Roles of the Heparin and Low Density Lipid Receptor-related Protein-binding Sites of Protease Nexin 1 (PN1) in Urokinase-PN1 Complex Catabolism THE PN1 HEPARIN-BINDING SITE MEDIATES COMPLEX RETENTION AND DEGRADATION BUT NOT CELL SURFACE BINDING OR INTERNALIZATION*

We have previously described thrombin (Th)-protease nexin 1 (PN1) inhibitory complex binding to cell surface heparins and subsequent low density lipid receptor-re-lated protein (LRP)-mediated internalization. Our present studies examine the catabolism of urinary plasminogen activator (uPA)-PN1 inhibitory complexes, which, unlike Th z PN1 complexes, bind almost exclusively through the uPA receptor. In addition, the binding site in PN1 required for the LRP-mediated internalization of Th z PN1 complexes is not required for the LRP-mediated internalization of uPA z PN1 complexes. Thus, the protease moiety of the complex partially determines the mechanistic route of entry. Because cell surface heparins are only minimally involved in the binding and internalization of uPA z PN1 complexes, we then predicted that complexes between uPA and the heparin binding-deficient PN1 variant, PN1(K7E), should be ca-tabolized at the same rate as complexes formed with native PN1. Surprisingly, the uPA z PN1(K7E) complexes were degraded at only a fraction of the rate of native complexes. Internalization studies revealed that both uPA z PN1(K7E) and native uPA z PN1 complexes were initially 3 once with 1 ml of ice-cold 50/50 acetone/H 2 O. Following a with 100% acetone and air the precipitates were resuspended in 75 m l of SDS-PAGE sample buffer and analyzed by SDS-PAGE on a 10% polyacrylamide gel. pH-dependent Binding Studies— To evaluate the binding of uPA z PN1 complexes to the surface of HF cells, a series of binding buffers were prepared at different pH levels using 20 m M Hepes as the starting buffer and titrating it with either HCl or NaOH. BSA was added to a final concentration of 0.1% prior to titration. Complexes were diluted into each of the buffers and incubated with the cells at 4 °C for 2 h. Binding and heparin dissociation of

Human protease nexin 1 (PN1) 1 is a 43-kDa member of the SERPIN family of protease inhibitors (1). PN1 is a potent inhibitor of thrombin (Th) and urinary plasminogen activator (uPA) (2) and is a less potent but still effective inhibitor of plasmin and trypsin. Although the precise molecular nature of the PN1-protease inhibitory complexes under native conditions remains controversial, it is most likely a 1:1 covalently linked stoichiometric complex or is at least very stable under physiological conditions (3)(4)(5). Complexes formed between PN1 and all of the proteases that it inhibits are stable to SDS-PAGE after boiling and reduction (6). Interestingly, in the presence of heparin, PN1 is a faster thrombin inhibitor than heparinactivated antithrombin III (ATIII), the most physiologically important inhibitor of Th in plasma (2). However, because of the physiological distribution of PN1, which is restricted primarily to tissues with the exception of a low abundance form in platelets (7), PN1 probably inactivates extravasated thrombin and does not participate directly in the fibrinolytic system. Because of the limited range of specificity of PN1 and the nature of the proteases it inhibits, PN1 has been implicated in neurite outgrowth (8), the protection of neuronal cells from proteolytic damage and thrombin-induced apoptosis in brain injury (9 -11), and the maintenance of neuromuscular junctions (11).
An integral aspect of PN1 physiology is the catabolism of PN1-protease inhibitory complexes, because the only way to ensure the removal of protease activity is through endocytosis and intracellular degradation. In addition, the endocytosis of inhibitory complexes may act as a feedback signaling mechanism to regulate inhibitor synthesis and secretion (12). Recent progress has revealed that the low density lipoprotein receptorrelated protein (LRP) is responsible for the cellular internalization of PN1-protease (12)(13)(14)(15). In the case of Th⅐PN1 complexes, LRP binding is regulated by the exposure of a cryptic binding site that is rendered LRP binding-competent only when PN1 is in complex with a protease (16). Using a synthetic peptide library, this binding site has been narrowed to a 12amino acid sequence, 47 PHDNIVISPHGI 58 (13). This is a predicted transition sequence between the A helix and strand 6 of sheet B in PN1, based on the projected structure of PN1 using ␣-1-antitrypsin coordinates as a model (1). More recent studies using an antibody directed against this sequence, as well as site-directed variants of PN1 with point, isosteric amino acid substitutions in this sequence, have confirmed this region as a structural determinant required for the LRP-mediated internalization of Th⅐PN1 complexes (17). Interestingly, the role of this sequence is limited to complex internalization by the LRP. The majority of the cellular binding of Th⅐PN1 complexes, which plays a substantial role in the overall rate of catabolism of the complexes, is mediated by the heparin-binding site in PN1 and its interactions with cell surface heparins or heparan sulfate proteoglycans (14).
The catabolism of uPA⅐PN1 complexes is also mediated by the LRP but reportedly proceeds through a pathway that utilizes the uPA receptor (uPAR) (15,18). The apparent mechanism is a sequential series of events that is initiated by the binding of uPA⅐PN1 complexes to uPAR, followed by LRPmediated internalization as a ternary complex (18). Because the association constant of uPA⅐PN1 complexes for the uPAR is much greater than the association constant of the uPA⅐PN1 complexes for heparin, this raises the question of the role of the heparin-binding site in PN1 in the catabolism of uPA⅐PN1 complexes. It also raises the question of the role of the LRPbinding site in PN1 in the catabolism of uPA⅐PN1 complexes. These questions have particular biological relevance for three reasons. First, the catabolic route of PN1-protease complexes may be determined by the identity of the protease present in the inhibitory complex. If indeed two different pathways exist, the pathway used may affect the cellular response to these inhibitory complexes and may augment the up-or down-regulation of inhibitor synthesis and secretion (12). Second, uPAR occupied by uPA has been shown to localize to the leading edge of migrating smooth muscle cells in a wound-healing model. The uPA-uPAR complexes at the cell surface activate the JAK/ STAT signaling pathway, and removal of the complexes by LRP-mediated internalization deactivates the pathway (19). Thus, PN1 has the potential to regulate this signaling pathway, depending on the entry route of uPA⅐PN1 complexes into the cell. Third, little is known about the clearance mechanism of SERPIN-protease inhibitory complexes in general. Based on several reports from one group, it was initially believed that all SERPIN-enzyme complexes were cleared by a common receptor, referred to as the SERPIN-enzyme complex receptor or SEC receptor (20,21). This receptor was thought to recognize a common structural motif in the carboxyl-terminal region of all of the SERPINs. This original hypothesis has not held true (22), and there are now several published reports showing that the LRP acts as the endocytosis receptor for nearly all of the SERPINs examined (13-15, 18, 20, 23, 24). It has also been determined that the structural site responsible for binding to LRP does not reside within the carboxyl-terminal domain (13,17).
In the present studies we have investigated the catabolic pathways of Th⅐PN1 complexes and uPA⅐PN1 complexes using a polyclonal antibody raised against the LRP-binding site in PN1 (13,17) and a heparin binding-deficient variant of PN1 that forms inhibitory complexes with Th and uPA but is unable to bind to cell surface heparins (14). The results of these studies demonstrate that the catabolism of uPA⅐PN1 complexes and Th⅐PN1 complexes do indeed proceed through two different pathways that converge at the point of the LRP. The cell surface binding of uPA⅐PN1 complexes is heparin-independent and internalization does not involve the LRP-binding site in PN1. In contrast, the cell surface binding of Th⅐PN1 complexes is mediated by heparins, and the internalization of Th⅐PN1 complexes is dependent on the LRP-binding site in PN1. We have also made the unexpected observation that although the heparin-binding site in PN1 is not involved in the cell surfacebinding of uPA⅐PN1 complexes, it is apparently required for the endosomal retention of uPA⅐PN1 complexes after endocytosis.

EXPERIMENTAL PROCEDURES
Materials-Cell culture media and reagents were purchased from Irvine Scientific and JRH Scientific. Cell culture plastics were from Corning. Thrombin, 3,000 NIH units/mg, and high molecular weight urokinase (uPA), 80,000 IU/mg, were purchased from Calbiochem. Na 125 Iodine was from Amersham Pharmacia Biotech. Porcine mucosal heparin was from Calbiochem. Soybean trypsin inhibitor and monensin were from Sigma. Protein G-Sepharose beads, high trap heparin-Sepharose and Cibacron blue-Sepharose were all from Amersham Pharmacia Biotech. All other common reagents were from Sigma, Calbiochem, or Irvine Scientific. Peptide synthesis (AE78 and Pro 47 -Ile 58 -Cys) and antiserum generation in rabbits, using Pro 47 -Ile 58 -Cys coupled to ovalbumin, have been previously described (25). The LRP agonist, receptor-associated protein (RAP), was expressed as a fusion protein (RAP-GST) consisting of an amino-terminal glutathione S-transferase sequence followed by the rat RAP sequence. The fusion protein was affinity purified on a glutathione-Sepharose column as described (26). The generation, purification and characterization of the K7E variant of PN1 that shows decreased affinity for heparin has been described elsewhere (14).
Cell Culture-Human foreskin fibroblasts (HF) were grown and maintained in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum as described previously (13,14,17). Experimental cultures were seeded at 1.0 ϫ 10 5 cells/well into 24-well plates. When the cells reached confluence, they were changed to serum free medium and used 48 h later.
Determination of PN1 and PN1(K7E) Activity-Purified samples of PN1 and PN1(K7E) were titrated with active thrombin to determine the percentage of activity. 30 ng of thrombin were added to various amounts of PN1 or PN1(K7E), which were then diluted to a final total volume of 80 l in PBS, pH 7.2, containing 0.1% BSA. At the end of a 30-min incubation at 37°C, the reactions were chilled on ice for 15 min, followed by the addition of a 200-fold molar excess of Chromozym-Th. The reactions were returned to room temperature for 30 min to allow for color development as a measure of residual thrombin activity. Absorbance measurements were taken at 405 nm to quantify color development.
Protein Radioiodination-125 I-Thrombin (16) and 125 I-uPA (27) were prepared as described previously using the Iodogen method. Specific activities for 125 I-thrombin ranged from 8,000 to 15,000 cpm/ng of protein. Specific activities for the 125 I-uPA ranged from 6,000 to 15,000 cpm/ng of protein.
Complex Formation and Analysis-Complexes were formed by combining a known amount of 125 I-Thrombin or 125 I-uPA with a 3-fold molar excess of either PN1 or PN1(K7E) as determined by the active site titration described above. Reactions were carried out in 300 l of PBS. At the end of a 30-min incubation at 37°C, the reactions were diluted with binding medium or buffer to the appropriate concentration, typically 100 -200 ng/ml. After dilution, 5-l aliquots were removed and added to 15 l of SDS-PAGE sample buffer and analyzed by SDS-PAGE on 10% polyacrylamide gels (28). 2-l aliquots of the complexes were also collected and quantified by ␥ counting to ensure that equal concentrations of complexes were added to cells.
Cell Binding, Internalization, and Degradation Assays-Binding and internalization experiments were done in binding medium that consisted of serum-free, bicarbonate-free Dulbecco's modified Eagle's medium, containing 20 nM Hepes buffer, pH 7.2, and 0.1% BSA. When binding assays were done at 4°C, all reagents were prechilled, and the cells were placed on ice in a 4°C cold room. Competing ligands were added simultaneously with radiolabeled ligands. Concentrations of ligands are indicated in the text and figure legends. At the completion of the incubations, unbound ligand was removed, and the cells were washed rapidly four times with 1 ml of PBS, and finally lysed with 1 ml of 10% SDS. Radioactivity in the samples was quantified by ␥ counting. Nonspecific binding, which accounts for approximately 25% of total binding, was determined by incubating three wells with a 400-fold excess of unlabeled complexes. Degradation assays were done by monitoring increases in trichloroacetic acid nonprecipitable radioactivity as described previously (13,14).
Internalization/surface-associated complex assays were done in the same binding medium described above. 300-l samples of complexes were added to triplicate wells at 37°C. At the indicated times the cells were rapidly chilled to 4°C and washed to remove free ligand. 1 ml of ice-cold trypsin solution in PBS (200 g/ml) was then added to the cells on ice and allowed to proteolytically digest cell surface proteins for 10 min. The trypsinized cells were transferred to microcentrifuge tubes containing 200 l of ice-cold soybean trypsin inhibitor at 5 mg/ml. Following centrifugation at 10,000 ϫ g for 3 min, the supernatants and cells were quantified separately by ␥ counting to discriminate between cell surface-associated and internalized complexes. The trypsinization efficiency was determined to be 90% in control experiments where endocytosis was inhibited by incubation at 4°C, and the ability of trypsin to release cell surface-bound complexes was tested.
Immunoprecipitation-125 I-uPA⅐PN1 and 125 I-uPA⅐PN1(K7E) complexes were incubated with either anti-(Pro 47 -Ile 58 -Cys) IgG or with preimmune IgG overnight at 4°C. 50 l of a 1:1 protein G-Sepharose bead slurry in PBS was added, followed by an additional 30 min incubation at room temperature. The samples were centrifuged for 1 min in a refrigerated microcentrifuge at 10,000 ϫ g. Following one wash with PBS, the centrifugation step was repeated, and the antigen-antibody complexes were released from the Sepharose beads by the addition of 100 l of reducing SDS-PAGE sample buffer. Samples were analyzed by SDS-PAGE on 10% polyacrylamide gels. The gels were fixed with methanol:acetic acid:water (5:1:5), and exposed to a Bio-Rad Phospho-Imager screen for 30 min. The image was developed on a Bio-Rad GS-250 Molecular Imager.
Detection of Released Complexes after Endocytosis-125 I-uPA⅐PN1 and 125 I-uPA⅐PN1(K7E) complexes at a concentration of 200 ng/ml were added to 10.0-cm plates of HF cells (5 ϫ 10 6 cells) in 2.5 ml of binding medium. The plates were placed on a rocker in a warm room at 37°C and were allowed to incubate for 2 h to ensure an internalized pool of complexes. After the incubation, free complexes were removed by aspiration, and the plates were washed four times with 5 ml of ice-cold PBS. BSA-free binding medium (1.5 ml) was added back to the plates, and they were placed back on the rocker in the warm room. At the indicated times, the supernatants were removed from the cells and added to 10 ml of 12.5% trichloroacetic acid for 1 h on ice in a 4°C cold room. The precipitates were collected by centrifugation at 10,000 ϫ g for 10 min, and washed once with 1 ml of ice-cold 50/50 acetone/H 2 O. Following a second wash with 100% acetone and air drying, the precipitates were resuspended in 75 l of SDS-PAGE sample buffer and analyzed by SDS-PAGE on a 10% polyacrylamide gel.
pH-dependent Binding Studies-To evaluate the binding of uPA⅐PN1 complexes to the surface of HF cells, a series of binding buffers were prepared at different pH levels using 20 mM Hepes as the starting buffer and titrating it with either HCl or NaOH. BSA was added to a final concentration of 0.1% prior to titration. Complexes were diluted into each of the buffers and incubated with the cells at 4°C for 2 h.
Binding and heparin dissociation of complexes were determined as described above.

RESULTS
The Heparin-binding Site in PN1 Does Not Play a Significant Role in the Cell Surface Binding of uPA⅐PN1 Complexes but Is Required for Their Efficient Catabolism-The ability to bind heparin after SERPIN-protease complex formation is unique to PN1 (29) and plasminogen activator inhibitor-1 (30). Other heparin-binding SERPINs (such as ATIII) lose affinity for heparin after complex formation with a protease (31). This heparin binding capacity is an important feature of Th⅐PN1 complex catabolism because it provides cells with a mechanism for concentrating Th⅐PN1 complexes at the cell surface for subsequent internalization via the LRP (14). In recent studies we have carefully characterized the role of the PN1 heparin-binding site in Th⅐PN1 complex catabolism, but its potential role in uPA⅐PN1 complex catabolism has not been previously addressed. The fact that uPA⅐PN1 complexes bind to the cell surface primarily through the uPAR afforded us the unique opportunity to examine any role for the heparin-binding site in the catabolic pathway separate from its role in cell surface binding.
We have previously generated and characterized a heparin binding-deficient variant of PN1, designated PN1(K7E). PN1(K7E) reacts with thrombin and forms complexes in a manner indistinguishable from native PN1 and also binds to the LRP (14). Because this variant had not been examined for its interaction with uPA, however, its ability to form complexes with uPA was first established (Fig. 1A). Complexes of 125 I-uPA FIG. 1. The heparin-binding site of PN1 is not involved in the cell surface binding of uPA⅐PN1 complexes but is required for their efficient degradation. A, 125 I-uPA (600 ng) was incubated at 37°C for 30 min with a 2-fold molar excess of active PN1 or active PN1(K7E) in a final volume of 100 l. The reactions were then diluted 12-fold with binding medium. 10-l aliquots of the diluted complexes were mixed with an equal volume of reducing SDS-PAGE sample buffer and analyzed by SDS-PAGE on a 10% polyacrylamide gel. 125 I-uPA not reacted with PN1 was run as a control. The dried gel was exposed to a Bio-Rad Phospho-Imager screen for 30 min, and the digitized image was prepared using a Bio-Rad GS-250 Molecular Imager. Lane 1, 125 I-uPA only; lane 2, 125 I-uPA⅐PN1; lane 3, 125 I-uPA⅐PN1(K7E). B, triplicate confluent cultures of HF cells in 24-well plates were incubated at 4°C with either 125 I-uPA⅐PN1 (open bars) or 125 I-uPA⅐PN1(K7E) (closed bars) complexes, prepared as described for A. Peptide AE78 (10 g/ml) or soluble heparin (1 g/ml) in addition to peptide AE78 were added to test wells concurrently. After 3 h, wells were aspirated to remove unbound complexes and washed four times with PBS. The cell monolayers were solubilized in 1 ml of 10% SDS, and radioactivity was quantified by ␥ counting. Nonspecific binding of complexes was determined as described under "Experimental Procedures" and has been subtracted. Error bars indicate one standard deviation from the mean of the triplicate samples in B and C. C, triplicate confluent cultures of HF cells in 24-well plates were incubated at 37°C for the indicated times with 200 ng/ml of 125 I-uPA⅐PN1 (closed circles) or 125 I-uPA⅐PN1(K7E) complexes (open circles). 100-l aliquots of the culture supernatants were removed and precipitated in 12.5% trichloroacetic acid to measure the changes in trichloroacetic acid soluble radioactivity. Background trichloroacetic acid soluble radioactivity was measured in parallel controls that were incubated at 37°C but not exposed to cells and was subtracted from the mean of the triplicate samples.
with either active PN1 and PN1(K7E) were prepared for addition to HF cells as described in the figure legend. Aliquots of 10 l were removed and resolved by SDS-PAGE on 10% polyacrylamide gels. In the digitized image, the position of free 125 I-uPA is indicated as well as the position of 125 I-uPA⅐PN1 and 125 I-uPA⅐PN1(K7E) complexes. Importantly, the K7E variant of PN1 formed complexes with 125 I-uPA as well as native PN1, and there were only trace amounts of free 125 I-uPA present in both the variant and native complexes. This is an important point, because free 125 I-uPA is also able to bind to the uPAR and could lead to misleading conclusions if present in high concentrations.
These same radiolabeled complexes were then added to HF cells in binding medium at 4°C to evaluate cell surface binding. At the end of a 3-h incubation at 4°C, the HF cultures were washed to remove unbound ligand, lysed with 10% SDS and quantified by ␥ counting (Fig. 1B). As shown by the control bars, nearly identical amounts (approximately 17 fmol) of variant 125 I-uPA⅐PN1(K7E) and native 125 I-uPA⅐PN1 complexes bound to the cell surface. This contrasts to previous studies of PN1(K7E) in complex with a different target protease, thrombin (Th), where the cell surface binding of these variant complexes was shown to be 10-fold lower than native Th⅐PN1 complexes (14). Cell surface binding of the native Th⅐PN1 complexes was shown in the same studies to be mediated primarily by cell surface heparins, accounting for the dramatically lowered binding of the heparin binding-deficient Th⅐PN1(K7E) complexes.
Unlike thrombin, uPA binds with high affinity to the uPAR present on the cell surface, and native uPA⅐PN1 complexes have been previously shown to bind via the uPAR (15). To demonstrate that cell surface binding of our variant uPA⅐PN1(K7E) complexes also proceeded via high affinity binding of the uPA moiety of the complex to the uPAR, we tested the ability of a peptide that specifically binds uPAR to compete for cell surface binding. Peptide AE78, derived from the aminoterminal uPAR-binding region of uPA (32,33), was able to effectively compete for over 75% of the binding of the 125 I-uPA⅐PN1(K7E) complexes (Fig. 1B, ϩAE78). In contrast, peptide AE78 did not compete for the binding of native 125 I-uPA⅐PN1 complexes. Because native uPA⅐PN1 complexes contain the heparin-binding site of PN1, they are able to bind to cell surface heparins alternatively when the uPAR-binding site is occupied by peptide AE78. Interestingly the alternative binding of native uPA⅐PN1 complexes to heparin in the presence of peptide AE78 was approximately equal to the high affinity binding of uPA⅐PN1 complexes to the uPAR under control conditions. The addition of both peptide AE78 and soluble heparin effectively reduced the cell surface binding of both native and PN1(K7E) heparin binding-deficient complexes to less than 5 fmol, which may represent the amount of these complexes that is bound directly to the LRP. Because heparin was able to compete for the binding of native uPA⅐PN1 complexes in the presence of peptide AE78, this suggests that the heparin-binding site in PN1 may only play a significant role in total cell surface binding of complexes when a high affinity uPAR-binding site is unavailable (ϩAE78). Considering that both native and variant complexes bound to the cell surface via uPAR approximately equally (control), we expected that the heparin binding-deficient uPA⅐PN1(K7E) complexes would be catabolized comparably to native uPA⅐PN1 complexes.
Measuring no significant decrease in cell surface binding of the variant uPA⅐PN1(K7E) complexes when the uPAR was available, we predicted that degradation of these heparin binding-deficient complexes would also be similar to that of native uPA⅐PN1 complexes. Aliquots of media taken from HF cells incubated at 37°C at 2 and 4 h were subjected to trichloroacetic acid precipitation to measure degradation. Unexpectedly, the degradation rate of 125 I-uPA⅐PN1(K7E) complexes was approximately 3-fold lower than native 125 I-uPA⅐PN1 complexes (Fig.  1C). The data in Fig. 1 (A and B) show that this surprising result is not due to an artifact of inefficient complex formation or cell surface binding and suggested a novel role for the heparin-binding site in complex catabolism.
The LRP-mediated Internalization of uPA⅐PN1 Complexes Proceeds by a Mechanism That Does Not Require the LRPbinding Site in PN1 (Pro 47 -Ile 58 )-uPA⅐PN1 and Th⅐PN1 complexes are both internalized by the LRP, and this appears to be a point of convergence in two separate pathways (14,15,18). uPA⅐PN1 complexes first bind directly to the uPAR and are subsequently internalized as part of a uPA⅐PN1-LRP ternary complex (15,18). In contrast, Th⅐PN1 complexes first bind to cell surface heparins and are subsequently internalized by the LRP without uPAR involvement (14). Whether the Th⅐PN1 complexes are "transferred" to the LRP or internalized as a Th⅐PN1⅐heparin⅐LRP quaternary complex is not known. Based on our unexpected finding of a potential role for heparin in the catabolism of uPA⅐PN1 complexes, we next investigated whether the LRP-binding site in PN1, which was identified in PN1 complexed with thrombin, also participated in uPA⅐PN1 catabolism.
We recently developed a polyclonal antibody that inhibits the LRP-mediated internalization of Th⅐PN1 complexes (17). It is a synthetic peptide directed antibody generated against the PN1derived sequence, 47 PHDNIVISPHGI 58 , a peptide previously shown to inhibit Th⅐PN1 complex internalization (14). Previous studies with this antibody characterized its effect on Th⅐PN1 complexes, so we first wanted to determine the relative accessibility of 47 PHDNIVISPHGI 58 of the immune IgG in uPA⅐PN1 and Th⅐PN1 complexes. Factors such as a difference in conformation, inaccessibility of this site when complexed to uPA or differential exposure of the site after complex formation could all affect the ability of the antibody to bind to the LRP-binding site of PN1. To test this, the ability of the anti-LRP-binding site IgG to immunoprecipitate 125 I-Th⅐PN1 and 125 I-uPA⅐PN1 complexes was evaluated over a broad concentration range ( Fig.  2A). Shown is a digitized image of the samples after resolution by SDS-PAGE. In the graph below the image, band densities were measured and normalized to the respective specific activities of the radioiodinated species. At concentrations of complexes ranging from 10 to 200 ng/ml, there was no significant difference in immunoprecipitation efficiency. Only a slight difference was seen at the complex concentration of 500 ng/ml, but the next experiments were performed within the region where the efficiency of immunoprecipitation of the antibody was equivalent for both Th⅐PN1 and uPA⅐PN1 complexes.
To test the effect of the antibody on the catabolism of uPA⅐PN1 complexes, HF cells were incubated with either 125 I-Th⅐PN1 or 125 I-uPA⅐PN1 complexes at 37°C for 3 h in the presence and absence of immune IgG (Fig. 2B). At the end of the incubation, aliquots of the media were removed and subjected to precipitation with 10% trichloroacetic acid. The appearance of trichloroacetic acid soluble radioactivity in this assay is a direct measurement of the appearance of low molecular weight degradation products (13,14). Over the 3-h time course, approximately 45 fmol of 125 I-uPA⅐PN1 complexes were degraded, and the presence of anti-LRP-binding site IgG had no significant effect on the rate of degradation. 20 fmol of 125 I-Th⅐PN1 complexes were degraded during the same time period. But in sharp contrast, the degradation of Th⅐PN1 complexes was inhibited 80% by the anti-LRP-binding site IgG. These data suggest that the LRP-binding site in PN1 is not required when uPA⅐PN1 complexes are internalized by the LRP, because the immunoprecipitation experiment shown in Fig. 2A demonstrate equal accessibility of this site in both types of complexes. The simplest interpretation of these data is that uPA⅐PN1 complexes and Th⅐PN1 complexes interact with the LRP through different mechanisms. Although Th⅐PN1 complexes bind to the LRP through the PN1 sequence, 47 PHD-NIVISPHGI 58 , uPA⅐PN1 complexes must interact with the LRP through another structural determinant in the complex, most likely located in either uPA or uPAR.
Soluble Heparin Does Not Significantly Dissociate Cell Surface-bound 125 I-uPA⅐PN1 or 125 I-uPA⅐PN1(K7E) Complexes-Because of the unexpected differential in the degradation rate of 125 I-uPA⅐PN1 and 125 I-uPA⅐PN1(K7E) complexes (Fig. 1C), the potential role of cell surface heparin in the binding of complexes was more rigorously examined. In previous studies we have shown that 125 I-Th⅐PN1 complexes bind almost exclusively to cell surface heparins prior to LRP-mediated internalization (14,17). In addition, because of the large number of cell surface heparins present, the Th⅐PN1 complexes do not dissociate significantly even when their concentration in the binding medium is effectively reduced to zero but are rapidly released by the addition of soluble heparin (14,17). To more carefully determine whether any of the cell surface binding of uPA⅐PN1 complexes was heparin-mediated, 125 I-uPA⅐PN1, 125 I-uPA⅐PN1-(K7E), and 125 I-Th⅐PN1 complexes, each at 100 ng/ml, were bound to HF cells at 4°C for 3 h. At the end of the incubation, the monolayers were rapidly washed four times and placed in binding media containing soluble heparin to measure dissociation (Fig. 3). In agreement with previously published data, 125 I-Th⅐PN1 complexes were rapidly released by the addition of soluble heparin (14,17). In contrast, 125 I-uPA⅐PN1(K7E) heparin binding-deficient variant complexes were not released significantly, demonstrating that their binding to the cell surface is non-heparin-mediated. This is consistent with the data shown in Fig. 1B, where the cell surface binding of uPA⅐PN1-(K7E) complexes was shown to occur primarily via the uPAR and was greatly decreased only when the uPAR was occupied by peptide AE78. In the case of native 125 I-uPA⅐PN1 complexes, a small fraction (20%) was released by soluble heparin, suggesting that the majority of these complexes were bound to the uPAR as well. Consistent with Fig. 1B again, these native complexes can also use the heparin-binding site of PN1 as an alternative and therefore show some dissociation in the presence of soluble heparin. However, when a high affinity uPARbinding site is available, the heparin-binding site in PN1 ap- parently plays very little role in the total cell surface binding of native 125 I-uPA⅐PN1 complexes, because only 20% of the complexes are dissociated in media containing soluble heparin. Importantly, these data clearly show that no measurable amount of 125 I-uPA⅐PN1(K7E) complexes were bound to the cell surface by heparin, although almost equivalent amounts of both uPA⅐PN1 and uPA⅐PN1(K7E) are cell surface-associated (Fig. 1B). Both native uPA⅐PN1 and variant uPA⅐PN1(K7E) complexes are bound to the cell surface predominantly via the uPAR (high affinity), although a small percentage of the native complexes are also bound via cell surface heparins. The inability of the variant complexes to bind to heparin appears to have a much greater than expected inhibitory effect on internalization (Fig. 1C), which led us to consider alternative roles for heparin in the catabolism pathway. However, because some binding of the native uPA⅐PN1 complexes is mediated by heparin, it was important to determine whether this small difference in the initial step in internalization could account kinetically for the difference in degradation rate observed in Fig. 1C.
The Heparin-binding Site of PN1 Is Required for the Intracellular Retention of 125 I-uPA⅐PN1 Complexes after Endocytosis-We next examined the initial internalization step to determine whether only the native complexes were being initially internalized, potentially because of the heparin dissociable species. The endocytosis of 125 I-uPA⅐PN1(K7E) variant complexes was directly compared with that of native 125 I-uPA⅐PN1 complexes by determination of the ratio of internalized ligand to cell surface-bound ligand (34). If only heparin dissociable native complexes can be internalized effectively, the initial ratio of internalized native complexes to surface-bound native complexes (In/Sur ratio) should be much greater than the initial In/Sur ratio of variant complexes. HF cells were incubated with 125 I-uPA⅐PN1 or 125 I-uPA⅐PN1(K7E) complexes at 37°C. At the indicated times, triplicate cultures were chilled to 4°C and subjected to trypsinization to remove cell surface-bound complexes. Internalized radiolabeled ligand in the cell pellet and surface-bound trypsin-sensitive ligand were quantified by ␥ counting. Control incubations done only at 4°C demonstrated that the trypsinization procedure was over 90% efficient in removing cell surface-bound complexes (data not shown). Shown in Fig. 4A is the ratio of internalized/cell surface-bound complexes (In/Sur) plotted versus time (34). At the 1-min time point, the In/Sur ratios for 125 I-uPA⅐PN1 and 125 I-uPA⅐PN1-(K7E) complexes were close, 0.8 and 0.6, respectively. At the 3-min time point however, the In/Sur ratio for the 125 I-uPA⅐PN1(K7E) complexes dropped to 0.5, whereas the In/Sur ratio for the 125 I-uPA⅐PN1 complexes climbed to 1.6. The In/Sur ratios for the 125 I-uPA⅐PN1(K7E) complexes remained fairly constant throughout the remainder of the 15-min time course, whereas the In/Sur ratios for the 125 I-uPA⅐PN1 complexes reached a near maximum of 2.5 at 10 min and then increased at a much slower rate thereafter. The slower rate of increase after 10 min is consistent with the time expected for the earliest internalized complexes to reach the lysosomes (34).
To exclude the possibility that the 125 I-uPA⅐PN1(K7E) complexes are never endocytosed, the In/Sur ratios were also determined in the presence of the LRP agonist, RAP (Fig. 4A,  open symbols). In the presence of RAP, the In/Sur ratios of both 125 I-uPA⅐PN1 and 125 I-uPA⅐PN1(K7E) complexes were markedly reduced initially and did not increase, indicating that RAP inhibited the internalization of both complexes. Importantly, the much lower initial In/Sur ratio for the variant uPA⅐PN1-(K7E) complexes in the presence of RAP strongly suggests that, in the absence of RAP, 125 I-uPA⅐PN1(K7E) complexes are internalized initially. Because the initial internalization at the 1-min time point of both normal and heparin binding-deficient complexes is very similar, the quick decrease of the In/Sur ratio of 125 I-uPA⅐PN1(K7E) complexes at the 3-min time point and beyond in the absence of RAP may be caused by a lack of retention because of retro-endocytosis.
Following this line of reasoning, we noted that one of the earliest changes that occurs after internalization and endosome formation is a rapid lowering of the pH, and we decided to explore the potential role of pH in the endosomal retention of uPA⅐PN1 complexes. Monensin, a sodium ionophore that prevents the lowering of pH in the endosomes, was tested for its effect on the In/Sur ratio of normal uPA⅐PN1 complexes. Interestingly, in the presence of monensin, native 125 I-uPA⅐PN1 complexes displayed endocytosis kinetics nearly identical to 125 I-uPA⅐PN1(K7E) complexes. This suggests a possible role for pH in the retention of the complexes in the endosomes. The endocytosis kinetics of uPA⅐PN1(K7E) complexes did not change significantly in the presence of monensin (data not shown).
Because the uPA⅐PN1(K7E) complexes are internalized, initially we next wanted to determine whether they were simply Each error bar indicates Ϯ one standard deviation from the mean of the triplicate samples. B, confluent 10-cm plates of HF cells were incubated at 37°C for 2 h with either 125 I-uPA⅐PN1 or 125 I-uPA⅐PN1(K7E) complexes. Excess complexes were removed, and BSA free binding medium was added back to the cells. At the indicated times, 50-l samples of the culture supernatants were removed and added to an equal volume of SDS-PAGE sample buffer. Samples were analyzed by SDS-PAGE gel on a 10% polyacrylamide gel. The gel was exposed for 2 h to a Bio-Rad Phospho-Imager screen, and the digitized image shown was prepared by scanning with a Bio-Rad GS 250 Molecular Imager. retained in the endosome (which seems unlikely because there was little time-dependent increase in the intracellular pool), degraded, or perhaps released back into the medium. To determine whether 125 I-uPA⅐PN1(K7E) complexes might be released from the cells in an intact, partially degraded or fully degraded form, we characterized released material from cells incubated with complexes at 37°C, followed by a washout. Four 10-cm plates of HF cells, approximately 7 ϫ 10 6 cells, were incubated with 200 ng/ml of 125 I-uPA⅐PN1 and 125 I-uPA⅐PN1(K7E) complexes for 30 min at 37°C so they could establish a pool of endocytosed complexes. The cells were chilled to 4°C, washed four times to remove free complexes, and returned to 37°C. At the 30-and 60-min time points, 50-l samples of the media were removed and analyzed by SDS-PAGE, and the digitized image shown in Fig. 4B was prepared.
At the 30-min time point radioactive bands corresponding to the molecular weight of the protease-inhibitor complexes were apparent in the media from both cultures that received uPA⅐ PN1 and those that were preincubated with uPA⅐PN1(K7E). We were unable to detect any smaller degradation products. These bands most likely represent the small fraction of the complexes that were bound to cell surface uPAR at 37°C and then dissociated and have not yet rebound because the concentration of complexes in the medium has been reduced greatly by the washout procedure. This would be consistent with the data in Fig. 3 clearly indicating that the majority of both 125 I-uPA⅐PN1 and 125 I-uPA⅐PN1(K7E) complexes are bound to uPAR and not to heparin. However, by 60 min, the intensity of the 125 I-uPA⅐PN1(K7E) complex band nearly doubled, whereas the radioactivity visible in the native 125 I-uPA⅐PN1 complex lanewasatbackgroundlevels.Thereappearanceof 125 I-uPA⅐PN1-(K7E) complexes in the medium at a higher level than 125 I-uPA⅐PN1 complexes cannot be explained solely by continued dissociation of cell surface-bound complexes, because the data in Fig. 1 show that the total amount of cell surface-bound 125 I-uPA⅐PN1 and 125 I-uPA⅐PN1(K7E) complexes is nearly identical. The data are consistent with the release of 125 I-uPA⅐PN1(K7E) complexes from the endocytic pool, after their initial internalization.
The Heparin-binding Site in PN1 Is Required for the Binding of 125 I-uPA⅐PN1 Complexes to Components on the Cell Surface at Acidic pH-Taken together, the data in Fig. 4 suggest that the heparin-binding site in PN1 interacts with a component putatively located in the endosomal lumen and potentially derived from the cell surface, that is required for retention after endocytosis. Because the external cell surface membrane is topologically similar to the lumenal side of the endosomal membrane, we exploited this attribute to test the effect of acidic pH on 125 I-uPA⅐PN1 complex dissociation from the cell surface. We hypothesized that the heparin-binding site of PN1 is required for complex binding and retention only after the uPAR interaction is disrupted by acidic pH and predicted that there would be little change in the binding of 125 I-uPA⅐PN1 complexes that contain this retention site as the membrane environment became acidified. Following the same line of reasoning, the binding of 125 I-uPA⅐PN1(K7E) complexes should decrease substantially, because the putative retention site is nonfunctional. Drawing on the previously mentioned topological similarity between the cell surface composition and the lumenal surface of the endosomes, we tested this hypothesis by measuring the binding of 125 I-uPA⅐PN1 and 125 I-uPA⅐PN1(K7E) complexes to the cell surface over a broad range of pH. Cultures of HF cells were incubated with 125 I-uPA⅐PN1 and 125 I-uPA⅐PN1(K7E) complexes at the indicated pH levels for 2 h at 4°C (Fig. 5A). At the end of the incubation, the labeled ligands were removed, and the cultures were washed four times to remove any residual unbound ligand. The cultures were then solubilized in 10% SDS and quantified by ␥ counting. From pH 5.75 to 5.25, the binding of 125 I-uPA⅐PN1 and 125 I-uPA⅐PN1-(K7E) complexes was remarkably similar, approximately 24 fmol. Between pH 5.25 and 5.0 there was a marked decrease in the cell surface association of 125 I-uPA⅐PN1(K7E) complexes, whereas the association of 125 I-uPA⅐PN1 was relatively unchanged. Below pH 5.0, the association of 125 I-uPA⅐PN1 complexes continued to decrease and approached that of 125 I-uPA⅐PN1(K7E) complexes as the pH was lowered. The large difference in cell surface complex association in the pH 5.25 to pH 5.0 range has physiological relevance, because this is the reported pH of late endosomes (35) The simplest interpretation of these results is that below pH 5.2, the binding of 125 I-uPA⅐PN1 complexes switches from the uPAR, to an unidentified component through a mechanism that involves the heparin-binding site of PN1.
To probe the identification of this entity further, we next I-uPA⅐PN1 complexes in medium buffered at the indicated pH levels. Following the removal of unbound complexes the cell monolayers were washed four times with ice-cold PBS. Binding medium alone (closed circles) or containing 100 nM soluble heparin (closed squares) was added back to the cultures that were then incubated for an additional 15 min at 4°C. The medium was then removed, and the PBS washes were repeated. The cell monolayers were solubilized in 1 ml of 10% SDS, and the radioactivity was quantified by ␥ counting. Each error bar indicates Ϯ one standard deviation from the mean of triplicate samples. examined the ability of heparin to compete for the cell surface binding of native uPA⅐PN1 complexes over the same pH range (Fig. 5B). Test cultures were incubated with native uPA⅐PN1 complexes under the same experimental conditions as shown in Fig. 5A and then were placed in binding medium alone or containing 100 nM soluble heparin for an additional 15 min. From pH 5.75 to pH 5.25, none of the bound complexes were competed for by the addition of exogenous soluble heparin. Interestingly, at pH 5.0, just over 80% of the bound native complexes suddenly became dissociable in the presence of added heparin. These data are most consistent with a mechanism that predicts a shift in the binding of native uPA⅐PN1 complexes from the uPAR to heparin at between pH 5.25 and pH 5.0.
A Heparin-binding Peptide Derived from the Heparin-binding Domain of PN1 Specifically Inhibits the Degradation of uPA⅐PN1 Complexes-As an independent means of probing the potential role of the heparin-binding site in PN1 in intracellular retention and perhaps trafficking of uPA⅐PN1 complexes, we employed a synthetic peptide strategy. The heparin-binding site in PN1 has been narrowed to residues 70 -87 in the mature protein (29). A synthetic peptide, Gly 70 -Asp 87 (GKILKKIN-KAIVSKKNKD) was synthesized and purified as described previously (13). A control peptide representing the homologous heparin-binding site of ATIII (residues 124 -145, AKLN-CRLYRKANKSSKLVSANR) was also prepared.
The heparin binding properties of the ATIII heparin-binding peptides have been well characterized in previous studies (25,36), but the peptide derived from the PN1 heparin-binding sequence is not as well studied. The experiment shown in Fig.  6A was done to illustrate that the PN1-derived synthetic peptide, Gly 70 -Asp 87 , does indeed bind heparin in a specific and saturable manner. 1-g quantities of Gly 70 -Asp 87 in individual dots on nitrocellulose were incubated in duplicate with 100 ng/ml of 125 I-F-heparin in the absence and in the presence of increasing concentrations of nonlabeled soluble heparin. Relative to the control incubation with 125 I-F-heparin only, increasing the concentration of nonlabeled heparin initially increased binding slightly but eventually competed for nearly all of the binding at the 500 ng/ml concentration.
Having established that the PN1 heparin-binding peptide, Gly 70 -Asp 87 , does bind heparin specifically and saturably, the effect of Gly 70 -Asp 87 on the cell surface binding of uPA⅐PN1 complexes was evaluated. Confluent HF cell cultures were incubated with 100 ng/ml of uPA⅐PN1 complexes, in the absence and presence of Gly 70 -Asp 87 over a concentration range of 1 to 50 g/ml at 4°C (Fig. 6B). Even at the highest concentration (50 g/ml), Gly 70 -Asp 87 was able to compete for only about 10% of uPA⅐PN1 complex binding. This is consistent with the data shown in Fig. 1, where it was shown that the majority of the complexes were bound to the uPAR, and in Fig. 3, where it was shown that only a small fraction of the complexes were dissociable from the cell surface by the addition of soluble heparin. It also supports a model in which the uPA⅐PN1 complexes do not release from high affinity binding to the uPAR at the cell surface under normal conditions but only after internalization occurs.
The effect of Gly 70 -Asp 87 on the degradation of uPA was next evaluated at intervals of 1 and 2 h. Triplicate cultures of HF cells were incubated with uPA⅐PN1 complexes, in the presence and absence of Gly 70 -Asp 87 (Fig. 6C). At 1-and 2-h intervals, samples were subjected to trichloroacetic acid precipitation to measure the degradation of uPA⅐PN1. Degradation was fairly linear, with 30 fmol degraded in the first hour and 72 fmol degraded by the end of the second hour. At the1-h time point, at a concentration of 10 M, Gly 70 -Asp 87 decreased degradation by approximately 70%. At the 2-h time point, the decrease was not as prominent, approximately 50%, which might be explained by a rapid degradation of the peptide. At a concentration of 20 M Gly 70 -Asp 87 , the effect on degradation was much more pronounced, with an 87% decrease in the degradation of uPA⅐PN1 at the 1-h time point. When a second addition of Gly 70 -Asp 87 at 20 M at the end of the 1-h time point, nearly the same magnitude of effect was observed at the 2-h time point, supporting the idea that the Gly 70 -Asp 87 may have a relatively short half-life in presence of cells at 37°C.
Finally, we wanted to rule out the possibility that the effect of Gly 70 -Asp 87 might be explained by artifact because of its high positive charge density. As a control, we repeated the degradation experiment shown in Fig. 6C using Gly 70 -Asp 87 and Ala 124 -Arg 145 , which represents the heparin-binding site of human ATIII, in parallel. The PN1-derived heparin-binding peptide, Gly 70 -Asp 87 again markedly reduced the degradation of uPA⅐PN1 complexes, whereas in contrast, the ATIII heparinbinding peptide, Ala 124 -Arg 145 , had no effect (Fig. 6D). This is an important and convincing control, because the ATIII-and PN1-derived peptides are nearly identical in charge density, both bind heparin, and are predicted to have the amphipathic helical structure. These data suggest that there is a specific heparin sequence responsible for the endosomal retention of uPA⅐PN1 complexes. DISCUSSION The present studies were undertaken to determine whether the pathways and components involved in the cellular binding, internalization, and degradation of uPA⅐PN1 complexes differed significantly from that of Th⅐PN1 complexes. Utilizing a genetically engineered PN1 variant that is deficient in heparin binding and a recently developed polyclonal antibody specific for the LRP-binding site in PN1 that inhibits the LRP-mediated internalization of Th⅐PN1 complexes, we conclude that the pathways are distinct but share the LRP as a common endocytosis mechanism.
There are two major differences in how uPA⅐PN1 and Th⅐PN1 complexes are catabolized. First, the binding of Th⅐PN1 complexes to the cell surface is mediated primarily by heparins (14,17). This is an important step in the catabolism of the complexes, because it serves to keep the concentration of the complexes at the cell surface relatively high, which in turn promotes interaction with the LRP, a relatively low affinity interaction. In contrast, the binding of uPA⅐PN1 complexes is nearly completely independent of heparins and is primarily mediated by the uPAR. The second major difference is the interaction of uPA⅐PN1 and Th⅐PN1 complexes with the LRP. In both pathways the LRP functions only as an endocytosis mechanism (14,15). Parameters affecting the rate of transfer of protease-PN1 from heparin or from uPAR to the LRP could have a marked effect on turnover. To date it is not clear whether transfer to LRP occurs as a separate step in the pathway or whether the heparin or uPAR are co-internalized with the complex. In the case of Th⅐PN1 complexes, the endocytosis step requires a specific structural region in PN1, amino acid residues Pro 47 -Ile 58 (13,17). Amino acid substitutions in this region, as well as a polyclonal antibody generated against this sequence, markedly impair Th⅐PN1 complex internalization by the LRP. Interestingly, this same region is not required for the LRP-mediated internalization of uPA⅐PN1 complexes. There is apparently an intermediate affinity binding site in uPA that is sufficient for LRP internalization (30). Whether the LRP-binding sites in PN1 and in uPA can be used simultaneously is not known, but clearly the blockage of the PN1 site does not affect the overall rate of uPA⅐PN1 catabolism. Thus, although the LRP-mediated endocytosis is a common step in the catabolism of uPA⅐PN1 and Th⅐PN1 complexes, there are major molecular differences in the details of the pathways.
One of the more interesting observations made in the present studies is the apparent involvement of the heparin-binding site in PN1 in the post-endocytic retention of uPA⅐PN1 complexes. After initial internalization, uPA⅐PN1(K7E) complexes that have a nonfunctional heparin-binding site appear to be released from the cells in an intact form. uPA⅐PN1 complexes made with native PN1, on the other hand, proceed on to the lysosomes where they are degraded. Examination of the binding of 125 I-uPA⅐PN1 and 125 I-uPA⅐PN1(K7E) complexes to the cell surface at different pH levels revealed that between pH 5.25 and 5.0 the binding of 125 I-uPA⅐PN1 complexes was relatively unchanged, whereas the binding of 125 I-uPA⅐PN1(K7E) complexes was reduced approximately 3-fold. Interestingly, the fold difference in binding at pH 5.0 correlates well with the overall decreased rate of catabolism of 125 I-uPA⅐PN1(K7E) complexes, which is 3-fold. The most obvious candidate molecule to participate in the binding of 125 I-uPA⅐PN1 at pH 5.0 is heparin or a heparan sulfate proteoglycan for the following reasons. First, the only known difference between PN1 and PN1(K7E) is the inability of the latter to bind heparin. PN1(K7E) displays normal kinetics of thrombin inhibition, forms covalent complexes with both thrombin and uPA, and FIG. 6. The heparin-binding peptide derived from PN1 displays specificity in markedly reducing catabolism of 125 I-uPA⅐PN1 complexes without competing for their cell surface binding. A, 1-g quantities of Gly 70 -Asp 87 synthetic peptide were applied to nitrocellulose in 0.5-cm circles. After blocking with 3% BSA, they were incubated in duplicate with 100 ng/ml of 125 I-F-Heparin, in the absence and in the presence of increasing concentrations of nonlabeled soluble heparin for 2 h at room temperature. After removal of the labeled heparin solution, the nitrocellulose was washed as described under "Experimental Procedures" and exposed for 2 h to a Bio-Rad Phospho-Imager screen. A digitized image was prepared by scanning with a Bio-Rad GS 250 Molecular Imager, and the pixel density of each dot was determined by integration. Pixel density was normalized to known quantities of radiolabeled heparin. The graph shows the amount of radiolabeled heparin that remained bound to the nitrocellulose in the presence of increasing concentrations of unlabeled heparin, relative to control dots incubated with 125 I-F-heparin only. B, triplicate cultures of HF cells were incubated with 125 I-uPA⅐PN1 complexes at a concentration of 100 ng/ml and the indicated concentrations of Gly 70 -Asp 87 synthetic peptide for 2 h at 4°C. At the end of the incubation, the labeled ligand was removed, and the cultures were washed and processed as described in the legend to Fig. 5A. Each error bar indicates Ϯ one standard deviation from the mean. C, triplicate cultures of HF cells were incubated with 125 I-uPA⅐PN1 complexes at a concentration of 100 ng/ml at the 37°C in the absence (closed circles) and in the presence of Gly 70 -Asp 87 at a concentration of 10 m (open circles) or 20 M (open triangles). At the indicated times, 100-l aliquots of the culture supernatants were removed and assayed for the appearance of trichloroacetic acid soluble radioactivity as described in the legend to Fig. 1C. Each error bar indicates Ϯ one standard deviation from the mean. D, triplicate cultures of HF cells were incubated for 2 h at 37°C with 100 ng/ml of 125 I-uPA⅐PN1 complexes alone or in the presence of the ATIII heparin-binding peptide (124 -145) or the PN1 heparin-binding peptide (70 -87), each at a concentration of 20 M. At the end of the incubation, aliquots were removed and assayed for the appearance of trichloroacetic acid soluble radioactivity as described in the legend to Fig. 1C. Each error bar indicates Ϯ one standard deviation from the mean. binds to the LRP (14). Second, the biochemical requirements for heparin binding are most consistent with a binding interaction that would not be adversely affected at pH 5.0. There would be no change in the status of lysine and arginine protonation from pH 7.0 to 5.0 because of the high pK a of these amino acids. In addition the protonation of the sulfates on heparin would not change until a much lower pH when the pK a of the sulfate groups is reached. Heparin and heparan sulfate differ in sulfate content, although previous work has shown almost no difference in their biological activities with respect to PN1 binding and activation (29,37). At the present level of analysis we cannot discern between the interaction of uPA⅐PN1 complexes with heparin compared with heparan sulfate proteoglycans on the cell surface or in the endosomes.
We directly addressed the role of heparin in the post-endocytic retention of 125 I-uPA⅐PN1 complexes in two different experimental paradigms. In the first, we demonstrated that the binding of 125 I-uPA⅐PN1 complexes shifted from a state of heparin insensitivity at pH 5.25 to a state of heparin sensitivity at pH 5.0. In the second, we demonstrated that a synthetic heparin-binding peptide derived from PN1 specifically inhibited the catabolism of 125 I-uPA⅐PN1 complexes, despite the fact that it had no effect on the cell surface binding of the complexes. A structurally similar peptide with an identical charge density derived from the ATIII heparin-binding sequence had no effect on catabolism. This ruled out a simple charge-charge interaction between the heparin-binding peptides and heparins, indicating some degree of specificity. This specificity is not all that surprising, because it is known that the synthetic heparin pentasaccharide specifically activates ATIII toward factor Xa and thrombin but binds poorly if at all to PN1. What these data suggest is that after release from uPA in the low pH environment of the endosomes, binding to a heparin or heparin-like molecule is required for procession to the next step in transit to the lysosomes. It is interesting to note that all known LRP ligands, with the exception of ␣-2-macroglobulin, bind heparin (24). It will be important to determine whether the heparinbinding sites in these ligands serve a similar function and to determine more precisely the molecular nature of the glycosaminoglycan molecule responsible for the retention of PN1protease complexes in the endosomes.
Finally, the results of these studies make an important point about the clearance of SERPIN-protease complexes. There is no universal mechanism, and no structural determinant in the SERPINs that stretches across the entire SERPIN family required for cellular uptake and catabolism as has been previously proposed (21). Indeed, the data in the present report clearly illustrate that the structural requirement for the endocytosis of a single SERPIN can be determined by the protease moiety in the inhibitory complex. The larger picture that is now emerging is that the LRP acts as a common endocytosis mechanism for many of the SERPIN-protease complexes, but the delivery route to the LRP may be specific to individual SER-PINs or classes of SERPINs.