THE INSULIN-LIKE growth factors (IGF-I and -II) are

Transcription

1 /97/$03.00/0 Vol. 138, No. 9 Endocrinology Printed in U.S.A. Copyright 1997 by The Endocrine Society Human Pregnancy Serum Contains at Least Two Distinct Proteolytic Activities with the Ability To Degrade Insulin-Like Growth Factor Binding Protein-3* PETER BANG AND PAUL J. FIELDER Pediatric Endocrinology Unit, Department of Women and Child Health, Karolinska Hospital and Institute (P.B.), Stockholm Sweden; and Department of Pharmacokinetics and Metabolism, Genentech Incorporated (P.J.F.), South San Francisco, California ABSTRACT The presence of a proteolytic activity in sera from pregnant humans and rodents capable of degrading insulin-like growth factor binding protein-3 (IGFBP-3) has been known for some time. However, the identity of this activity has remained elusive. We have attempted to purify the IGFBP-3 protease activity from pregnant human serum (PHS) using the degradation of 125 I-IGFBP-3 as a marker. Following ammonium sulfate precipitation of PHS and further enrichment of active fractions by ion-exchange, protein-a Sepharose, and size-exclusion chromatography, a protease of approximately kda was isolated and subjected to N-terminal analysis. The N-terminal sequence was consistent with plasminogen, a known fibrinolytic enzyme. To further characterize the IGFBP-3 protease activities in both PHS and nonpregnant human serum (NHS), aliquots of serum were first enriched by polyethylene glycol-precipitation and subjected to size-exclusion chromatography. The size-separated fractions were then incubated with 125 I-IGFBP-3, and proteolytic activity was measured. PHS contained two separate proteases ( 150 kda and Received March 13, Address all correspondence and requests for reprints to: Peter Bang, Department of Woman and Child Health, Pediatric Endocrinology Unit, Karolinska Hospital, Stockholm, Sweden. peter.bang@netg.se. * This work was supported by grants from the Swedish Medical Research council (B96 19X A), the Foundation for Studies Without Animals, Swedish Foundation for Childrens Ward. It was presented in part at the 3rd International Meeting on Insulin-Like Growth Factor Binding Proteins, Tübingen, Germany, October, 1995 and in the proceedings from this meeting: Bang P Serum proteolysis of IGFBP-3. Prog Growth Factor Res 6: kda), whereas NHS contained only one (70 90 kda) that had a inhibitor profile similar to plasmin. However, inhibitors of plasmin had no effect on the activity of the 150-kDa protease. Plasminogen activators (PAs) greatly increased the activity of the 70- to 90-kDa protease, but had little effect on the 150-kDa protease activity. Addition of PAs greatly increased the ability of NHS to proteolyze IGFBP-3. In contrast, the ability of plasminogen-depleted plasma to degrade 125 I-IGFBP-3 was not affected by the addition of PAs. Both urokinase and tissue-type PA had the ability to proteolyze IGFBP-3 and were, in contrast to the 150-kDa protease activity, inhibited by the specific PA inhibitor D-PHE-PRO-ARG chloromethyl ketone. The present data suggest that sera has the ability to proteolyze IGFBP-3, and that this ability, as demonstrated by NHS, can be regulated by protease inhibitors and PAs. In addition, PHS does indeed contain an unique IGFBP-3 protease activity that is not present in NHS, and its identity is unknown at this time. (Endocrinology 138: , 1997) THE INSULIN-LIKE growth factors (IGF-I and -II) are potent endocrine and paracrine growth factors with important implications for growth and differentiation (1). The IGFs are present in high concentrations in circulation, predominantly in a less bioavailable form bound to IGF binding proteins (IGFBP-1 through -6) (2). Approximately 90 95% of circulating IGFs are bound in a ternary complex, consisting of IGF peptide, IGFBP-3, and acid labile subunit (ALS) (3), which in humans is limited to the intravascular space. A major question in IGF physiology is how IGFs gain access to IGF-receptors in the peripheral target tissues, and how the preceding dissociation of IGFs from the stable ternary complex is regulated. Recent reports have suggested that proteases may be involved in the release of IGFs from IGFBPs in serum (4 10). Proteolysis of IGFBP-3 was first demonstrated in pregnancy serum from humans and rodents (4, 5, 6, 11). This proteolytic processing of IGFBP-3 results in reduced affinity of IGFBP-3 for IGFs and, in particular, for radiolabeled IGF-I (12, 13) and has been demonstrated to increase IGF bioavailability in serum (8, 9). The increased IGFBP-3 protease activity in maternal serum may only reflect the activation of proteases within a specific organ and subsequent dilution of the activity into the general circulation. Thus, the resulting release of IGF from circulating stores in the ternary complex may be more pronounced within that organ. If the major production site of the pregnancy IGFBP-3 protease activity is the placenta, as has been postulated (14), this activity may help support fetal growth by increasing maternal IGF bioavailability to the placenta. In human fetal serum, which does not display increased IGFBP-3 protease activity, an increased IGF-I bioavailability is accomplished by low fetal ALS expression resulting in lack of the ternary complex (15, 16). The identity of the pregnancy-specific IGFBP-3 protease activity in human and rodent serum is not known, although it has been suggested to be divalent-cation dependent and inhibited by serine protease inhibitors (17, 18). In sera from pregnant rats, tissue inhibitors of matrix metallo-proteinases (TIMP-1) have recently been demonstrated to inhibit the proteolysis of IGFBP-3 (19). In the present study we partially purified the pregnancy-associated IGFBP-3 proteases from pregnant human serum (PHS). Two distinct protease activities were identified: one activity of approximate kda,

2 IGFBP-3 PROTEOLYTIC ACTIVITIES IN PREGNANCY SERUM 3913 which was present in both pregnant and nonpregnant human serum (NHS), and was further demonstrated to be plasminogen. The other activity, present only in PHS, was distinct from the 70- to 90-kDa protease because it was of larger molecular mass ( 150 kda), was unaffected by plasminogen/plasmin activators (PAs) and inhibitors, as well as serine protease inhibitors, but was inhibited by EDTA and zinc. We suggest that at least two protease activities may be involved in the regulation of IGF bioavailability in human serum, one of which is present during pregnancy. Materials and Methods PHS was obtained from women during the last trimester of a normal pregnancy. NHS was obtained from apparently healthy volunteers and used for comparison. PHS was collected under an exemption granted by the Committee on the Use of Human Subjects in Medical Research, Stanford University and from paid subjects following informed consent (Genentech, Inc. South San Francisco, CA). Serum was aliquoted and stored at 70 C until processing. Plasminogen, plasmin, streptokinase, urokinase (UK), aprotinin, and -2-antiplasmin, and D-PHE-PRO-ARG chloromethyl ketone (PPACK) were purchased from Calbiochem (La Jolla, CA). Plasminogen-depleted plasma (PDP) was purchased from American Diagnostic (Greenwich, CT). IGF-I, IGFBP-3, and tissue PA (tpa) are products of Genentech, Inc. IGFBP-3 protease assay Proteolytic activity was measured as the ability of a sample to degrade radiolabeled 125 I-IGFBP-3 (20). Serum (2 3 l), column fractions (10 20 l), or pure proteases were incubated with 30,000 cpm of 125 I-human glycosylated IGFBP-3 (DSL, Webster, TX) for 5hat37C( protease inhibitors) in a total volume of 50 l20mmhepes, ph 7.2, 5 mm Ca 2, and 0.1% BSA. Reactions were terminated by addition of SDS sample buffer, and the reaction mixture was separated by SDS-PAGE (12.5% or 15% gels) at 45V overnight as previously described (15). Gels were dried and exposed to x-ray films for 1 4 days at 70 C. In addition, to check for possible substrate competition or differences in the susceptibility for proteolytic cleavage introduced by radiolabeling of IGFBP-3, selected samples were reassayed using unlabeled IGFBP-3 as a substrate and Western immunoblotting for detection of proteolytic products. Samples were incubated with 50 ng IGFBP-3 in a total volume of 50 l 20mm HEPES, ph 7.2, with 1 mm CaCl 2, 0.1% BSA for 1 5 h at 37 C. The reaction were terminated by the addition of SDS sample buffer, and the mixtures were processed by SDS-PAGE (12.5% or 15% gels), electroblotted, and immunoblotted with an anti-igfbp-3 antibody using enhanced chemiluminescence technique as previously described (18). Protease purification PHS was pooled, and 10 ml was subjected to stepwise ammonium sulphate precipitation. Preliminary results using increments of 5% (vol/ vol) 3.8 m ammonium sulfate, demonstrated that active material precipitated between 30 50% (vol/vol) 3.8 m ammonium sulfate. The 30 50% precipitate was resuspended in 20 mm Tris, ph 8.0, and dialyzed (10 kda cutoff) overnight against 20 liter 20 mm Tris, ph 8.0, at 4 C. The sample was applied to a 6-ml Resource Q anion-exchange column (Pharmacia, Stockholm, Sweden) developed with 20 mm Tris, ph 8.0. Adsorbed proteins were eluted by a linear gradient of m of NaCl. Active fractions, as determined by the IGFBP-3 protease assay, typically eluting at m NaCl were pooled. In preliminary experiments, contamination with IgG was present at this step. To remove the IgG, the pooled fractions were adjusted to ph 7.2 with 100 mm HEPES, ph 7.2, and run over a protein-a Sepharose column (12 ml bed volume) developed with 50 mm HEPES, ph 7.2. The IgG-free eluate was concentrated by Centricon 30 (Amicon, Danvers, MA) ultrafiltration and processed by size-exclusion chromatography on a Superdex 200 HR column (Pharmacia) developed with a 50 mm phosphate buffer, ph 7.4, 150 mm NaCl. Fractions were assayed for IGFBP-3 protease activity. Active fractions were separated by SDS-PAGE (15% gels), electrotransfered to polyvinylidene difluoride membranes and stained for total protein. A band of approximately kda (on nonreducing gels) was then subjected to N-terminal amino acid analysis as previously described (21). Characterization of IGFBP-3 proteases in PHS To determine the molecular weight of the IGFBP-3 protease activities in PHS and NHS, 50% polyethylene glycol 8000 was added to 1.5 ml of serum [final concentration 12% (wt/vol)] to precipitate the proteolytic activity. The pellet was then solubilized in 1.5 ml 0.1 m NaHCO 3 and concentrated to approximately 0.4 ml using a Centricon 50, and the buffer was exchanged by successive addition of a total of 1.6 ml PBS, 0.01% Tween 80 and reconcentrated. The samples were then applied to a Superdex 200 HR size-exclusion column as described above. Fractions of 1 ml were collected, and 75- l aliquots assayed for IGFBP-3 protease activity. The fractions containing IGFBP-3 protease activity were divided into two pools based on the molecular mass kda and 150 kda. These two pools were further characterized by incubation with 125 I- IGFBP-3 with or without various protease inhibitors and/or PAs. These samples were analyzed as described in the IGFBP-3 protease assay. Results Purification of IGFBP-3 protease activity in PHS After enrichment of the IGFBP-3 protease activity in PHS by ammonium sulfate precipitation followed by anion exchange chromatography, the major protein component of the active fractions was IgG. Protein-A Sepharose chromatography was used to exclude IgGs from the IGFBP-3 protease activity. Subsequent size-exclusion chromatography demonstrated that the major IGFBP-3 proteolytic activity eluted at the approximate size of kda. There was good agreement between the results from IGFBP-3 protease assays using either 125 I-IGFBP-3 or unlabeled IGFBP-3 as a substrate (data not shown). SDS-PAGE under reducing conditions and silver staining demonstrated that the active material consisted of two protein components; a major 90-kDa band and a minor 70-kDa band (doublet bands of approximately 70 kda on nonreducing gels) (Fig. 1). N-terminal amino acid analysis of this material from a polyvinylidene difluoride membrane demonstrated a sequence of EPLDDYVNTQG, which was consistent with the N-terminal sequence of human plasminogen. FIG. 1. Purification of an IGFBP-3 proteolytic activity from PHS. Active samples were subjected to SDS-PAGE under both reducing and nonreducing conditions followed by silver staining. Approximate molecular weights as determined by SDS-PAGE are presented on left axis.

3 3914 IGFBP-3 PROTEOLYTIC ACTIVITIES IN PREGNANCY SERUM Characterization of two IGFBP-3 proteases in PHS FIG. 2. Presence of two different molecular mass IGFBP-3 protease activities in PHS. PHS and NHS were separated by size-exclusion chromatography, incubated with 125I-IGFBP-3, and subjected to SDSPAGE followed by autoradiography. Approximate molecular masses as determined by SDS-PAGE are presented on left axis and by sizeexclusion chromatography are on top of figure. pattern with two major fragments of approximately 19 and 15 kda and two minor fragments of approximately 28 and 3 kda, respectively. It is possible, though, that the 3-kDa fragment results from further degradation of larger fragments due to higher activity of the 70- to 90-kDa protease. Addition of 5 mm EDTA or 0.5 mm zinc21 completely inhibited the activity of the.150-kda protease, but EDTA had no effect and zinc21 only partially inhibited the activity of the 70-to 90-kDa pool (Fig. 3). Concentrations as low as 0.1 mm zinc21 completely inhibited the.150-kda protease activity (data not shown). Conversely, addition of 500 mg/ml aprotinin or 20 mg/ml a-2-antiplasmin had no effect on the activity of the.150-kda pool, whereas, these inhibitors completely blocked the ability of the 70- to 90-kDa pool to proteolyze 125 I-IGFBP-3. Another difference between the two protease activities was their response to addition of tpa. Addition of 1 mg/ml tpa to the.150-kda pool (A 1 tpa) had no enhancing effect on its activity, whereas, addition of tpa to the 70- to 90-kDa pool (B 1 tpa) greatly enhanced its ability to proteolyze 125I-IGFBP-3 (Fig. 3). These data strongly suggest that PHS contains at least two different molecular mass proteases capable of degrading 125 I-IGFBP-3, which have distinct inhibitor and activator profiles. To investigate the ability of plasminogen to act as a potential serum pool of IGFBP-3 proteolytic activity, we incubated PHS, NHS, and PDP with 125I-IGFBP-3 in the absence or presence of various PAs. As can be seen in Fig. 4, addition of 1 mg/ml tpa, UK, or streptokinase to PHS slightly in- FIG. 3. Characterization of two IGFBP-3 proteolytic activities present in PHS. Pooled fractions from each of two activities from PHS were then incubated with 125I-IGFBP-3 in absence or presence of protease inhibitors (5 mm EDTA, 0.5 mm zinc21, 500 mg/ml aprotinin, or 20 mg/ml a-2-antiplasmin), or tpa (1 mg/ml). A 1 tpa, Represents 150-kDa pool incubated with tpa; B 1 tpa, represents 70- to 90-kDa pool. Samples were then subjected to SDS-PAGE, followed by autoradiography. Approximate molecular masses as determined by SDSPAGE are presented on left axis. The purification steps resulted in the isolation of a 70- to 90-kDa protein with a N-terminal amino acid sequence consistent with that of plasminogen. As such, we characterized PHS and NHS to investigate whether the IGFBP-3 protease activities were plasmin-like. When PHS was separated by size-exclusion chromatography, two IGFBP-3 protease activities of $150 and kda could be demonstrated (Fig. 2a). In contrast to PHS, NHS contained only the IGFBP-3 protease activity of kda consistent with the molecular mass of plasminogen (Fig. 2b). Pooled fractions from each of the two activities from PHS were then incubated with 125IIGFBP-3 in the absence or presence of protease inhibitors and PAs, respectively. Figure 3 demonstrates that the.150-kda pool degraded 125I-IGFBP-3 into three major fragments of approximately 29, 19, and 15 kda, respectively. The 70- to 90-kDa pool degraded 125I-IGFBP-3 into a slightly different Endo 1997 Vol 138 No 9

4 IGFBP-3 PROTEOLYTIC ACTIVITIES IN PREGNANCY SERUM creased its ability to degrade 125I-IGFBP-3. This increase is likely due to activation of inactive plasminogen to plasmin. The degradation pattern of PHS was also slightly altered by addition of PAs; this is evidenced by the increase in intensity of a lower molecular mass band of approximately 10 kda. Addition of PAs to NHS greatly enhanced its ability to degrade 125I-IGFBP-3 (Fig. 4). However, addition of PAs to PDP had no significant effect on its ability to degrade 125IIGFBP-3, consistent with the absence of plasminogen. Some minor degradation bands were present in the PDP 1 tpa sample, but these were likely due to the ability of tpa itself to proteolyze IGFBP-3. This view was supported by the pattern of fragments produced by addition of tpa to PDP, in particular with the two fragments slightly above and below 30 kda, similar to the fragmentation pattern produced by tpa itself (see below). The ability of the 150-kDa protease, tpa, and UK to proteolyze 125I-IGFBP-3 is demonstrated in Fig. 5. In this experiment, 125I-IGFBP-3 was incubated with the.150-kda pool (75 ml), tpa (10 mg/ml), and UK (10 mg/ml) in the presence or absence of increasing concentrations of the PA inhibitor PPACK. As shown in Fig. 5, all proteases were capable of degrading 125I-IGFBP-3. The proteolytic fragments produced by the.150-kda protease in PHS differed from those produced by tpa and UK, with one band of approximately 30 kda as opposed to two bands slightly above and below 30 kda, respectively. However, whereas both tpa and UK were inhibited by PPACK, the.150-kda protease was not. We also incubated unfractionated PHS with PPACK and a-2antiplasmin at concentrations that inhibit IGFBP-3 proteolysis by plasmin, tpa, UK, and the 70- to 90-kDa protease. Neither PPACK nor a-2-antiplasmin decreased IGFBP-3 proteolytic activity of PHS (data not shown). NHS had no significant activity without addition of PAs. These data indicate that both tpa and UK can act as IGFBP-3 proteases alone, and that the 150-kDa protease activity does not appear to be FIG. 5. Effects of an inhibitor of PAs on ability of.150-kda protease, tpa, and UK to degrade 125I-IGFBP I-IGFBP-3 was incubated with.150-kda pool (0.075 ml), tpa (10 mg/ml), UK (10 mg/ml), or alone (BP-3) in presence or absence of increasing concentrations of PA inhibitor PPACK (0 100 mg/ml). Samples were then subjected to SDS-PAGE, followed by autoradiography. Approximate molecular masses as determined by SDS-PAGE are presented on left axis. related to the PAs based on the inability of PPACK to alter its activity. Discussion Numerous studies have suggested that the proteolysis of IGFBPs is involved in the regulation of IGF availability at the cellular level, thereby modulating cell growth and differentiation (22, for review see Refs. 23, 24). These studies have demonstrated an intimate regulation of cell growth involving coordinated changes in IGF synthesis and release, differential regulation of several IGFBPs, and modulation of IGFBP actions by cell-produced proteases. However, the first reports of IGFBP-protease activities focused on serum from pregnant women and rodents (4 6). In the present study, we demonstrated the presence of two proteases in PHS that can degrade intact glycosylated 125IIGFBP-3 into smaller fragments. We have purified one of these protease activities and demonstrated the major protein component to be identical with the known fibrinolytic enzyme, plasminogen. Moreover, plasminogen IGFBP-3 proteolytic activity could also be demonstrated in NHS. In contrast to NHS, the higher molecular mass (.150 kda) protease activity was present only in PHS. We propose that these two activities represent distinct proteolytic activities based on the following findings. Size- exclusion chromatography revealed that the two proteases are kda and.150 kda in size, respectively. The 70- to 90-kDa protease activity is inhibited by the serine protease inhibitors aprotinin and a-2antiplasmin, but is unaffected by EDTA. This was in contrast to the.150-kda protease activity, which demonstrated cation dependency, because it was inhibited by EDTA and was unaffected by the above protease inhibitors. Zinc completely FIG. 4. Ability of PAs to induce IGFBP-3 proteolytic activity in pregnant and nonpregnant sera. PHS, NHS, and PDP were incubated with 125 I-IGFBP-3 in presence of various plasminogen activators (1 mg/ml tpa or 1 mg/ml streptokinase). Samples were then subjected to SDSPAGE, followed by autoradiography. Approximate molecular masses as determined by SDS-PAGE are presented on left axis. 3915

5 3916 IGFBP-3 PROTEOLYTIC ACTIVITIES IN PREGNANCY SERUM Endo 1997 Vol 138 No 9 inhibited the activity of the 150-kDa protease, but only partly inhibited the 70- to 90-kDa protease. Addition of PAs markedly increased the activity of the 70- to 90-kDa protease, but did not alter the activity of 150-kDa protease. Most nonpregnant serum and plasma samples lack the ability to proteolyze IGFBP-3 because of the presence of plasmin inhibitors and low levels of circulating free tpa. It is not clear how our purification protocol and/or size-exclusion chromatography activated the fibrinolytic system to degrade IGFBP-3 in vitro. It is possible that specific serum inhibitors of the fibrinolytic cascade were separated from the protease during the purification, or that the physical manipulation of the serum resulted in the activation of plasmin. What is the physiological role of the fibrinolytic system in acting as an intravascular IGFBP-3 protease resulting in increased availability of circulating IGFs? Clearly, PAs can induce IGFBP-3 proteolytic activity in NHS, as has been described in this paper. Campbell et al. (22) previously showed that plasmin can act as an IGFBP-3 protease at the cellular level, where it was converted by osteosarcoma cells from endogenous plasminogen by UK or tpa residing at the cell membrane (25). Thus, increases in plasmin concentrations can enhance cell growth through the release of IGFs from IGFBPs (22, 26). We also demonstrated that PAs, including tpa and UK, have IGFBP-3 proteolytic activity, and that they can act more potently by converting endogenous serum plasminogen to plasmin. However, in vivo inhibitors such as plasminogen activator inhibitor, -2-antiplasmin, and -2-macroglobulin can act at each level of the proteolytic cascade to prevent these proteolytic processes (27). The ubiquitous presence of these inhibitors in serum likely prevents us from detecting an overall increase in serum IGFBP-3 protease activity in situations where local activation of the fibrinolytic system has taken place. As further support, we were not able to measure increased arterial-venous difference in IGFBP-3 protease activity in blood from exercising human muscle, although local tpa expression was elevated (Bang, P., V. Berg, B. Saltin, and K. Hall, unpublished observations). Other relevant physiological actions of the plasmin system may include IGFBP-3 proteolysis and IGF release at local sites of tissue injury or blood clot formation, thereby supporting mitogenesis of regenerating tissue. In fact, the presence of a 30-kDa IGFBP-3 fragment, which constitutes a significant proportion of circulating IGFBP-3 in normal healthy adults (18), may suggest that a constant low level of intravascular proteolysis of IGFBP-3 may occur in the absence of significant circulating IGFBP-3 proteolytic activity. However, it cannot be excluded that these IGFBP-3 fragments at least partially originate from extravascular proteolysis (26). In contrast to the fibrinolytic system, the distinct 150-kDa protease in term PHS is capable of exerting generalized IGFBP-3 proteolytic effects in serum. Whether this is due to a more widespread release of the protease in the circulation and/or less efficient inhibition in serum of an organ-specific activity is not known. Early mixing experiments have demonstrated that addition of NHS to PHS did not inhibit the IGFBP-3 protease (4, 6), which suggests that the latter case is true. If the protease activity is actually produced in the placenta, as it has been postulated, the protease that can be measured in serum may only be a fraction of the activity present in the maternal circulation at the maternal-placental interface. The present study demonstrates a distinct 150-kDa protease activity that is unique to PHS, which could be completely inhibited by EDTA and zinc. Inhibition of matrix metalloproteinase (MMP)-1 and MMP-3 in unfractionated rat pregnancy serum, either by the addition of TIMP-1 or by specific antibodies to these MMPs, have demonstrated a marked reduction in the IGFBP-3 proteolytic activity (19). Therefore these MMPs have been suggested to be responsible for the IGFBP-3 proteolytic activity in rat pregnancy serum. Because the molecular mass of MMP-1 and MMP-3 in human serum is in the range of kda, these proteases are not likely to explain the activity of the 150-kDa protease in human pregnancy serum. In addition, MMPs are generally zinc dependent, but the 150-kDa protease was completely inhibited by zinc. We are currently performing experiments with human pregnancy serum using specific inhibitors of MMPs such as TIMP-1 and specific anti-mmp antibodies to find further support for this view. However, eventually isolation and characterization of the human protease is needed to fully resolve these questions. It is well accepted that proteolysis of IGFBP-3 does not result in a disruption of the ternary complex between IGF, IGFBP-3, and ALS (5, 12, 18). However, Laserre et al. (7) demonstrated that IGF-I had a 10-fold and IGF-II a 2-fold decreased affinity for IGFBP-3 isolated from PHS compared with NHS. When compared with NHS, IGF-I and -II dissociated 10 and 6 times faster from IGFBP-3 in PHS, respectively, and the proportion of IGF-I in the free form was nearly 3 times greater in PHS than in NHS. These data suggest that although in PHS most IGF-I and -II is still present in the 150-kDa complex, its bioavailability is likely increased. Thus, the presence of at least two distinct proteases in human serum may prove to play an important role in the regulation of the endocrine actions of IGF-I. The challenge now is to determine the identity of the pregnancy-specific protease activity and possible other proteases, as well as to study their regulation in relation to physiological and pathophysiological states in humans. Acknowledgments The authors thank Dr. Linda Giudice for collection of PHS, Dr. David Riefsnyder for supplying the recombinant human IGFBP-3, Bill Henzel for the N-terminal analysis, and Yvonne Lin for reviewing the manuscript. References 1. Bang P, Hall K 1992 Insulin-like growth factors as endocrine and paracrine hormones. In: Schofield PN (ed) The Insulin-Like Growth Factors: Structure and Biological Functions. Oxford University Press, New York, pp Jones JI, Clemmons DR 1995 Insulin-like growth factors and their binding proteins. Endocr Rev 16: Baxter RC, Martin JL, Beniac VA 1989 High molecular weight insulin-like growth factor complex; purification and properties of the acid-labile subunit from human serum. J Biol Chem 264: Giudice LC, Farrell EM, Pham H, Lamson G, Rosenfeld RG 1990 Insulin-like growth factor binding proteins in maternal serum throughout gestation and in the puerperium: Effects of a pregnancy-associated serum protease activity. J Clin Endocrinol Metab 71: Hossenlopp P, Sergovia B, Lassarre C, Roghani M, Bredon M, Binoux M 1990 Evidence of enzymatic degradation of insulin-like growth factor-binding pro-

6 IGFBP-3 PROTEOLYTIC ACTIVITIES IN PREGNANCY SERUM 3917 teins in the 150K complex during pregnancy. J Clin Endocrinol Metab 71: Fielder PJ, Thordarson G, Talamantes F, Rosenfeld RG 1990 Characterization of insulin-like growth factor binding proteins (IGFBPs) during gestation in mice: effects of hypophysectomy and an IGFBP-specific serum protease activity. Endocrinology 127: Laserre C, Binoux M 1994 Insulin-like Growth Factor Binding Protein-3 is functionally altered in pregnancy plasma. Endocrinology 134: Blat C, Villaudy J, Binoux M 1994 In vivo proteolysis of serum insulin-like growth factor (IGF) binding protein-3 results in increased availability of IGF to target cells. J Clin Invest 93: Hasagawa Y, Takada M, Hasagawa T, Tsuchiya Y 1994 Clinical utility of newly developed free insulin-like growth factor-i by IRMA. Clin Pediatr Endocrinol 3[Suppl 5]: (Abstract) 10. Binoux M, Lalou C, Lassarre C, Blat C, Hossenloop P 1994 Limited proteolysis of insulin-like growth factor binding protein-3 (IGFBP-3): A physiological mechanism in the regulation of IGF bioavailability. In: LeRoith D, Raizada MK (eds) Current Directions in Insulin-Like Growth Factor Research. Plenum Press, New York, pp Davenport ML, Pucilowska JB, Clemmons DR, Lundblad R, Spencer JS, Underwood LE 1992 Tissue-specific expression of insulin-like growth factor binding protein-3 protease activity during rat pregnancy. Endocrinology 130: Suikkari A-M, Baxter RC 1992 Insulin-like growth factor-binding protein-3 is functionally normal in pregnancy serum. J Clin Endocrinol Metab 74: Lassarre C, Binoux M 1994 Insulin-like growth factor binding protein-3 is functionally altered in pregnancy plasma. Endocrinology 134: Giudice, LC, Irwin JC, Dsupin BA, de las Fuentes L, Jin IH, Vu TH, Hoffman AR 1994 Insulin-like growth factors (IGFs), IGF binding proteins (IGFBPs) and IGFBP protease in human uterine endometrium: their potential relevance to endometrial cyclic function and maternal-embryonic interactions. In: Baxter RC, Gluckman PD, Rosenfeld RG (eds) The Insulin-Like Growth Factors and Their Regulatory Proteins. Excerpta Medica, Amsterdam, The Netherlands, pp Bang P, Stangenberg M, Westgren M, Rosenfeld RG 1994 Decreased ternary complex formation and predominance of a 29 kda IGFBP-3 fragment in human fetal serum. Growth Regul 4: Lewitt MS, Scott FP, Clarke NM, Baxter RC 1995 Developmental regulation of circulating insulin-like growth factor binding proteins in normal pregnancies and in pre-eclampsia. Prog Growth Factor Res 6: Davies SC, Holly JM, Coulson VJ, Catterhill AM, Abdulla AF, Whittaker DG, Chard T, Wass JA 1991 The presence of cation-dependent proteases for insulin-like growth factor binding proteins does not alter the size distribution of insulin-like growth factors in pregnancy. Clin Endocrinol (Oxf) 34: Bang P, Brismar K, Rosenfeld RG 1994 Increased proteolysis of insulin-like growth factor binding protein-3 (IGFBP-3) in noninsulin-dependent diabetes mellitus serum, with elevation of a 29 kda glycosylated IGFBP-3 fragment contained in the approximately 130- to 150-kDa ternary complex. J Clin Endocrinol Metab 78: Fowlkes JL, Suzuki K, Nagase H, Thrailkill KM 1994 Proteolysis of insulinlike growth factor binding protein-3 during rat pregnancy: A role for matrix metalloproteinases. Endocrinology 135: Lamson G, Giudice LC, Rosenfeld RG 1991 A simple assay for proteolysis of IGFBP-3. J Clin Endocrinol Metab 72: Fielder PJ, Rosenfeld RG, Graves HCB, Grandbois K, Maack CA, Sawamura S, Sommer A, Cohen P 1995 Biochemical analysis of prostate specific antigen (PSA)-proteolyzed insulin-like growth factor binding protein-3. Growth Regul 1: Campbell PG, Novak JF, Yanosick TB, McMaster JH 1992 Involvement of the plasmin system in dissociation of the insulin-like growth factor-binding protein complex. Endocrinology 130: Clemmons DR, Nam TJ, Busby WH, Parker A 1994 Modification of IGF action by insulin-like growth factor binding protein-5. In: Baxter RC, Gluckman PD, Rosenfeld RG (eds). The Insulin-Like Growth Factors and Their Regulatory Proteins. Excerpta Medica, Amsterdam, The Netherlands, pp Conover CA 1995 Insulin-like growth factor binding protein proteolysis in bone cell models. Prog Growth Factor Res 6: Campbell PG, Novak JF, Wines K, Walton PE 1993 Localization of plasmin activity on osteosarcoma cells: cell surface proteolysis of insulin-like growth factor binding proteins. Growth Regulation 3: Lalou C, Silve C, Rosato R, Segovia B, Binoux M 1994 Interactions between insulin-like growth factor-i (IGF-I) and the system of plasminogen activators and their inhibitors in the control of IGF-binding protein-3 production and proteolysis in human osteosarcoma cells. Endocrinology 135: Sprengers ED, Kluft C 1987 Plasminogen activator inhibitors. Blood 69: