Analysis of a Structural Determinant in Thrombin-Protease Nexin 1 Complexes That Mediates Clearance by the Low Density Lipoprotein Receptor-related Protein*

We recently identified a synthetic peptide, Pro47–Ile58, derived from the mature protease nexin 1 (PN1) sequence, that inhibited the low density lipoprotein receptor-related protein (LRP)-mediated internalization of thrombin-PN1 (Th-PN1) complexes. Presently, we have analyzed this sequence in Th-PN1 complex catabolism using two independent approaches: 1) An antibody was generated against Pro47–Ile58, which inhibited complex degradation by 70% but had no effect on the binding of the complexes to cell surface heparins. This places the structural determinant in PN1 mediating complex internalization by the LRP outside of the heparin-binding site. 2) Site-directed genetic variants of PN1 with a single Ala substitution at His48, or two Ala substitutions, one at His48 and another at Asp49, were expressed in Sf9 insect cells. The catabolic rate of complexes formed between Th and the singly substituted and doubly substituted variants was lowered to 50 and 15%, respectively, when compared with the catabolic rate of native Th-PN1 complexes. This is the first analysis of a structural determinant in a serineprotease inhibitor (SERPIN) required for LRP-mediated internalization and in part may explain the cryptic nature of this site in the unreacted serine protease inhibitor.

Protease nexin 1 (PN1) 1 is a member of the SERPIN superfamily (1,2), and an important physiological regulator of thrombin (Th) and urinary plasminogen activator (uPA) (2). PN1 forms stoichiometric complexes with both Th and uPA (3) that ultimately results in their removal by cellular endocytosis and degradation (4). The precise biochemical nature of the complexes is still not completely clear, but they are extremely stable and possibly covalent (5)(6)(7). When complexes are formed between the SERPIN and its target protease, there is an accompanying conformational change in the SERPIN that either unmasks or causes the formation of a new binding site in the complexed SERPIN that is not present in the free SERPIN (8,9). The cryptic nature of the LRP-binding site in the free SERPIN makes sense biologically. It ensures that SERPINs will remain extracellular, either in plasma or in tissues near cell surfaces, until they have formed an irreversible complex with a protease.
The list of SERPINs dependent on the LRP for cellular internalization includes, protease nexin 1 (PN1) (10,11), heparin cofactor-II (12), antithrombin III (ATIII) (12) and ␣-1-antitrypsin (12). It is interesting to note that although the LRP acts as the internalization receptor, other cellular components are most likely required for the efficient catabolism of the SERPIN-Protease complexes (12)(13)(14)(15). In the case of the plasma SER-PINs these components remain to be identified. However PN1, which is primarily restricted to tissues, utilizes at least two different cell surface molecules to assist LRP-mediated internalization. When PN1 forms complexes with uPA, the uPA receptor is required for efficient concentration at the cell surface of uPA-PN1 complexes and subsequent internalization via the LRP (11,16). In contrast, when PN1 is in complex with thrombin, heparin chains present at the cell surface greatly facilitate the uptake and turnover of thrombin-PN1 complexes (15), and uPA receptor is not involved (11,16). Thus, in the case of PN1, the nature of the target protease directly plays a role in the clearance mechanism.
In a recent study using a synthetic peptide library strategy, a putative LRP-binding site was identified in PN1 (10). The library consisted of peptides 12 amino acids in length, and spanned nearly the entire PN1 sequence. A single peptide in the library, 47 PHDNIVISPHGI 58 was identified as a potent inhibitor of Th-PN1 internalization and degradation. Using ␣-1-antitrypsin structure and sequence alignments, this sequence is predicted to be a transition sequence that occurs just after helix A and continues to form the sixth strand of sheet B (1). Consequently, this site meets the criteria of being at least partially buried in the intact SERPIN, with the potential of becoming more exposed after complex formation with thrombin. There is no direct evidence, however, to distinguish this possibility from a simple conformational change in which the exposure of this site remains constant.
In the present studies we further investigate the potential role of this putative site in LRP-mediated internalization using two different approaches. In the first approach, a polyclonal IgG was generated against Pro 47 -Ile 58 with a cysteine residue added after the Ile to facilitate haptenization to ovalbumin. We demonstrate that this antibody specifically and selectively inhibits the binding of Th-PN1 complexes to the LRP, but does not affect the interaction of the complexes with cell surface heparins. In the second approach we utilize site-directed mutagenesis and baculovirus-driven expression in insect cells. Two variant forms of PN1 were expressed; one with an alanine substitution at the position of His 48 (H48A), and another with alanine substitutions at the positions of His 48 and Asp 49 (H48A,D49A). Each PN1 variant was characterized biochemi-cally by determining the k assoc value for thrombin inhibition and ability to form SDS-resistant complexes with thrombin. Additionally, the PN1 variant-thrombin complexes and native PN1-thrombin complexes were assayed for their capacity to bind to cell surface heparins. While the PN1 variants were found to be very similar to native PN1 in their ability to inactivate thrombin and bind to cell surface heparins, complexes made with each of the PN1 variants showed decreased rates of catabolism. These experiments define a critical role for the structural determinant, Pro 47 -Ile 58 , in the LRP-mediated internalization of Th-PN1 complexes. These data also demonstrate that with the use of anti-(Pro 47 -Ile 58 ) antibody, cell surface heparin binding and LRP-mediated internalization of Th-PN1 complexes can be studied as independent events, even though they act cooperatively to facilitate complex catabolism.

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, was purchased from Calbiochem, and Na 125 I was from Amersham Pharmacia Biotech. High-trap heparin-Sepharose and Cibacron blue-Sepharose were from Amersham Pharmacia Biotech. Recombinant enzymes were from Boehringer Mannheim. All other common shelf reagents were either from Sigma or Calbiochem. The synthesis of Pro 47 -Ile 58 -Cys has been described previously (10). Antibodies against this peptide were raised in rabbits, using Pro 47 -Ile 58 -Cys coupled to ovalbumin as described previously (17). The LRP agonist, receptor-associated protein, was expressed as a receptor-associated protein-glutathione S-transferase 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 (15).
Cell Culture-Human foreskin fibroblasts (HF) were grown and maintained as described previously (8). Experimental cultures were seeded at 1 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.
Preparation and Expression of Recombinant Proteins-The methods used to prepare and express recombinant forms of PN1 in a baculovirusdriven expression system have been described in detail elsewhere (18) and are briefly summarized here. Recombinant cDNA constructs of PN1 in pBluescript with Ala substitutions at His 48 and Asp 49 were prepared using overlapping polymerase chain reaction to introduce the desired nucleotide substitutions as described previously (19). The constructs were sequenced and cloned into the pVL baculovirus shuttle vector and co-transfected into Sf9 insect cells along with BaculoGold baculovirus. Recombinant viruses were purified by a single round of plaque purification. For protein expression, Sf9 cells grown in T175 flasks were infected at a multiplicity of infection of 10:1. Five to seven days later the media were harvested, and the recombinant forms of PN1 were purified by affinity chromatography on Cibacron blue-Sepharose (20).
Determination of PN1 Activity and k assoc Constants-Purified samples of PN1 and the recombinant variants of known protein concentration (21) were titrated with active thrombin to determine the percentage of activity. 30 ng of thrombin were added to various amounts of the PN1 samples in a final volume of 100 l of PBS, pH 7.2, containing 0.1% bovine serum albumin. At the end of a 30-min incubation, the reactions were chilled on ice, and a 200-fold molar excess of Chromozym-Th was added. 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. k assoc values were determined as described previously (2).
Protein Radioiodination-125 I-Thrombin was prepared as described previously, using the Iodogen method (22). Specific activities ranged from 8,000 to 15,000 cpm/ng of protein.
Th-PN1 Complex Formation and Analysis-600 ng of 125 I-thrombin was added to amounts of PN1, PN1(H48A), and PN1(H48A,D49A) required to achieve complete thrombin inhibition as determined in the titration assay described above. Reactions were carried out in 300 l of PBS containing 0.1% bovine serum albumin. At the end of a 30-min incubation, the reactions were diluted with binding medium to a final concentration of 200 ng/ml. Prior to 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 (23).
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 mM Hepes buffer, pH 7.2, and 0.1% bovine serum albumin. When binding assays were done at 4°C, all reagents were pre-chilled, 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 four times with 1 ml of PBS and finally lysed with 10% SDS. Radioactivity in the samples was quantified by ␥ counting. Internalization assays were done in the same binding medium in a 37°C water bath. 125 I-Th-PN1 complexes were added to the wells at a concentration of 200 ng/ml, and at the indicated times, triplicate samples were rapidly chilled to 4°C. Cell surface bound complexes and internalized complexes were determined as described previously (8).
Degradation assays were also done in the same binding medium in a 37°C water bath. 125 I-Th-PN1 complexes were added to each well at a final concentration of 200 ng/ml in a volume of 250 l. At the indicated times, 100-l aliquots from triplicate samples were each added to 1 ml of ice-cold 12% trichloroacetic acid. After a minimum incubation of 2 h on ice, trichloroacetic acid precipitable material was removed by centrifugation at 10,000 ϫ g for 10 min in a refrigerated microcentrifuge. Aliquots of the supernatants were quantified by ␥ counting to measure nonprecipitable radioactivity. Samples from control incubations performed in the absence of cells were processed identically to determine background levels of trichloroacetic acid nonprecipitable radioactivity and were subtracted from the values determined in parallel samples.

Anti-(Pro 47 -Ile 58 ) Antibody Selectively Blocks the LRP-mediated Internalization of Th-PN1 Complexes but Does Not Affect Heparin-mediated Cell Surface Binding-A rabbit polyclonal
IgG was generated against Pro 47 -Ile 58 , and the total IgG fraction was purified on protein G-Sepharose and brought to a final concentration of 1.0 mg/ml in PBS. To test the effect of this IgG on Th-PN1 complex degradation, 125 I-Th-PN1 complexes were formed and pre-incubated with various dilutions of the antibody for 30 min at 37°C. The complexes were then added to cells for 3 h at 37°C, and release of trichloroacetic acid soluble radioactivity into the incubation medium was measured. At a 1:10 dilution of the IgG, complex degradation was inhibited approximately 70% (data not shown). Because the peptide sequence, Pro 47 -Ile 58 , lies in close proximity to the heparinbinding site, additional experiments were done to determine whether the decrease in complex degradation was due to inhibiting the LRP-binding site in the complexes or to a steric hindrance of heparin binding. Inhibition of heparin binding could also account for the lower degradation rate in the presence of the antibody, because we have previously shown that a heparin binding deficient variant of PN1 in complex with thrombin is degraded very slowly when compared with native PN1 (15). Shown in Fig. 1 are parallel experiments done at 4 and 37°C, where we simultaneously measured the effect of anti-(Pro 47 -Ile 58 ) IgG on both the degradation and cell association of Th-PN1 complexes. Relative to cultures that received only 125 I-Th-PN1 complexes or a control pre-immune IgG, experimental cultures that received anti-(Pro 47 -Ile 58 ) IgG degraded significantly lower amounts of complexes during a 3-h incubation at 37°C; approximately 70% less when compared with the complex only controls (Fig. 1A). As expected, degradation was reduced by 80% in the presence of the LRP agonist, RAP-GST. Insignificant quantities of trichloroacetic acid soluble radioactivity were generated at 4°C, demonstrating that complex degradation is dependent on endocytosis and occurs intracellularly. In contrast to the marked effect that anti-(Pro 47 -Ile 58 ) IgG had on complex degradation, it had no effect on total cell surface binding of the Th-PN1 complexes at 4°C or on total cell association of the complexes at 37°C (Fig. 1B). These data indicate that the inhibitory effect of anti-(Pro 47 -Ile 58 ) IgG on Th-PN1 complex degradation is due to the specific inhibition of binding of the complexes to the LRP and that the binding of complexes to cell surface heparins occurs via a structural determinant that is distinct from the heparin-binding site.
To rule out the possibility that these data might be an artifact of antibody bivalency or the possibility that these complexes may be associated with another cell surface component in the presence of the antibody, we next directly demonstrated that the complexes are indeed bound to cell surface heparins in the presence of anti-(Pro 47 -Ile 58 ) IgG. The 4°C binding experiment shown in Fig. 1B was repeated, this time using complexes in the presence and absence of a 1:10 dilution of anti-(Pro 47 -Ile 58 ) IgG. At the end of the 3-h incubation, the cells were rapidly rinsed four times with ice-cold PBS, and 1 ml of binding medium was added back with or without 200 nM soluble heparin. The dissociation and release of the complexes in the medium was then followed over a 12-min time course. In the absence of soluble heparin, dissociation was not detectable either in the presence or absence of anti-(Pro 47 -Ile 58 ) IgG (Fig.  2). In contrast, dissociation in the presence of soluble heparin was extremely rapid, greater than 70% in 3 min. The addition of anti-(Pro 47 -Ile 58 ) IgG, at the same concentration that effectively inhibited complex degradation, had no effect on the dissociation rate. These data clearly demonstrate that anti-(Pro 47 -Ile 58 ) IgG does not inhibit binding of Th-PN1 complexes to the cell surface and that complex binding is indeed to cell surface heparins, because dissociation was accelerated by the addition of soluble heparin. The simplest interpretation of the data in Figs. 1 and 2 is that cell surface heparin binding and LRP binding are mediated by distinct structural determinants in PN1.
Characterization of PN1 Variants with Alanine Substitutions at His 48 and Asp 49 -Both of the PN1 variants, PN1(H48A) and PN1(H48A,D49A), were expressed in Sf9 insect cells under control of the polyhedron promoter in baculovirus and purified using Cibacron blue-Sepharose as described previously (20). Native PN1 was purified from the serum-free conditioned medium harvested from HF cells using heparin-Sepharose affinity chromatography (8). Active site titrations using thrombin in a chromogenic substrate assay were performed on both native PN1 and each of the variants. Based on the active site titrations and the actual protein concentration, the percentage of active protein was calculated (Table I). Very little difference in the activity of the protein preparations was measured. In fact, PN1(H48A) and PN1(H48A,D49A) were 74 and 76% active, respectively, which was slightly higher than native human fibroblast PN1 that displayed 70% activity. Using the activity measurements, the concentrations of each of the PN1 variants and native PN1 were then adjusted to equivalent active concentrations, and the k assoc were determined for each of the forms of PN1 as described previously (24). For each variant of PN1, the k assoc constants for thrombin were found to be similar to the k assoc constant of native PN1 for thrombin. The PN1(H48A) variant had the highest measured k assoc (6. 100-l aliquots of the culture supernatants were removed and precipitated in 12% trichloroacetic acid to measure the changes in trichloroacetic acid nonprecipitable radioactivity. Background trichloroacetic acid nonprecipitable radioactivity was determined as described under "Experimental Procedures" and subtracted from the mean of triplicate samples. B, remaining culture supernatants were removed by aspiration, and the cells were then washed four times with PBS. The cell monolayers were solubilized in 1 ml of 10% SDS, and radioactivity was quantified by ␥ counting. Error bars indicate the standard deviation from the mean of the triplicate samples in both panels. In addition to the k assoc constant and activity analyses presented above, which argue strongly that the introduced substitutions do not significantly affect the biochemical activities of PN1, we also examined the ability of the variant forms of PN1 to form complexes stable to SDS-PAGE with 125 I-thrombin. The characterization of the complex formation was critical, because any variation in the capacity of the PN1 variants to form stable complexes with thrombin due to the introduced substitutions could have a major effect on their LRP-mediated catabolism. Human recombinant native PN1, purified from Sf9 cells by the same method as each variant of PN1, was used for comparison. Shown in Fig. 3 is a digitized image of complexes formed between each form of PN1 and 125 I-thrombin. In each lane where recombinant native PN1 or a variant form of PN1 was present, greater than 90% of the thrombin appeared in complex with the PN1 or PN1 variant, demonstrating that each of the variants formed stable complexes with 125 I-thrombin at levels comparable with the recombinant native PN1. Note that some 125 I-thrombin becomes inactivated and unable to form complexes, probably due to oxidation during the radioiodination procedure, but all lanes showed a similar low level of this nonreactive thrombin. Cumulatively, the close similarity in activity, k assoc constants for thrombin, and ability to form SDSresistant high molecular weight complexes with 125 I-thrombin, argue that the alanine substitutions introduced at the positions of His 48 and Asp 49 do not affect the stability nor thrombin inhibitory activity of either of the PN1 variants.

The Heparin-mediated Binding of 125 I-Th-PN1 Variant Complexes to the Cell Surface Is Identical to Native 125 I-Th-PN1
Complexes-In recent studies (15), we have shown that the binding of Th-PN1 complexes to cell surface heparins acts synergistically to enhance LRP-mediated internalization. Heparin binding plays a similar role in the catabolism of thrombospondin 1 (25) and lipoprotein lipase (26). The aim of our next experiments was to measure the ability of the PN1 variant complexes to bind to cell surface heparins, because this could also significantly affect catabolism rates. Confluent cultures of HF cells in 24-well plates were incubated with native or variant 125 I-Th-PN1 complexes for 4 h at 4°C, in the presence and absence of the indicated concentrations of heparin (Fig. 4). At the end of the incubation, unbound ligand was removed, and the cultures were assayed for bound complexes by ␥ counting. The absolute binding of native 125 I-Th-PN1 complexes, 125 I-Th-PN1(H48A,D49A) complexes, and 125 I-Th-PN1(H48A) complexes were all virtually identical, as were the competition curves with soluble heparin. In all three cases, binding was inhibited by 50% at a heparin concentration of around 5 nM. Importantly, the concentration of complexes used in this experiment (200 ng/ml) is the same concentration used in the internalization and degradation studies presented below. From these results we determined that any change observed in the rate of catabolism was not due to a difference in the initial binding interaction of these PN1 variant complexes to cell surface heparins.
Alanine Substitutions at His 48 and Asp 49 Markedly Impair Th-PN1 Complex Catabolism-Based on our recent studies, which identified the peptide sequence 47 PHDNIVISPHGI 58 as a putative binding site in PN1 required for LRP-mediated  PN1(H48A,D49A) form complexes with 125 I-thrombin that are stable to SDS-PAGE. 600 ng of 125 I-thrombin were incubated with amounts of native PN1 and each of the recombinant variants required to achieve complete thrombin inactivation within 30 min at 37°C in a final volume of 300 l. At the end of the incubation, 5-l aliquots of the reactions were removed and added to 15 l of SDS-PAGE sample buffer. The reactions were analyzed by SDS-PAGE on 10% polyacrylamide gels to resolve free 125 I-Th from 125 I-Th-PN1 complexes. After drying, the gel was exposed to a Bio-Rad Phospho-Imager screen for 30 min. The digitized image was developed on a Bio-Rad GS-250 Molecular Imager. internalization and catabolism of 125 I-Th-PN1 complexes, we next evaluated the effect of alanine substitutions as His 48 and Asp 49 on 125 I-Th-PN1 catabolism. 125 I-Thrombin-PN1 complexes were formed from each variant of PN1 and from native PN1 and used at a final concentration of 200 ng/ml in binding medium as described under "Experimental Procedures." Complexes were added to confluent HF cell cultures in 24-well plates in triplicate. At the indicated time points, trichloroacetic acid nonprecipitable radioactivity in the media, which corresponds to 125 I-tyrosine released from lysosomal degradation of the radioiodinated thrombin (22) was determined (Fig. 5). Relative to native 125 I-Th-PN1 complexes, the 125 I-Th-PN1(H48A) variant complexes were degraded at about a 50% slower rate. 125 I-Th-PN1(H48A,D49A) variant complexes, in which the PN1 has an additional replacement of Asp 48 with Ala, reduced the rate of degradation to about 15% of native 125 I-Th-PN1 complexes. The control experiments presented in Figs. 3 and 4 and in Table I demonstrate that this differential rate of degradation is not due to a difference in PN1 inhibitory activity, complex stability, or a differential capacity to bind to cell surface heparins. Therefore, these data would strongly suggest that there is an altered interaction of the PN1 variant complexes with their receptor, LRP, which is required for the internalization of the complexes prior to intracellular degradation.
The LRP-mediated Internalization of 125 I-Th-PN1(H48A) and 125 I-Th-PN1(H48A,D49A) Complexes Is Sharply Lowered-The results shown in Fig. 5 measure the intracellular degradation of native and variant 125 I-Th-PN1 complexes as judged by an increase in trichloroacetic acid soluble radioactivity. The diminished degradation rates of 125 I-Th-PN1(H48A) and 125 I-Th-PN1(H48A,D49A) suggest an impaired interaction with the LRP but do not demonstrate this directly. To address this, we measured the internalization rates of 125 I-Th-PN1(H48A) and 125 I-Th-PN1(H48A,D49A) relative to native 125 I-Th-PN1 complexes (Fig. 6). Confluent cultures of HF cells in 24-well plates were incubated with native 125 I-Th-PN1 com-plexes, 125 I-Th-PN1(H48A) and 125 I-Th-PN1(H48A,D49A), each at a concentration of 200 ng/ml at 37°C. At the indicated times, ranging from 5 to 30 min, triplicates wells were rapidly chilled to 4°C and processed to remove cell surface bound complexes as described previously (8). Not surprisingly, the internalization data directly mirror the results of the degradation studies. Relative to native 125 I-Th-PN1 complexes, 125 I-PN1(H48A) were internalized at a 50% slower rate. 125 I-Th-PN1(H48A,D49A) complexes were internalized at only about 15% the rate of native 125 I-Th-PN1 complexes.

DISCUSSION
The present studies were undertaken to further probe the potential role of the PN1 peptide sequence, 47 PHD-NIVISPHGI 58 , in the LRP-mediated clearance of Th-PN1 complexes and to determine what structural features of this sequence might be important in this process. The potential importance of this sequence was discovered using a synthetic peptide library of PN1 sequences and screening for peptides that inhibited the catabolism of complexes by HF cells (10). The requirement for heparin-mediated binding of the complexes to the cell surface to promote an efficient interaction between the complexes and the LRP (15) precluded more detailed studies of this sequence using a synthetic peptide approach, because the sites in PN1 that interact with heparin and the LRP are distinct. In the present studies, two different approaches have been used to more clearly define the role of the structural determinant, Pro 47 -Ile 58 , in Th-PN1 complex catabolism and to identify critical residues within this determinant. Cell surface bound complexes were stripped at 4°C with a solution of EDTA and heparin as described previously (8). The remaining monolayers were solubilized in 10% SDS to determine the amount of internalized complexes. In both cases the radioactivity was quantified by ␥ counting. Error bars represent the standard deviation from the mean.
To independently confirm and more narrowly define the role of this structural determinant in Th-PN1 catabolism, an anti-(Pro 47 -Ile 58 ) polyclonal IgG was generated. This antibody specifically inhibited the internalization and subsequent degradation of Th-PN1 complexes but had no effect on the binding of complexes to cell surface heparins. These data independently confirm the role of Pro 47 -Ile 58 in LRP-mediated internalization, which had previously been based solely on synthetic peptide competition studies (10). In addition, they strongly suggest that the LRP and heparin-binding determinants in PN1 are structurally distinct, despite the fact that they act cooperatively to promote efficient complex catabolism (15).
In addition to the antibody studies, a genetic approach was used in which alanine substitutions were introduced at positions His 48 and Asp 49 of the intact PN1 protein. The rationale for these choices is derived from the predicted structural location of 47 PHDNIVISPHGI 58 in PN1 based on its homology to ␣-1-antitrypsin (1). This sequence is predicted to be at least partially interior to the protein surface. Pro 47 most likely represents the beginning of the transition sequence between helix A and strand 6B(1). His 48 and Asp 49 are transition amino acids, and Asn 50 through Pro 55 become strand 6 of sheet B. Given this structural information, we hypothesized that the charged residues, His 48 and Asp 49 , might be important for either direct interaction with the LRP receptor or assisting in the attainment of the active conformation of this structural determinant when fully exposed to the hydrophilic exterior of the molecule.
To evaluate the role of these specific amino acids we generated two different PN1 variants: one in which only His 48 was replaced by Ala and another where both His 48 and Asp 49 were replaced by Ala. Replacement of the first charged residue, His 48 by Ala, had a significant effect on the catabolism of PN1(H48A) complexes, reducing it by 50%. The additional substitution of Ala for Asp 49 , reduced complex catabolism by 85%. In both cases this was demonstrated to be due to a decreased rate of LRP-mediated internalization. Control studies revealed that the substitutions had no effect on the heparin-mediated binding of the complexes to the cell surface nor on the biochemical characteristics of the PN1 variant complexes with thrombin. Taken together, it is clear that both His 48 and Asp 49 play very important roles in the interaction of the PN1 structural determinant, 47 PHDNIVISPHGI 58 , with the LRP. Whether this is due to a direct interaction of these residues with the LRP or because these residues are required to establish a particular structural conformation of this determinant is presently unknown. Given the overall hydrophobic character of Pro 47 -Ile 58 , (PHDNIVISPHGI), it is likely that removal of the charged residues affects the solubility of the sequence. The genetic evidence using PN1 variants with point substitutions and the anti-(Pro 47 -Ile 58 ) antibody experiments represent two important and independent lines of evidence that support and extend the original observation that the sequence Pro 47 -Ile 58 in PN1 is required for the LRP-mediated clearance of Th-PN1 complexes.
A common pathway for many LRP internalized ligands seems to be emerging that involves cell surface proteoglycans and perhaps other accessory proteins in many cases. Several of the ligands first bind to cell surface heparin sulfate proteoglycans and are subsequently internalized by the LRP (15,(25)(26)(27). Even Th-ATIII complexes, which display a negligible affinity for heparin, use hepatic heparin sulfate proteoglycans as a part of their clearance mechanism but do so by an association with vitronectin (Vn) (14). Th-ATIII complexes first bind to Vn and then utilize the heparin-binding site in Vn, which is exposed only after it binds to Th-ATIII complexes (14). Recently, another accessory molecule, cytokeratin 18, has been shown to play an important role in the clearance of Th-ATIII-Vn ternary complexes (13). Antibodies specific for cytokeratin 18 were shown to markedly reduce the LRP-mediated internalization of Th-ATIII-Vn ternary complexes. Although the heparin-mediated pathway appears to be common for many LRP internalized ligands, there is at least one example where the involvement of heparin has not yet been documented (27). The uPA receptor binds complexes between high molecular weight urokinase and plasminogen activator inhibitor 1 or PN1 prior to LRP-mediated internalization (27). There are data suggesting that the uPA receptor, as well as its bound ligand, is co-internalized along with the LRP (16). Because plasminogen activator inhibitor 1 does contain a heparin-binding site, however, the potential involvement of heparin in this pathway should be examined more carefully.
Structural information on binding sites in ligands that interact with the LRP is, however, limited. The highest resolution studies have been done on the LRP-binding site in activated ␣ 2 -macroglobulin, which identified lysine residues 1370 and 1374 as essential for binding to the LRP (28). The LRPbinding site in thrombospondin 1 has been localized to the same amino-terminal fragment that contains the heparin-binding domain (25), and the carboxyl-terminal folding domain of lipoprotein lipase, which also contains the heparin-binding domain, has been implicated in binding to the LRP (26). Most recently, studies on the binding of plasminogen activator inhibitor 1-protease complexes to the LRP have implicated two residues located in the heparin-binding site in LRP binding (27). These data may be subject to alternate interpretations, because the opposite charge nature of the amino acid substitutions in plasminogen activator inhibitor 1 variants that resulted in a decreased LRP affinity may impart structural changes in the general region that are not necessarily part of the heparin-binding site (27). In addition, data supporting any type of a universal structural motif shared by all LRP ligands are not very compelling, because the ligands are diverse, and many of the ligands do not cross-compete for binding (28). Although the binding of several LRP ligands to cell surface heparins has been shown to play an important role in their clearance by the LRP, it remains to be determined whether the heparin and the LRP-binding sites overlap in some cases. The data in the present report, along with previously published studies using synthetic peptides strongly argue against this in the case of PN1.