-1 was incubated with 1 g of chymotrypsin (23) in the presence of ␣ 2 M (0-10 g). ments are apparent, suggesting that chymotrypsin activity is reduced by a low concentration of ␣ 2 M. IGFBP-1 proteolysis was completely abolished by the presence of 10 g of ␣ 2 M; here the majority of 125 I-IGFBP-1 was detected in high molecular weight complexes which co-migrate with the radiolabeled species observed when IGFBP-1 and ␣ 2 M are incubated in the absence of chymotrypsin. This suggests that in this instance, ␣ 2 M protects against proteolysis by associating with the substrate (IGFBP-1) rather than the protease.

Effect of ␣ 2 M on IGF-I-stimulated [ 3 H]Thymidine
Uptake-␣ 2 M influences the action of other growth factors and therefore the effect of ␣ 2 M on IGF-I stimulated [ 3 H]thymidine uptake by 3T3-L1 fibroblasts was investigated to determine whether ␣ 2 M could be responsible for the enhanced stimulation observed in the presence of plasma-derived IGFBP-1. Conversion of ␣ 2 M into the activated isoform, which is recognized by the ␣ 2 M receptor LRP (Fig. 6A (i)), was achieved by reaction with methylamine. Activated ␣ 2 M (100 g/ml) but not the native isoform (100 g/ml), was able to enhance IGF-I-stimulated [ 3 H]thymidine uptake 2-fold (p Ͻ 0.001; Fig. 6A (ii)). Importantly, ␣ 2 M could also abrogate the inhibitory effect of IGFBP-1; an HPLCpurified preparation of phosphorylated IGFBP-1 could reduce IGF action (Fig. 6B, p Ͻ 0.01), however, when 100 g/ml ␣ 2 M was also included in the incubation, IGF-I-stimulated [ 3 H]thymidine uptake (p Ͻ 0.01) was enhanced. Activated ␣ 2 M (200 g/ml) also had an independent effect on mitogenesis (p Ͻ 0.05), whereas phosphorylated IGFBP-1 (40 ng/ml) alone had no effect (Fig. 6B).

DISCUSSION
Analysis of the highly phosphorylated IGFBP-1 found in the circulation of non-pregnant adults has demonstrated association with ␣ 2 -macroglobulin, another plasma protein. ␣ 2 M can enhance IGF-I-stimulated proliferation of fibroblasts, even in the presence of phosphorylated IGFBP-1, however, when ␣ 2 M is absent, phosphorylated IGFBP-1 is inhibitory.
Highly phosphorylated IGFBP-1 has a high affinity for IGF-I (9, 11) and can therefore sequester IGF-I from its cell surface receptors. In order for IGFBP-1 enhancement of IGF action to occur, it is generally thought that IGFBP-1 must be either in the nonphosphorylated isoform, which has a lower affinity for IGF-I (5), or associated with cell membranes (6, 24). We found that increased IGF action in the presence of a preparation of IGFBP-1 purified from plasma was not the result of dephosphorylation by 3T3-L1 cells. BeWo choriocarcinoma cells, however, are able to dephosphorylate IGFBP-1 (23) and this isoform of IGFBP-1 is normally only detected in the human during pregnancy (10), raising the possibility that altering IGFBP-1 phosphorylation status may predominantly be a mechanism for regulating IGF bioavailability in pregnancy.
An early report from Clemmons and Gardner (25) suggested that a factor present in plasma was necessary for IGFBP-1 to potentiate IGF stimulated smooth muscle cell DNA synthesis. This group discerned the factor to be a macromolecule and excluded mitogens such as platelet-derived growth factor, epidermal growth factor, and fibroblast growth factor and the carrier proteins transferrin, albumin, and fibronectin, although its identity remained elusive. These studies led us to suspect that our purified preparation of plasma IGFBP-1 contained another plasma component that had co-purified on the anti-IGFBP-1 monoclonal antibody immunoaffinity column by virtue of its association with IGFBP-1 in plasma.
Radioimmunoassay of the IGFBP-1 preparation refuted our initial assumption that IGF-I was the contaminating protein and so we then used a combination of MALDI-TOF mass spectrometry and co-imunoprecipitation/Western blot studies, to identify the co-purified factor as ␣ 2 -macroglobulin.
Early size exclusion chromatographic studies of plasma did not detect the association between IGFBP-1 and ␣ 2 M. Several explanations for this are tenable. First ␣ 2 M is a large protein (Ϸ700,000) which would have been outside the molecular weight range of most chromatographic studies. Appearance of material within the void volume could have been discounted as being due to protein solubility/aggregation. Second, the association of IGFBP-1 with ␣ 2 M may be low affinity resulting in dissociation of the complex on size exclusion chromatography; these studies indicate a k d of 2.75 ϫ 10 Ϫ3 s Ϫ1 , which supports this hypothesis. Third it is possible that some IGFBP-1 antibodies cannot recognize the peptide when bound to ␣ 2 M. Furthermore, initial chromatographic studies of other IGFBPs did not demonstrate associations with plasma proteins and yet it has recently been recognized that this is the case. For example, IGFBP-3 specifically binds to lactoferrin (26), transferrin (27), type I collagen (28), and fibronectin (29) although the physiological significance of these associations is as yet unclear.
␣ 2 M is a homotetrameric glycoprotein that circulates at concentrations of 2-4 mg/ml (30). Each subunit contains multiple reactive sites suggesting that ␣ 2 M has diversified functions as a binding, carrier, and targeting protein and it may therefore be important for several aspects of IGFBP-1 function. ␣ 2 M is well known as a protease inhibitor and has the unique ability of being able to inhibit proteinases from all four mech-anistic classes (30). Protease cleavage of native ␣ 2 M results in a conformational change to form activated-␣ 2 M which traps the protease so that it is sterically hindered from access to substrate. Nonproteolytic peptides trapped by ␣ 2 M become largely protected from exogenous proteases and our own results show that when associated with ␣ 2 M, IGFBP-1 is protected from proteolysis by chymotrypsin. Thus in plasma, ␣ 2 M may be acting as a chaperone to IGFBP-1 and this may explain why there are no reports of IGFBP-1 fragments in the circulation despite the fact that IGFBP-1 has many of the cleavage motifs displayed by other circulating IGFBPs which are proteolysed.
Another physiological consequence of IGFBP-1/␣ 2 M association may be in regulating IGF activity. ␣ 2 M is known to associate with several other growth factors including fibroblast growth factor (31), vascular endothelial growth factor (32), epidermal growth factor (33), transforming growth factor-␤1 (34), and platelet-derived growth factor (35) to synergistically enhance their action on cell proliferation (36). We have found that ␣ 2 M also enhances IGF-I-stimulated mitogenesis and importantly, that ␣ 2 M can also influence how IGFBP-1 modulates this effect. In our initial studies using the IGFBP-1 preparation purified from plasma, IGF-I stimulated [ 3 H]thymidine uptake by 3T3-L1 fibroblasts was seemingly enhanced by phosphorylated IGFBP-1 and the preparation also appeared to have an independent mitogenic effect. However, we have now demonstrated that this effect was more likely due to the presence of ␣ 2 M.
The mechanism behind ␣ 2 M enhancement of IGF-I activity is unclear, although it probably involves one of the two ␣ 2 M receptors, LRP/␣ 2 MR, or the recently described ␣ 2 M signaling receptor (␣ 2 MSR). The LRP/␣ 2 MR is a member of the low density lipoprotein receptor superfamily (37-39) which recognizes both free and growth factor-associated ␣ 2 M once it has been activated by proteases or amines such as methylamine (40 -42). Thus ␣ 2 M, through IGFBP-1, could serve to increase the concentration of IGF-I in the local environment of the cell; dissociation at or near the cell surface could release IGF-I to interact with its cell surface receptor. Our preparation of native ␣ 2 M was unable to enhance IGF-I activity, whereas activated ␣ 2 M did enhance IGF actions. These data strongly suggest that such synergy was due to ␣ 2 M/receptor interactions and not the presence of an additional ␣ 2 M-associated mitogen such as platelet-derived growth factor. Alternatively, the LRP/␣ 2 MR scavenges ␣ 2 M complexes resulting in their rapid clearance from the circulation and thus internalization of IGFBP-1 along with ␣ 2 M could alleviate the inhibitory effect on paracrine or autocrine IGF-I, or, if both IGF-I and IGFBP-1 were internalized, their dissociation within an endocytic compartment could influence signal transduction in a manner analogous to that of epidermal growth factor (43)(44)(45).
A further mechanism by which ␣ 2 M may regulate cellular growth involves a second ␣ 2 M receptor (46), which also recognizes only activated ␣ 2 M. ␣ 2 M interaction with this receptor increases phosphatidylinositol 3-kinase activity (47) and elevates p21 Ras -GTP (48), which may explain our finding that ␣ 2 M acts as an independent mitogen. However, signaling through the type 1 IGF receptor also involves activation of the phosphatidylinositol 3-kinase and Ras/Raf pathways (1) and so it is possible that ␣ 2 M could enhance endogenous and exogenous IGF-I actions as a result of synergy between their intracellular signaling components.
In summary, we have demonstrated that ␣ 2 M is a binding protein of IGFBP-1 which modifies IGFBP-1/IGF interaction. This represents a novel and potentially important mechanism for controlling the transport and biological activity of IGFs since the effect of IGFBP-1 on IGF activity depend not only on IGFBP-1 phosphorylation status, but also on whether IGFBP-1 FIG. 6. A, confirmation of ␣ 2 M activation by methylamine. ␣ 2 M was activated by incubation with 200 mM methylamine and then analyzed by native Tris borate electrophoresis and silver staining. Activation was confirmed by comparing uptake of activated () and native () 125 I-␣ 2 M by mouse embryo fibroblast cells expressing the LRP receptor, which only recognizes ␣ 2 M in the activated form (i). 3T3-L1 cells were then used to compare the effect of active and native ␣ 2 M on IGF-I activity (ii). Serum-starved cells were incubated with 10 ng/ml IGF-I Ϯ 100 g/ml ␣ 2 M for 20 h before the addition of 0.25 Ci/ml [ 3 H]thymidine for a further 4 h. Uptake is expressed as percentage increase over control and is shown as the mean (Ϯ S.D.) of three experiments performed in triplicate. B, effect of ␣ 2 M and pIGFBP-1 on IGF-I stimulated [ 3 H]thymidine uptake by 3T3-L1 fibroblasts. IGF-I (10 ng/ml) and IGFBP-1 (40 ng/ml) were incubated in the presence or absence of activated ␣ 2 M (100 g/ml) and effects on [ 3 H]thymidine uptake were determined as described above. The effect of ␣ 2 M (200 g/ml) and IGFBP-1 (40 ng/ml) alone or in combination was also determined. Uptake is expressed as percentage increase over control and is shown as the mean (Ϯ S.D.) of three experiments performed in triplicate.