INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

THE INSULIN-LIKE GROWTH FACTOR-BINDING PROTEINS (IGFBPs) constitute a family of six high-affinity insulin-like growth factor (IGF)-binding proteins found in the circulation and produced by several types of cells. Of these six proteins, IGFBP-3 is the most prevalent, circulating primarily as a 140-kDa ternary complex with IGF-I or IGF-II and an acid-labile subunit (ALS) (3, 14). Several proteases have been shown to cleave IGFBP-3 (24), including plasmin (1, 4, 6, 9, 13, 20), thrombin (6, 26), kallikrein (13), prostate-specific antigen (11), matrix metalloproteases (16, 19), and cathepsin D (10, 12).

The position of a number of plasmin and thrombin cleavage sites on IGFBP-3 has been determined (6). In addition to these sites, additional plasmin cleavage sites must be present, because further digestion with plasmin led to the disappearance of both intact IGFBP-3 and IGFBP-3 digestion products (6).

IGFBP-3 binds to several cell types, including cultured bovine microvessel endothelial cells (5, 15, 23). The affinity of IGFBP-3 for IGF is greater than the affinity of the type 1 IGF receptor, whereas binding of IGFBP-3 to cells appears to decrease the affinity of IGFBP-3 for IGF (3, 23). It has been suggested that this decrease in affinity of cell-bound IGFBP-3 for IGF increases the release of IGF from cell-surface IGFBP-3, facilitating IGF binding to IGF receptors. (3, 23).

IGFBP-3 contains a highly basic heparin-binding area, the P3 region, in its COOH-terminal portion (5). This region contains the cell-surface recognition site for IGFBP-3 (18), a nuclear localization sequence (17, 21, 24), and is important in ALS binding (15, 25). A similar region, P5, is found in IGFBP-5 and appears to serve a comparable function (5, 8, 24, 25). Recent studies have shown that the P3 region of IGFBP-3, or the P5 region in IGFBP-5, serves as a binding site for the proenzymes plasminogen and prekallikrein (9, 13). Activation of plasminogen to plasmin or prekallikrein to kallikrein resulted in proteolysis of IGFBP-3. Proteolysis of IGFBP-3 by plasmin presumably decreases its association with IGF, freeing IGF to interact with its receptor (4, 9, 13, 20).

Chimeras of IGFBP-3 and IGFBP-4, IGFBP-3₄ and IGFBP-4₃, were constructed; the P3 region in IGFBP-3 was replaced by the equivalent region of IGFBP-4 and vice versa (18). These chimeras show changes in cellular binding properties; although IGFBP-4 does not bind to endothelial cells, cellular binding of IGFBP-4₃ is equal to or greater than cellular binding of IGFBP-3, whereas binding of IGFBP-3₄ is minimal (18). The purpose of the present study was to further investigate, using these IGFBP chimeras, the effect of the P3 region on protease sensitivity to plasmin and thrombin. The chimeras IGFBP-3₄ and IGFBP-4₃ show altered plasmin and thrombin sensitivity relative to IGFBP-3 or IGFBP-4. The apparent importance of the P3 region was not restricted to binding of a proenzyme followed by activation but was demonstrated using the active protease. Perhaps most significantly, we demonstrate that binding of IGFBP-3 or IGFBP-4₃ to endothelial cells protects these proteins from degradation by plasmin or thrombin.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Recombinant IGFBP-3 and IGFBP-4, and their chimeras, IGFBP-3₄ and IGFBP-4₃, were isolated as secreted proteins in a baculovirus system and purified on IGF affinity columns, as previously described (18). Unlabeled IGFBP-3, IGFBP-3₄, IGFBP-4, and IGFBP-4₃ (20 µg/20 µl) were incubated with 0.05 µl (1×) or 0.1 µl (2×) plasmin, stock concentration 2 µg/µl (American Diagnostica, Greenwich, CT) for 20 min at 37°C in PBS, pH 7.3, or PBS containing 90 mM Tris, pH 7.5 (PBS-Tris). Plasmin was used at molar ratios of ~1:300 and ~1:600 plasmin/IGFBP. Incubation was stopped by the addition of phenylmethylsulfonyl fluoride (PMSF), final concentration 1 mM, and 2× SDS-PAGE sample buffer. Samples were immediately boiled for 5 min and electrophoresed on nonreducing 12% SDS gels. Gels were stained with Coomassie blue, scanned, and analyzed using Sigma Gel software (Jandel Scientific). The integrated area of the IGFBP band in the control (no protease) lane was compared with the integrated areas of the IGFBP bands from protease-treated samples. Unlabeled binding proteins, at 20 µg/20 µl, were also incubated with thrombin from bovine plasma (1,500-2,500 NIH units/mg protein; Sigma, St Louis, MO) at 37°C for 2 h in PBS-Tris buffer and analyzed as for plasmin.

IGFBPs were iodinated and purified on Sephadex G100 columns, as previously described (18). Concentrations of plasmin or thrombin were established that would achieve measurable ¹²⁵I-IGFBP-3 degradation in solution after a 20-min incubation at 22°C in HLB (in mM: 100 HEPES, 120 NaCl, 1.7 MgSO₄, 25 sodium acetate, 2.4 KCl, 0.8 EDTA, 10 glucose, and 10 mg/ml BSA, pH 7.8) or serum-free Medium 199 (M199; Life Technologies, Rockville, MD) plus 50 mM HEPES, pH 7.3. Aliquots were analyzed by SDS-PAGE, and dried gels were exposed to X-ray film (Eastman Kodak, Rochester, NY), scanned, and analyzed as described for stained gels. All experiments were performed at least twice, with duplicate samples in each experiment. When replicate samples were electrophoresed, the integrated areas of the ¹²⁵I-IGFBP band in two or more lanes were averaged, and the standard deviation was calculated. Data were analyzed by ANOVA, with the Newman-Keuls post test, or by t-tests where appropriate, with the use of GraphPad Prism (GraphPad Software, San Diego, CA).

Microvessel endothelial cells were prepared from bovine heart adipose tissue and characterized as previously described (2). Cells were preincubated with ¹²⁵I-IGFBPs for 1 h at 22°C in 12-well trays in HLB or M199. In some experiments, the supernatant was removed, and the cells were washed and then incubated with the appropriate protease. In other experiments using ¹²⁵I-IGFBP-3₄ (which does not bind significantly to cells) after the preincubation, the proteases were added directly to the medium, and the supernatant was not removed. Parallel incubations were conducted in solution (i.e., in the absence of cells) for 20 min at 22°C. All reactions were performed in duplicate.

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Unlabeled IGFBP-3, IGFBP-3₄, IGFBP-4, and IGFBP-4₃ (1 µg/µl) were incubated with plasmin for 20 min at 37°C. Similar conditions were previously determined to digest the majority, but not all, of the IGFBP-3 (6). Aliquots were analyzed by 12% SDS-PAGE and stained with Coomassie blue (Fig. 1). In the control ("0") lanes, there was no degradation of the binding proteins after 20 min at 37°C when compared with the starting IGFBP. After digestion with 1× plasmin, 58% of IGFBP-3 vs. 81% of IGFBP-3₄ remained intact; after digestion with 2× plasmin, 18% of IGFBP-3 vs. 72% of IGFBP-3₄ remained intact. IGFBP-4 was relatively insensitive to plasmin. As predicted on the basis of the importance of the P3 region as a plasmin-binding site, substitution of the P3 for the P4 region in IGFBP-4 (IGFBP-4₃, Fig. 1) increased susceptibility to plasmin. After digestion with 1× plasmin, 101% of IGFBP-4 vs. 54% of IGFBP-4₃ remained intact, whereas after digestion with 2× plasmin, 88% of IGFBP-4 vs. 58% of IGFBP-4₃ remained intact. Although the presence of the P3 region increased plasmin digestion of IGFBP-4₃, presumably by serving as a binding site for plasmin, there was less digestion of IGFBP-4₃ than of IGFBP-3 with 2× plasmin (58% intact for IGFBP-4₃ vs. 18% intact for IGFBP-3). Similar results were seen when IGFBP-3 and IGFBP-3₄ were treated with either plasmin or thrombin (Fig. 2). In the controls, there was no digestion after incubation at 37°C (compare C with C₀). After treatment with plasmin, 17% of IGFBP-3 and 50% of IGFBP-3₄ remained intact; after treatment with thrombin, 26% of IGFBP-3 and 44% of IGFBP-3₄ remained intact.

View larger version (62K): [in this window] [in a new window]   Fig. 1.   Treatment of insulin-like growth factor-binding protein (IGFBP)-3, IGFBP-34, IGFBP-4, and IGFBP-43 with plasmin. IGFBPs (20 µg/20 µl) were treated in PBS with 0, 1× (0.1 µg), and 2× (0.2 µg) plasmin for 20 min at 37°C. Reactions were stopped by the addition of phenylmethylsulfonyl fluoride (PMSF, final concentration 0.1 mM), 2× SDS-PAGE sample buffer, and immediate boiling. Samples were electrophoresed on a 12% SDS gel, and the gel was stained with Coomassie blue. Lanes of the scanned gel were analyzed using SigmaGel software, and the integrated area of the intact IGFBP was determined. For each binding protein, the untreated (0 plasmin) control was considered as 100%. Data are expressed as %IGFBP remaining intact after enzyme treatment.

View larger version (79K): [in this window] [in a new window]   Fig. 2.   Treatment of IGFBP-3 and IGFBP-34 with plasmin or thrombin. IGFBPs (20 µg/20 µl) were treated in PBS-Tris with plasmin (0.1 µg) for 20 min or with thrombin (6 µl) for 2 h at 37°C. Reactions were stopped by the addition of PMSF (final concentration 0.1 mM), 2× SDS-PAGE sample buffer, and immediate boiling. Samples were electrophoresed on a 12% SDS gel, and the gel was stained with Coomassie blue. Lanes of the scanned gel were analyzed using SigmaGel software, and the integrated area of the intact IGFBP was determined. For each binding protein, the untreated control was considered as 100%. Data are expressed as %IGFBP remaining intact after enzyme treatment. C0, IGFBPs at time 0; C, no enzyme; P, plasmin; T, thrombin after 20 min.

To determine whether the presence of cells could protect IGFBPs containing the P3 region from digestion, IGFBP-3, IGFBP-4₃, and IGFBP-3₄ were iodinated and exposed to plasmin and thrombin. For these studies, several preliminary experiments were performed. The concentrations of plasmin or thrombin that gave measurable digestion of ¹²⁵I-IGFBP-3 or ¹²⁵I-IGFBP-4₃ were determined. Next, studies were performed in HLB or M199 to determine conditions of preincubation to achieve near-maximal binding of IGFBPs to cells without substantial degradation in the control samples. After 1 h at 22°C in HLB, total binding was 17% of added ¹²⁵I-IGFBP-4₃, and specific binding was 14%. After 1 h in M199, total binding was 11% of added ¹²⁵I-IGFBP-4₃, and specific binding was 8%. We then determined whether plasmin or thrombin treatment of cells in HLB or M199 would affect cell morphology. Microvessel endothelial cells, in HLB or M199, were treated with appropriate concentrations of plasmin or thrombin and compared visually with untreated cells over a period of 5 h. No differences in cell morphology or attachment were observed between protease-treated and -untreated cells.

Endothelial cells were preincubated with ¹²⁵I- IGFBP-3 or ¹²⁵I-IGFBP-4₃ in HLB. The medium containing the free binding protein was removed, the cells were washed once with HLB, and then fresh HLB and the proteases were added. After 20 min, an aliquot of the supernatant was saved for SDS-PAGE analysis, the rest was discarded, and the cells were lysed in SDS-PAGE sample buffer. Parallel incubations were conducted in solution. In solution, 9 ± 2% (n = 2, ±SD) of the ¹²⁵I-IGFBP-3 remained intact after plasmin treatment, and 47 ± 2% remained intact after thrombin treatment (Fig. 3). As seen with use of the unlabeled binding proteins, in solution, ¹²⁵I-IGFBP-4₃ was less sensitive to either enzyme than ¹²⁵I-IGFBP-3, with 20 ± 11% remaining after plasmin, and 85 ± 12% remaining after thrombin treatment. After incubation in the presence of endothelial cells, there was less degradation of both cell-attached binding proteins with either enzyme. For ¹²⁵I-IGFBP-3, 53 ± 5% of cell-attached binding protein remained intact after plasmin (vs. 9 ± 2% in solution), and 70 ± 4% remained intact after thrombin (vs. 47 ± 2% in solution). For ¹²⁵I-IGFBP-4₃, 67 ± 2% remained after plasmin (vs. 20 ± 11% in solution) and 104 ± 3% after thrombin (vs. 85 ± 12% in solution) (Fig. 3). After 20 min, there was some dissociation of the bound ¹²⁵I-IGFBP into the cell supernatant. The counts in the supernatant were similar for untreated and protease-treated samples. These supernatants, i.e., the material dissociated from the cells, were analyzed (data not shown). For plasmin, the percentage of intact binding protein in the supernatant (¹²⁵I-IGFBP-3, 13 ± 1% and ¹²⁵I-IGFBP-4₃, 15 ± 0%) was similar to that seen in the solution reaction (¹²⁵I-IGFBP-3, 9 ± 2% and ¹²⁵I-IGFBP-4_3, 20 ± 11%). For thrombin, the material dissociated from the cells (¹²⁵I-IGFBP-3, 19 ± 2% and ¹²⁵I-IGFBP-4₃, 27 ± 3%) was digested to a greater extent than that seen in the solution reaction (¹²⁵I-IGFBP-3, 47 ± 2% and ¹²⁵I-IGFBP-4_3, 85 ± 12%).

View larger version (89K): [in this window] [in a new window]   Fig. 3.   Treatment of 125I-IGFBP-3 or 125I-IGFBP-43 with plasmin or thrombin in the absence (Solution) or presence (Cells) of endothelial cells. Confluent cells in 12-well trays were preincubated with 125I-IGFBPs (0.3 × 106 counts/min) in 0.3 ml HLB at 22°C for 1 h. After 1 h, the supernatant was removed. Cells were washed with HLB, and 0.3 ml of fresh HLB was added. Plasmin (1.2 µg) or thrombin (30 µl) was added, and cells were incubated in a final volume of 360 µl at 22°C. After 20 min, PMSF was added (final concentration 0.1 mM), the supernatant was removed, and an aliquot was saved for analysis. SDS-PAGE sample buffer was added to the wells, and the cells were lysed and boiled for 5 min. Simultaneously, reactions in solution were performed in HLB for 20 min at 22°C with the same concentrations of proteases. Reactions were terminated by addition of PMSF and SDS-PAGE sample buffer and were boiled for 5 min. All reactions were carried out in duplicate. Aliquots were electrophoresed on 12% SDS gels. Gels were exposed to X-ray film, scanned, and analyzed as described in Figs. 1 and 2. Duplicate samples (means ± SD) are shown on each gel. *P < 0.05; **P < 0.01, Solution vs. Cells by t-test. C, no enzyme; P, plasmin; T, thrombin after 20 min.

There are several possible explanations for the "cell- protective" effect we observed: 1) protection of the protease-binding site (P3 region) on the binding proteins by cell attachment at that site, thus preventing plasmin or thrombin attachment; 2) protection of the binding proteins by protein(s) secreted by the endothelial cells during the 20-min incubation; 3) protection of the binding proteins by cell membrane proteins or components unrelated to the putative IGFBP-3 basic cellular binding region. To determine whether the protective effect was indeed due to the cellular binding region (P3), we designed a series of similar experiments with ¹²⁵I-IGFBP-3₄. Because only a minimal amount of this protein binds specifically to endothelial cells (<2% binding for ¹²⁵I-IGFBP-3₄ compared with 14% for ¹²⁵I-IGFBP-4₃), the cell-associated fraction and the cell-supernatant fraction were analyzed after a 1-h preincubation in HLB followed by incubation with the proteases. The nonbound material was not removed after the preincubation; the enzymes were added directly to the incubation after 1 h, and samples were incubated for a further 20 min at 22°C. ¹²⁵I-IGFBP-3₄ and ¹²⁵I-IGFBP-3 were both included in these experiments to show that, under these slightly modified conditions, the ¹²⁵I-IGFBP-3 was still protected by the cells. Additionally, a longer (40-min) incubation with thrombin (T2) was performed in an attempt to increase the degradation of ¹²⁵I-IGFBP-3₄. One set of gels is shown in Fig. 4, which indicates the percentage of ¹²⁵I-IGFBP-3 or ¹²⁵I-IGFBP-3₄ remaining intact in the solution reaction, the percentage in the supernatant of the cells (i.e., the media in which the cells were incubated), and the percentage attached to the cells. As in the previous experiments, ¹²⁵I-IGFBP-3 was protected in the presence of cells. However, ¹²⁵I-IGFBP-3₄ was not protected from plasmin digestion (Fig. 4). There was insufficient thrombin digestion of ¹²⁵I-IGFBP-3₄ in the solution reaction to study the effect of the cells. The amount of IGFBP digestion in the cell supernatant was intermediate to that seen in the solution reaction and that seen for cell-associated binding protein (Fig. 4).

View larger version (55K): [in this window] [in a new window]   Fig. 4.   Treatment of 125I-IGFBP-3 or 125IGFBP-34 with plasmin or thrombin in HLB in the absence (Solution Rx) or presence (Cell Supn, Cells) of endothelial cells. Confluent cells in 12-well trays were preincubated with 125I-IGFBPs (1 × 106 counts/min) in 0.3 ml HLB at 22°C for 1 h. Plasmin (1.2 µg) or thrombin (30 µl) was added, and cells were incubated in a final volume of 360 µl at 22°C. After 20 min (or 40 min for T2), PMSF was added to give 0.1 mM, the supernatant was removed, and an aliquot was boiled in SDS-PAGE sample buffer and saved for analysis. SDS-PAGE sample buffer was added to the cells, and the cells were lysed and boiled for 5 min. Simultaneously, reactions in solution were performed in HLB for 20 min at 22°C with the same concentrations of proteases. Digestion was terminated by addition of PMSF and SDS-PAGE sample buffer, and samples were boiled for 5 min. All reactions were carried out in duplicate. Aliquots were electrophoresed on duplicate 12% SDS gels. Gels were exposed to X-ray film, scanned, and analyzed as described in Figs. 1 and 2. One set of the duplicate gels is shown. Data are expressed as means ± SD of samples on duplicate gels. Data were analyzed by ANOVA and the Newman-Keuls post test. 125I-IGFBP-3: Plasmin, P < 0.01 Solution Rx or Cell Supn vs. Cells; Thrombin, P < 0.001 Solution Rx vs. Cells, P < 0.001 Solution Rx vs. Cell Supn, P < 0.01 Cell Supn vs. Cells; Thrombin2, P < 0.05 Solution Rx vs. Cells, P < 0.05 Solution Rx vs. Cell Supn. 125I-IGFBP-34: Plasmin, Thrombin, Thrombin2, all comparisons not significant (NS). C, no enzyme; P, plasmin; T, thrombin after 20 min; T2, thrombin after 40 min.

Because HLB contains 10 mg/ml BSA and 0.8 mM EDTA, which might affect thrombin activity, the experiments were repeated using serum-free M199 (Fig. 5). Again, the cells protected ¹²⁵I-IGFBP-3 from digestion by plasmin. ¹²⁵I-IGFBP-3 was 1 ± 1% (mean ± SD, n = 4) intact in solution, vs. cells 63 ± 24% intact. With thrombin, although there was an apparent protective effect of cells, it was not statistically significant, with the reaction in solution 39 ± 19% intact vs. that in cells 62 ± 10% intact. With ¹²⁵I-IGFBP-3₄, which binds minimally to cells, there was no significant protective effect of cells for plasmin digestion (compare 21 ± 4 with 30 ± 19%). With thrombin, however, there appeared to be increased degradation of the small amount of the ¹²⁵I-IGFBP-3₄ attached to cells (compare 103 ± 11 with 47 ± 6%). The material in the cell supernatant (predominantly unbound ¹²⁵I-IGFBP rather than dissociating bound ¹²⁵I-IGFBP) was generally "not protected" by the presence of cells.

View larger version (64K): [in this window] [in a new window]   Fig. 5.   Treatment of 125I-IGFBP-3 or 125I-IGFBP-34 with plasmin or thrombin in Medium 199 (M199) in the absence (Solution Rx) or presence (Cell Supn, Cells) of endothelial cells. Confluent cells in 12-well trays were preincubated with 125I-IGFBPs (1 × 106 counts/min) in 0.3 ml serum-free M199 at 22°C for 1 h. Plasmin (1.2 µg) or thrombin (30 µl) was added, and cells were incubated in a final volume of 360 µl at 22°C. After 20 min, PMSF was added to give 0.1 mM, the supernatant was removed, and an aliquot was boiled in SDS-PAGE sample buffer and saved for analysis. SDS-PAGE sample buffer was added to the cells, and the cells were lysed and boiled for 5 min. Simultaneously, reactions in solution were performed in M199 for 20 min at 22°C with the same concentrations of proteases, terminated by addition of PMSF and SDS-PAGE sample buffer, and boiled for 5 min. All reactions were carried out in duplicate. Aliquots were electrophoresed on duplicate 12% SDS gels. Gels were exposed to X-ray film, scanned, and analyzed as described in Figs. 1 and 2. The identical experiment was performed twice. One set of gels is shown. Data are expressed as means ± SD (n = 4). Data were analyzed by ANOVA with the Newman-Keuls post test. 125I-IGFBP-3: Plasmin, P < 0.001 Solution Rx or Cell Supn vs. Cells; Thrombin, all comparisons NS. 125I-IGFBP-34: Plasmin, all comparisons NS; Thrombin, P < 0.001 Solution Rx vs. Cells, P < 0.05 Solution Rx vs. Cell Supn, P < 0.001 Cell Supn vs. Cells. C, no enzyme; P, plasmin; T, thrombin.

It is unclear whether the P3 region, as the putative plasmin (and perhaps thrombin) binding site, is itself being cut by either of these enzymes. In our previous studies (6, 7), we sequenced major plasmin and thrombin digestion products of IGFBP-3. We did not obtain an NH₂-terminal sequence beginning in this region. However, these results do not mean that there is not a cleavage site within the P3 region, because the pieces may be too small to isolate on a 12% SDS gel. To determine whether plasmin or thrombin could cleave the P3 peptide, we incubated 20 µg of P3 (MW ~2,300) in M199 with several concentrations of plasmin and thrombin and assessed the degradation of P3 on a Coomassie-stained 16.5% Tris-Tricine gel. Plasmin digested the P3 peptide after 20 min at 37°C. Thrombin had little or no effect on the P3 peptide, even after 2 h of incubation at 37°C (data not shown).

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

The presence of the P3 region in IGFBP-3, when introduced into IGFBP-4 (i.e., IGFBP-4₃), increased the digestion of the IGFBP by plasmin and thrombin. However, although IGFBP-4₃ was more sensitive to digestion than IGFBP-4, it remained less sensitive to plasmin than IGFBP-3. When ¹²⁵I-IGFBPs were used, the presence of the P3 region similarly increased digestion of the binding protein "in solution." Perhaps of greatest importance, IGFBP-3 and IGFBP-4₃, in the presence of endothelial cells, were protected from digestion. IGFBP-3₄, which does not contain the P3 region and which binds minimally to endothelial cells (18), was not protected from plasmin or thrombin proteolysis. There was some dissociation of cell-bound ¹²⁵I-IGFBP-3 or ¹²⁵I-IGFBP-4₃ during the 20-min incubation. The plasmin digestion of this material was similar to that seen for the samples in solution, but the thrombin digestion was greater than that seen for the samples in solution.

While associated with the endothelial cell surface, binding proteins containing the P3 region were significantly protected from digestion by plasmin and thrombin. Protection of endothelial cell-associated IGFBP-3 is likely to be important, because both plasmin and thrombin are found in the endothelial milieu, especially in the case of injury to the vascular endothelium. Several studies have indicated that NH₂-terminal and possibly COOH-terminal proteolytic fragments of IGFBP-3 have biological activity unrelated to their IGF binding (1, 4, 7). Cytoprotection would decrease the generation of these fragments and their activity and potentially alter the pattern of fragments produced. It should be noted that the cytoprotection was not complete; some of the labeled binding protein in the cell fraction was not intact. A portion of this may represent cleavage of loosely or nonspecifically bound IGFBP-3.

The highly basic P3 region of IGFBP-3 serves a multitude of purposes. It contains a heparin-binding sequence (3, 5), serves as a cellular binding domain (5, 18, 23), contains a nuclear localization signal sequence (17, 21, 24), is important for ALS binding (15, 25), and binds the precursors of the serine proteases kallikrien and plasmin (8, 9, 13). On the basis of the current results, the P3 region may interact with the active proteases plasmin and thrombin, as well as with the proenzyme plasminogen. Although this region appears to be important in digestion of IGFBPs by plasmin and thrombin, several major protease cleavage sites are outside of the P3 region (6, 7). Cleavage of IGFBP-3 can take place in the absence of plasmin interaction with the P3 region; otherwise, plasmin would be totally unable to cleave IGFBP-3₄. One possibility is that major plasmin or thrombin cleavage sites exist within or at the periphery of the P3 region and that these cleavage sites are blocked by IGFBP-3 association with endothelial cells. Cleavage at other sites might be inhibited, but not blocked, by IGFBP-3 association with cells. Adherent IGFBP-3 cleaved in the P3 region would lose its ability to interact specifically with the cell surface while binding protein cleaved in other regions would remain cell associated. The P3 peptide was digested by plasmin, suggesting that there are one or more plasmin cleavage sites within the P3 region of IGFBP-3. Thrombin did not cleave the P3 peptide, which contains no Arg-Gly sequences. However, because the final residue of P3 is an Arg, followed by a Gly, IGFBP-3 may be cleaved by thrombin at the COOH terminus of the P3 region. It is important that the protease be able to detach from the BP binding site (i.e., the P3 region) to proceed on to the cleavage of further molecules. One mechanism for detachment would be cleavage of the binding protein within the P3 region. It is also possible that these serine proteases interact with the P3 region and then cleave the binding protein at other sites. Supporting this possibility is the relative insensitivity of IGFBP-4 to plasmin, the increased sensitivity of IGFBP-4₃, and the cytoprotective effect of endothelial cell binding of IGFBP-4₃ Specifically, the addition of the P3 region provides a site for plasmin interaction, increasing digestion, whereas blockage of the P3 region by cell attachment eliminates this site, and digestion is decreased. However, it should be noted that, although plasmin digestion of IGFBP-3 or IGFBP-4₃ is decreased in the presence of cells, it is not eliminated. Thus plasmin interaction with the P3 region is not essential for all of plasmin's action on IGFBP-3. Digestion is decreased but not eliminated by the exchange of the P3 region in IGFBP-3 for the P4 region (IGFBP-3₄). Cells do not protect IGFBP-3₄ from digestion. In fact, when experiments were conducted in M199, thrombin digestion of cell-associated IGFBP-3₄ was actually increased, suggesting that the small amount of cell-associated IGFBP-3₄ may be in proximity to thrombin concentrated on or near cell surface thrombin receptors.

Proteolysis of IGFBP-3 decreases its affinity for IGF, allowing the freed IGF to interact with the IGF receptor (9, 13, 20). IGFBP-3 proteolysis could thus lead to an increase in the effects of IGF or to increased transendothelial transport of IGF. Proteolysis of IGFBP-3 may also affect transport and action of IGFBP-3 fragments. IGFBP-3 containing the P3 region was detected by both fluorescence labeling and immunologic techniques in the nuclei of cells (17, 21, 24). Additionally, the physiological role of the ALS is not well defined. It has been presumed that the larger ALS molecule prevents entry of IGFBP-3 or IGFBP-3-IGF complex into subendothelial tissue. Because the P3 region is also important in ALS binding, the ternary complex may have decreased protease sensitivity. Thus the amount and positions of proteolytic cleavage of IGFBP-3 occurring at or near the endothelial surface may influence both IGF and IGFBP-3 actions.