The Low Density Lipoprotein Receptor-related Protein Modulates Protease Activity in the Brain by Mediating the Cellular Internalization of Both Neuroserpin and Neuroserpin-Tissue-type Plasminogen Activator Complexes*

Proteases contribute to a variety of processes in the brain; consequently, their activity is carefully regulated by protease inhibitors, such as neuroserpin. This inhibitor is thought to be secreted by axons at synaptic regions where it controls tissue-type plasminogen activator (tPA) activity. Mechanisms regulating neuroserpin are not known, and the current studies were undertaken to define the cellular pathways involved in neuroserpin catabolism. We found that both active neuroserpin and neuroserpin·tPA complexes were internalized by mouse cortical cultures and embryonic fibroblasts in a process mediated by the low density lipoprotein receptor-related protein (LRP). Surprisingly, despite the fact that active neuroserpin is internalized by LRP, this form of the molecule does not directly bind to LRP on its own, indicating the requirement of a cofactor for neuroserpin internalization. Our studies ruled out the possibility that endogenously produced plasminogen activators (i.e. tPA and urokinase-type plasminogen activator) are responsible for the LRP-mediated internalization of active neuroserpin, but could not rule out the possibility that another cell-associated proteases capable of binding active neuroserpin functions in this capacity. In summary, neuroserpin levels appear to be carefully regulated by LRP and an unidentified cofactor, and this pathway may be critical for maintaining the balance between proteases and inhibitors.

A variety of physiological processes in the brain are modulated by serine proteases, such as thrombin and tissue-type plasminogen activator (tPA). 1 Increasing evidence implicates an important role for tPA in synaptic plasticity and memory development. For example, tPA is induced as an immediateearly gene during long term potentiation (1) and during learning of a complex motor task (2). Furthermore, tPA is directly involved in late phase long term potentiation (3,4) through a mechanism that appears to involve its binding to the LDL receptor-related protein (LRP) (5). LRP is a large endocytic receptor that is a member of the LDL receptor family (6), and is highly expressed in the brain and other organs. LRP has diverse functions in a variety of processes including lipoprotein metabolism (7) and the homeostasis of proteases and their inhibitors (8,9).
The activity of serine proteases is regulated by inhibitors, which are often members of the serpin gene family. Neuroserpin is a member of the serpin family that is primarily expressed in neuronal and microglial cells in the brain (10,11). This inhibitor rapidly inhibits tPA (10), and is expressed in similar locations as tPA within the brain (2,10,(12)(13)(14), suggesting that tPA may be a target protease for neuroserpin. This is further supported by studies showing that injection of neuroserpin markedly delays the progression of seizure activity in wildtype, but not tPA-deficient, mice (15). The importance of neuroserpin in modulating neuronal function was revealed by deleting the neuroserpin gene in mice and demonstrating that neuroserpin-deficient animals show defective exploratory behavior and react abnormally to novel stimuli (16). Further, mutations in the neuroserpin gene are associated with an autosomal-dominant dementia and a form of progressive myoclonic epilepsy (17)(18)(19)(20)(21).
In the present investigation, we have characterized mechanisms that regulate levels of neuroserpin and identified receptors responsible for binding and mediating the catabolism of this molecule. The results demonstrate that both active neuroserpin and the tPA⅐neuroserpin complex are rapidly internalized in an LRP-dependent process. Surprisingly, active neuroserpin does not directly bind to LRP, implying that a cofactor molecule may facilitate this process.

EXPERIMENTAL PROCEDURES
Proteins and Antibodies-Neuroserpin was prepared in insect cells as described (10). The protein was labeled with [ 125 I]iodine (Amersham Biosciences) using IODOGEN (Pierce) to a specific activity of 3-25 Ci/g as described by the manufacturer. Fluorescence-labeled neuroserpin was prepared by conjugating neuroserpin with Alexa Fluor 546 dye (Molecular Probes, Eugene, OR) as described by the manufacturer. The cleaved form of neuroserpin was prepared by incubating 125 Ilabeled neuroserpin (800 nM) with 1.3 M tPA for 4 h at 37°C. Following incubation, SDS-PAGE followed by autoradiography confirmed that all of the 125 I-labeled neuroserpin was present as the cleaved form. LRP was purified from human placenta as described (22). Human singlechain tPA was purchased from Xtrana, Inc. (Broomfield, CO). Receptorassociated protein (RAP) was prepared as a fusion protein with glutathione S-transferase and was cleaved and purified as described (23). Rabbit anti-LRP R2629 (24) and mouse anti-LRP 5A6 (25) have been described. Mouse polyclonal anti-VLDLR IgG was prepared by immunizing VLDL receptor knockout mice with recombinant fragment of the human VLDL receptor prepared as described (26), and the antiserum was purified on protein G-Sepharose (Amersham Biosciences). Purified mouse IgG and purified rabbit IgG (Sigma) were used as controls for mouse anti-VLDLR and rabbit anti-LRP antibodies, respectively. All IgG (except Alexa Fluor-labeled anti-LRP 5A6) were heat-inactivated (30 min, 50°C) before use. For microscopy mouse anti-LRP IgG 5A6B6 were labeled with Alexa Fluor 488 (Zenon TM One IgG 1 labeling kit, Molecular Probes, Eugene, OR).
Solid Phase Binding Assay-96-well microtiter plates (Dynatech, Chantilly, VA) were coated with LRP or bovine serum albumin (BSA) (4 g/ml, 100 l/well) in 50 mM Hepes buffer, pH 7.5, with 1 mM CaCl 2 overnight at 4°C. The plates were then blocked with 3% BSA in Hepes, 5 mM CaCl 2 (300 l/well, 2 h at room temperature). Complex of neuroserpin and tPA was made by incubating the two proteins at a neuroserpin⅐tPA molar ratio of 1:1.6 (800:1330 nM) for 5 min at room temperature in Hepes, pH 7.5, with 3% BSA, 5 mM CaCl 2 . Neuroserpin was added to wells (100 l/well) in increasing concentrations (from 4 to 64 nM) in Tris-buffered saline, 3% BSA, 5 mM CaCl 2 , and 0.05% Tween 20. Preformed neuroserpin⅐tPA complex was added to wells to a final neuroserpin concentrations of 4 to 64 nM. The selectivity of the binding for LRP was assessed by measuring binding to wells coated with just BSA. LRP-and BSA-coated microtiter wells were incubated for 1.5 h at room temperature and then washed with the wash buffer (the blocking buffer with 0.05% Tween 20). The wells were broken apart, and their radioactivity was measured. Measurements were performed in duplicate.
Fluorescent Microscopy-The primary cortical cultures were used after 7 days of culturing on coverglasses (Fisher Scientific) coated with poly-L-lysine (Sigma). One hour before the experiment, cells were washed and preincubated with the assay media (Dulbecco's modified Eagle's medium, 50 mM Hepes, pH 7.5, 1% Nutridoma® NS media supplement (Roche Molecular Biochemicals), 0.5% BSA). Cells were incubated with fluorescently labeled (Alexa546) neuroserpin (20 nM) and/or Alexa Fluor 488-mouse anti-LRP IgG 5A6B6 for times indicated in figure legends. Cells were then washed with PBS, fixed in 4% formaldehyde and 5% sucrose in PBS for 15 min, washed with PBS, rinsed with distilled water, and mounted with FluorSave TM reagent (Calbiochem). For detection of the endogenous LRP fixed cells were permeabilized with 0.4% Triton X-100 for 5 min, blocked with 5% donkey serum for 1 h at 37°C, and incubated with rabbit anti-LRP 2629 antibody (25 g/ml) for 1 h at 37°C. The coverglasses were then incubated with rhodamine-conjugated donkey anti-rabbit IgG (25 g/ml) (Jackson Im-munoResearch Laboratories, West Grove, PA) for 1 h at 37°C, washed with PBS, rinsed with distilled water, and mounted with FluorSave TM reagent. Fluorescent and phase contrast images were obtained on a Nikon ECLIPSE E800 microscope using Nikon Plan Apo 100ϫ/1.4 oil objective and immersion oil type FF (Cargille Laboratories, Cedar Grove, NJ) at room temperature. Images were acquired with a Radiance 2100 laser scanning system (Bio-Rad) using Laser Sharp 2000 version 4.1 software (Bio-Rad) and processed using Adobe PhotoShop software.
Cell Internalization and Degradation Assays-Mouse fibroblasts were plated on 12-well tissue culture dishes (Corning Inc., New York, NY) coated with polylysine 1 day prior to the experiment and grown in Dulbecco's modified Eagle's medium (Mediatech, Washington, DC) supplemented with 10% fetal calf serum and penicillin/streptomycin. The primary cortical cultures were used after 7 days in culture. The experiments were performed when cells were ϳ80% confluent. One hour before the experiment, the cells were washed and preincubated with the assay media (Dulbecco's modified Eagle's medium, 50 mM Hepes, pH 7.5, 1% Nutridoma® NS medium supplement (Roche Molecular Biochemicals), 0.5% BSA). In certain experiments, the cells were incubated with 2 M RAP, 100 M chloroquine, the serine protease inhibitor 4-(2-aminoethyl)benzenesulfonyl fluoride (AEBSF), antibodies, or the assay media for 15-75 min at 37°C prior to the addition of 12 nM 125 I-neuroserpin (active, complexed with tPA, or cleaved) to the wells. The cells were incubated with the labeled neuroserpin with or without RAP or antibody for 10 min to 10 h at 37°C. Degradation is defined as the radioactivity in the cell culture medium that is soluble in 10% trichloroacetic acid. The amount of radioactivity in the absence of cells is subtracted. Trypsin-Versene mixture (BioWhittaker (a Cambrex company), Walkersville, MD) with proteinase K (0.05 mg/ml) were added to washed cells, and the cells were collected and centrifuged. Internalization of 125 I-neuroserpin was determined as radioactivity that is resistant to release from cells by trypsin with proteinase K (i.e. the radioactivity of the pellet). The cell numbers or total protein contained in well for each experimental condition were measured in parallel wells to normalize counts to cell number in well and to ensure that the treatment did not cause cell detachment. To analyze internalized radiolabeled protein, the cells were collected with trypsin-proteinase K, and then subjected to centrifugation in Eppendorfs with 10-fold molar excess of soybean trypsin inhibitor. Cell pellets were diluted in the cell extraction buffer (50 mM Hepes, 0.5 M NaCl, 1% Triton X-100, 0.05% Tween 20, pH 7.5) with Complete TM proteinase inhibitor mixture tablets (Roche Molecular Biochemicals). The extracts were subjected to SDS-PAGE under nonreducing conditions; the gels were dried and exposed to MS film.

RESULTS
The Neuroserpin⅐tPA Complex Is Relatively Unstable-Barker-Carlson et al. (30) previously demonstrated that tPA rapidly reacts with neuroserpin to form a covalent complex that can be detected as a higher molecular weight form upon SDS-PAGE. They found, however, that in contrast to other serpin-enzyme complexes, the neuroserpin⅐tPA complex readily dissociates releasing free tPA and cleaved neuroserpin. Prior to examining the binding of neuroserpin and neuroserpin⅐tPA complexes to purified LRP, we performed experiments to determine the extent of complex formation and stability under experimental conditions used for in vitro binding studies and cell uptake experiments. As reported by Barker-Carlson et al., we found that tPA rapidly reacts with 125 I-labeled neuroserpin to form a

125
I-Labeled neuroserpin (311 nM) was incubated with tPA (529 nM) in 50 mM Hepes, pH 7.0, at room temperature. At the indicated times, aliquots were removed and the reaction was stopped by addition of SDS-PAGE buffer, and the samples were analyzed by SDS-PAGE under nonreducing conditions. Following electrophoresis, the gel was dried and exposed to film. The position of molecular size markers (kDa) are indicated on the left, and the migration position of the complex, active, and cleaved forms of neuroserpin are indicated on the right. covalent complex evident by the slower mobility band upon SDS-PAGE ( Fig. 1) (30). However, only a portion of the total 125 I-labeled neuroserpin was detected as a covalent complex with tPA after incubation with a molar excess of tPA. With increasing time of incubation, significant amounts of cleaved forms of neuroserpin appeared, most likely arising from dissociation of the tPA⅐neuroserpin complex (30). For the experiments in the current study, we reacted neuroserpin with tPA for 5 min before diluting and adding to cells or microtiter wells coated with LRP. As evident from the analysis in Fig. 1, the neuroserpin⅐tPA complex contains significant amounts of both active and cleaved forms of neuroserpin.
Neuroserpin Binds to LRP in Vitro Only after Incubation with tPA-The ability of active neuroserpin and neuroserpin incubated with tPA (hereafter referred to as the neuro-serpin⅐tPA complex) to bind LRP was assessed by employing a solid phase binding assay. Purified LRP was immobilized in microtiter wells, and incubated with increasing concentrations of 125 I-labeled native neuroserpin ( Fig. 2A), 125 I-labeled cleaved neuroserpin (Fig. 2B), or 125 I-labeled neuroserpin⅐tPA complexes (Fig. 2C). The results demonstrate low binding of native and cleaved forms of 125 I-labeled neuroserpin to immobilized LRP (Fig. 2, A and B). In contrast, when tPA was premixed with 125 I-labeled neuroserpin prior to the assay, the binding of 125 I-labeled neuroserpin to LRP increased substantially (Fig.  2B). The increased binding of 125 I-neuroserpin to immobilized LRP when tPA is present suggests that the tPA⅐neuroserpin complex has much higher affinity for LRP than active or cleaved neuroserpin. This observation is consistent with previous studies examining the catabolism of antithrombin III, heparin cofactor II, ␣ 1 -antitrypsin, and PAI-1 by LRP (9, 31), where it was found that LRP only binds to the serpin⅐enzyme complex, and no binding of the active or cleaved serpin forms to LRP was detected. In the experiment shown in Fig. 2C, multi- ple species of neuroserpin are present in the mixture, not all of which bind to LRP. Thus, the affinity of the interaction is not possible to estimate from the data.
We also investigated whether RAP is able to block binding of 125 I-labeled neuroserpin⅐tPA to immobilized LRP. RAP is a known antagonist of LDL receptor family members and is capable of blocking binding of all known ligands to this class of receptors (23,(32)(33)(34). Our results demonstrate that RAP significantly reduces the binding of 125 I-labeled neuroserpin⅐tPA to immobilized LRP (Fig. 2D), confirming that RAP antagonizes binding of 125 I-labeled neuroserpin⅐tPA complexes to LRP like all other known ligands for this receptor.
Both 125 I-Neuroserpin and 125 I-Neuroserpin⅐tPA Complex Are Efficiently Internalized by Murine Primary Cortical Cultures in an LRP-mediated Process-Because neuroserpin is primarily expressed in neurons, we used cortical cultures from mouse embryos to assess whether 125 I-labeled neuroserpin and 125 I-labeled neuroserpin⅐tPA complexes are internalized by these cells. Several LDL receptor family members are expressed in the brain, and thus we subjected cell extracts from these cultures to immunoblot analysis. The results (Fig. 3) revealed that both LRP and the VLDL receptor are expressed in cells present in these cultures, whereas no LRP-2 (megalin) was detected in these cells (data not shown). Initial experiments were performed to characterize LRP localization and function in the primary cortical cultures employing fluorescent microscopy. The primary cultures of mouse cortex were plated on glass coverslips and maintained in culture for 7 days, during which time cells develop an elaborated network of axons and dendrites (Fig. 4A). To examine LRP localization and to confirm its function as an endocytic receptor in these cells, fluorescently labeled monoclonal antibody 5A6, which binds the LRP ectodomain on the ␤-subunit, was incubated with cell cultures for 7 min at 37°C to label the pool of receptors undergoing internalization from the cell surface. The cells were then fixed, permeabilized, and stained with a rabbit polyclonal anti-LRP IgG to detect total cellular LRP. Confocal fluorescent microscopy revealed that LRP is localized in punctuate structures throughout the cell body and neuronal processes (Fig. 4B), and we conclude from these data that the majority of LRP is located within compartments of the endocytic pathway. When fluorescence-labeled monoclonal antibody 5A6 was incubated with the cells, immunofluorescence was detected in LRP-positive endosomes on the cell periphery (Fig. 4C), demonstrating that LRP undergoes internalization from the surface of neuronal cells. Of   FIG. 3. Expression of LDL receptor family members in murine primary cortical cultures. Cell extracts were prepared from the murine primary cortical culture at day 7 following plating. Fifty g of total protein was subjected to SDS-PAGE under nonreducing conditions, transferred to nitrocellulose membrane, and the membrane was incubated overnight at 4°C with mouse polyclonal anti-VLDLR IgG (1 g/ml) or rabbit anti-LRP IgG (1 g/ml). The blots were washed and incubated for 1 h with goat anti-mouse IgG-horseradish peroxidase conjugate or goat anti-rabbit IgG-horseradish peroxidase conjugate. Bound antibodies were visualized by use of Super-Signal® West Pico chemiluminescent substrate. interest, the internalized 5A6 antibodies can be detected not only in neuronal somas but also in processes, revealing LRP activity in remote dendritic elements, such as branching regions and/or synapses (Fig. 4, B and C, insets).
We next investigated the ability of these cells to mediate the uptake of various forms of 125 I-labeled NS (Fig. 5A). The results reveal that cortical cultures efficiently internalized both 125 Ilabeled active neuroserpin and 125 I-labeled neuroserpin⅐tPA complexes, and their internalization was significantly reduced when excess RAP was included in the assay. The ability of RAP to block the internalization of both neuroserpin and neuroserpin⅐tPA complex indicates the involvement of an LDL receptor family member in this process. The results in Fig. 5A also demonstrate that cleaved 125 I-labeled neuroserpin was not effectively internalized by these cells, confirming that this form of neuroserpin is not recognized by LRP.
To assess which receptor is responsible for internalization, specific antibodies were utilized. Anti-LRP IgG blocked ϳ60% of the 125 I-neuroserpin internalized, whereas anti-VLDL receptor IgG blocked ϳ26% of the 125 I-neuroserpin internalized (Fig.  5B). These results indicate that both LRP and the VLDL receptor are major receptors for the internalization of 125 Ineuroserpin in primary cortical cultures. In the case of 125 I-neuroserpin⅐tPA complexes, anti-LRP IgG reduced internalization by ϳ40%, whereas anti-VLDL receptor IgG had no effect on internalization of this ligand (Fig. 5C). These results confirm that LRP also contributes to the internalization of neuroserpin⅐tPA complex in these cells, but that the VLDL receptor does not appear to be involved in the catabolism of neuroserpin⅐tPA complexes. The ability of RAP to completely block internalization of neuroserpin⅐tPA complexes suggests that other LDL receptor family members must be able to internalize neuroserpin⅐tPA complexes in these cells. At this time the identity of this receptor(s) remains unknown, but it is interesting to note that LRP1b, an LDL receptor family member closely related to LRP, is also expressed in the brain and has recently been shown to bind to tPA (35). Thus, LRP1b may also contribute to the internalization of NS⅐tPA complexes. Currently, blocking antibodies to LRP1b are not available to test this possibility.
A primary function of LRP and the VLDL receptor is to transport cargo into cells, where it is sorted to lysosomes for degradation. Thus, we also measured the ability of the cortical cultures to degrade internalized neuroserpin and neu-roserpin⅐tPA complexes. Once internalized, 125 I-labeled neuroserpin was degraded as measured by the appearance of acid soluble material secreted into the medium (Fig. 6A). The degradation was blocked by RAP, confirming the specificity of the process for LDL receptor family members. The degradation of 125 I-labeled neuroserpin was also inhibited by chloroquine, an inhibitor of lysosomally mediated degradation. Although the neuroserpin⅐tPA complex was internalized in a RAP-sensitive manner, unlike active neuroserpin, the complex was not as efficiently degraded (Fig. 6B). These results indicate that the trafficking of NS⅐tPA complexes differ significantly from that of active NS. The reason for this is not clear, and will require more experiments.
Internalized Neuroserpin and LRP Co-localize in Neuronal Cells-The fact that LRP mediates the internalization of active neuroserpin, yet does not seem to bind this molecule with high affinity is surprising. To further characterize the mechanism of LRP-mediated endocytosis of neuroserpin, we compared intracellular localization of internalized LRP (detected by internalized monoclonal antibody 5A6-Alexa488 conjugate) with that of internalized neuroserpin-Alexa546. Because monoclonal antibody 5A6 recognizes a region on the ectodomain of the light chain of LRP, and does not interfere with ligand binding by LRP, we were able to measure uptake of this antibody and neuroserpin simultaneously. The results of this experiment are shown in Fig. 7, and demonstrate that anti-LRP antibodies decorate neuronal processes with a characteristic punctuate pattern (Fig. 7A). The internalized neuroserpin is also found in a punctuate pattern (Fig. 7B) and co-localizes with LRP as revealed by the overlap of neuroserpin fluorescence with 5A6 fluorescence (Fig. 7C). Co-localization of anti-LRP antibodies and neuroserpin was also detected in endosomal compartments within neuronal somas (data not shown). Interestingly, the amount of internalized 5A6 antibody varied significantly between cells, suggesting different LRP activity within cells in the culture. Although the mechanism for this is not clear, importantly, cellular accumulation of anti-LRP antibody and neuroserpin always correlated, consistent with the role of LRP in neuroserpin uptake. 125 I-Labeled Neuroserpin and 125 I-Labeled Neuroserpin⅐tPA Complexes Are Not Internalized in LRP-deficient Fibroblasts-To confirm our antibody data suggesting that LRP is a major receptor mediating the internalization of neuroserpin and neuroserpin⅐tPA complexes, murine LRP-deficient fibroblasts were employed. When 125 I-labeled neuroserpin and NS⅐tPA complexes were incubated with wild-type mouse embryonic fibroblasts, these ligands were efficiently internalized (Fig. 8A), and their internalization was inhibited by RAP. However, in contrast, LRP-deficient mouse embryonic fibroblasts failed to internalize either neuroserpin or the neuroserpin⅐tPA complex (Fig. 8A). These studies confirm that LRP is capable of mediating the internalization of both 125 I-neuroserpin and 125 I-neuroserpin⅐tPA complexes. Further, the results confirm that, in murine fibroblasts, LRP appears to be the major receptor responsible for internalization of both of these ligands as mouse fibroblasts have little, if any, VLDL receptor (36).

The Internalization of 125 I-Neuroserpin by Mouse Embryonic Fibroblasts Is Not Mediated by Endogenous tPA or uPA-The
LRP-mediated internalization of 125 I-neuroserpin, seen in multiple cell types, was surprising because this form of the molecule does not appear to bind LRP in in vitro solid phase assays ( Fig. 2A). Thus, we suspected that during the assay active neuroserpin formed a complex with endogenously produced proteases, such as tPA or uPA, and was then internalized as a complex. To determine whether endogenously produced tPA or uPA mediates neuroserpin uptake, fibroblasts were cultured from mice genetically deficient in either tPA (tPA Ϫ/Ϫ ) or uPA (uPA Ϫ/Ϫ ) and from mice genetically deficient in both tPA and uPA (tPA Ϫ/Ϫ ,uPA Ϫ/Ϫ ). Immunoblot analysis confirmed that LRP expression in these cells is comparable with wild-type fibroblasts (Fig. 8B). Internalization assays were performed using these cell lines in serum-free assay media, and the results reveal that tPA Ϫ/Ϫ , uPA Ϫ/Ϫ , and tPA Ϫ/Ϫ uPA Ϫ/Ϫ cells all mediate the internalization of both 125 I-neuroserpin and 125 I-neuroserpin⅐tPA complexes (Fig. 8A). In all cases, the internalization was significantly reduced when excess RAP was included in the incubation. Thus, these results indicate that internalization of active neuroserpin via LRP is not dependent upon endogenously produced tPA or uPA.
Active Forms of Neuroserpin Are Internalized by LRP-To confirm that the active form of neuroserpin can be internalized in cells, we first investigated the effect of protease inhibitors on the ability of cortical neurons to mediate the internalization of 125 I-labeled neuroserpin. The results of this experiment indicate that the serine protease inhibitor AEBSF had no effect on neuroserpin internalization by primary mouse cortical cultures (Fig. 9A). As a control for these experiments, 4-(2-aminoethyl)benzenesulfonamide (AEBS) was also shown to have no effect on neuroserpin internalization. This compound is structurally related, but lacks inhibitory capability. As AEBSF should inhibit serine protease activity and therefore prevent formation of neuroserpin complexes with serine proteases, the results support the notion that active forms of neuroserpin are being internalized by these cell lines.
We next evaluated the forms of 125 I-labeled neuroserpin (ac- tive, cleaved, or complexed) that are actually internalized. In these experiments, mouse embryonic fibroblasts were incubated with 125 I-labeled active neuroserpin. The cells were then washed and the cell surface-bound neuroserpin removed by treatment with trypsin-proteinase K solutions. Next, cell extracts were prepared and subjected to SDS-PAGE under non-reducing conditions, and analyzed by autoradiography. The results show that active neuroserpin is the major form of the molecule internalized (Fig. 9B, lane 2). Significantly, we did not detect any complexed form of neuroserpin internalized in this experiment, confirming that proteases are not required for the internalization of active neuroserpin. A small amount of  4) for 15 min at 37°C. 12 nM 125 I-neuroserpin was then added, and the incubation extended for 75 min in the absence of presence of the indicated molecules. Following this period, cells were washed, collected from plates with trypsin-proteinase K solution, and centrifuged. Proteins present in cells were separated by SDS-PAGE on 4 -20% gradient gels, and following electrophoresis the gels were dried and exposed to the MR film. Lane 1 shows the neuroserpin in media that was added to the cell cultures. cleaved neuroserpin along with additional lower molecular weight fragments were also detected within the cells. Considering that the cells do not readily internalize the cleaved form of 125 I-labeled neuroserpin, but that this form appears within cells following LRP-mediated internalization, we conclude that the cleaved form of neuroserpin is generated by cell-associated proteases following internalization of active neuroserpin. As expected, in the presence of RAP, no labeled ligand was internalized (Fig. 9B, lane 3) confirming the receptor-mediated specificity of this process. The lower molecular weight forms of neuroserpin was reduced by treatment of the cells with chloroquine, an inhibitor of lysosomal-mediated degradation indicating that the lower molecular weight forms are generated by lysosomal mediated degradation of neuroserpin (Fig. 9B, lane  4). Other protease inhibitors, such as lactacystin, had no effect (data not shown). Overall, these results show that neuroserpin is internalized in its active form. DISCUSSION It is clear from a variety of studies (4, 37-39) that perturbation of the balance between proteases and their inhibitors in the brain has significant consequences in neuronal function. For example, overexpression of protease nexin-1 (PN-1), a serpin expressed in neurons that primarily inhibits thrombin and tPA, increases long term potentiation (LTP) and alters motor behavior and sensorimotor integration. In confirmation of these results, deletion of the PN-1 gene in mice results in decreased LTP (37,38). Interestingly, both PN-1 overexpressors and PN-1 knockout mice develop epileptic activity (37,38). Like PN-1, neuroserpin also plays an important role in neuronal function. Thus, deletion of the neuroserpin gene in mice results in mice that display defective exploratory behavior and react inappropriately to novel stimuli (16). Excess neuroserpin appears to be neuroprotective, and, when added at pharmacological doses, neuroserpin reduce cerebral infarct volume and protect neurons from ischemia-induced apoptosis in a rat model of stroke (40). Together, these studies confirm the important role that protease inhibitors play in neuronal function.
In the current study, we found that LRP modulates neuroserpin levels by mediating the cellular internalization of both neuroserpin and neuroserpin⅐tPA complexes. The ability of LRP to mediate the rapid internalization of active neuroserpin indicates that levels of this serpin are carefully regulated. Interestingly, both neuroserpin (41) and LRP (42) are concentrated in synaptic regions and are therefore localized at appropriate sites to modulate protease turnover. Disruption of this regulation could contribute to synaptic degeneration. We hypothesize that these receptor-mediated pathways are critical for maintaining the balance of protease activity by regulating levels of both proteases and their inhibitors.
Internalization of the active form of a serpin by LRP is unusual, as all other inhibitory serpins studied to date are only internalized as the serpin-enzyme complexes (9,31). Further, active neuroserpin does not appear to bind to LRP with high affinity in vitro. We ruled out the possibility that endogenously produced tPA (or uPA) was responsible for the internalization of neuroserpin by demonstrating that fibroblast cell lines derived from mice genetically deficient in tPA, uPA or in both tPA and uPA were fully capable of internalizing 125 I-labeled neuroserpin. Further, SDS-PAGE and autoradiography of extracts from cells incubated with 125 I-labeled neuroserpin confirm that the active form of neuroserpin was indeed internalized. These results suggest that an unidentified accessory molecule is required for this process. Alternatively, it is possible that active neuroserpin undergoes alterations, such as polymerization, when added to cells resulting in increased avidity for LRP. Interestingly, active forms of recombinant neuroserpin un-dergo polymerization when incubated at 37°C (43).
Experiments with specific antibodies also suggest that another LDL receptor family member, the VLDL receptor, also participates in the internalization of active neuroserpin, but not neuroserpin⅐tPA complexes. Like LRP, the VLDL receptor recognizes a variety of ligands, and is also expressed in neurons. The VLDL receptor participates in a signal transduction pathways mediated by reelin (44 -46), a molecule secreted by Cajal-Retzius cell in the outermost layer of the cerebral cortex that controls the final position of neurons that migrate from the ventricular zone. Binding of reelin to the VLDL receptor induces tyrosine phosphorylation of disabled-1 (Dab-1) (45,46), an adaptor protein that interacts with the cytoplasmic domains of LDL receptor family members (47,48), and functions in tyrosine kinase signaling pathways. Currently it is not known whether neuroserpin modulates signaling of this pathway by inhibiting reelin/VLDL receptor interactions.
Although the exact target protease for neuroserpin is currently unknown, considerable evidence suggest that tPA may be an important in vivo target enzyme for neuroserpin. Neuroserpin is a rapid inhibitor of tPA activity (10), and both tPA and neuroserpin are expressed in similar regions in the central nervous system (11). Further, immunoblot analysis of mouse brain extracts with an antibody recognizing neuroserpin identified a 110-kDa species, which is the expected size of the neuroserpin⅐tPA complex (10). Finally, neuroserpin protects from seizure progression in wild-type, but not tPA-deficient mice, suggesting a role in regulating tPA activity (15). Several studies in recent years indicate an important role for tPA in the brain. Thus, in situ hybridization revealed that expression of tPA messenger RNA was increased in the Purkinje neurons of rats within 1 h of their being trained for a complex motor task (2). Studies in tPA-deficient mice reveal that tPA activity in the amygdala is critical for stress-induced anxiety-like behavior (39). Further, tPA-deficient mice manifest a selective reduction in L-LTP in hippocampal slices in both the Schaffer collateral-CA1 and mossy fiber-CA3 pathways (3,4), and can be restored by adding exogenous tPA. The association of tPA with LRP appears to be important in restoring this effect, because RAP inhibits exogenously added tPA-mediated LTP (5). Insight into possible mechanisms comes from the fact that association of tPA with LRP increases protein kinase A activity, which in turn might alter LTP, as protein kinase A is known to be involved in LTP. The signaling roles of LRP in response to tPA and perhaps other ligand binding as well may be important for normal synaptic plasticity. Very likely, neuroserpin modulates this process by inhibiting tPA, although its effect on this process is not yet known.
In summary, our results identify a receptor-dependent mechanism for the internalization of neuroserpin that requires LRP. The rapid internalization of active forms of neuroserpin by LRP indicates that levels of this serpin are carefully regulated at synaptic regions. It is likely that these receptor-mediated pathways are critical for maintaining the balance of protease activity by regulating levels of both proteases and their inhibitors, which in turn affect synaptic integrity and function.