Apolipoprotein E Receptors Mediate the Effects of β-Amyloid on Astrocyte Cultures*

We have previously shown that β-amyloid (Aβ) induces astrocyte activation in vitro and that this reaction is attenuated by the addition of exogenous apolipoprotein E (apoE)-containing particles. However, the effects of Aβ on endogenous apoE and apoJ levels and the potential role of apoE receptors in astrocyte activation have not been addressed. Three activating stimuli (lipopolysaccharide, dibutyryl cAMP, and aged Aβ 1–42) were used to induce activation of rat astrocyte cultures, as assessed by changes in morphology and an increase in interleukin-1β. However, only Aβ also induced ∼50% reduction in the amount of released apoE and apoJ and an 8-fold increase in the levels of cell-associated apoE and apoJ. Experiments using two concentrations of receptor-associated protein, an inhibitor of apoE receptors with a differential affinity for the low density lipoprotein receptor (LDLR) and the LDLR-related protein (LRP), suggest that LRP mediates Aβ-induced astrocyte activation, whereas LDLR mediates the Aβ-induced changes in apoE levels. Receptor-associated protein had no effect on apoJ levels or on activation by either dibutyryl cAMP or lipopolysaccharide. These data suggest that apoE receptors translate the presence of extracellular Aβ into cellular responses, both initiating and modulating the inflammatory response induced by Aβ.

ciated with amyloid plaques. Although the precise relationship between amyloid plaques and dementia remains unclear, genetic and experimental evidence suggests that ␤-amyloid (A␤) plays a critical role in AD. A␤ may initiate or exacerbate neuropathology by inducing glial activation, thereby promoting the release of inflammatory response compounds, including cytokines, nitric oxide, and other potentially neurotoxic agents.
Glia, in particular astrocytes, are the primary cell type in the central nervous system that synthesize apoE, whereas apoJ is expressed by glia and neurons (reviewed in Ref. 2). We have previously reported (3) that rat astrocytes secrete high density lipoprotein-like lipoprotein particles with apoE and apoJ as the primary protein components. In the periphery, apoE-containing lipoproteins participate in lipid and cholesterol transport, including the delivery of lipoprotein constituents to tissues expressing lipoprotein receptors that recognize apoE as a ligand. This process may also be operating in the parenchyma of the brain because neural cells express a variety of apoE receptors in the low density lipoprotein receptor (LDLR) family (4 -7). The role of apoJ in lipid transport in both the periphery and within the central nervous system is less clear, and megalin/LRP2, the only known receptor for mammalian apoJ, appears to be expressed only by ependymal and endothelial cells in the brain (8,9).
Several lines of evidence suggest that apoE and apoJ may be involved in neural homeostasis beyond their capacity to transport lipid. Both apoE and apoJ increase in response to neural injury or disease (10 -13). In addition, these proteins may play a role in the pathogenesis of AD, because apoE and apoJ immunoreactivity is localized to senile plaques (14,15), and both proteins appear to interact with A␤. In vitro, apoE and apoJ form stable complexes with A␤ (11, 16 -20), alter the aggregation of various A␤ peptides (21)(22)(23), and affect A␤ neurotoxicity (24 -27). In humans, apoE exists as three naturally occurring isoforms (apoE2, apoE3, and apoE4), and apoE4 is a risk factor for AD via a mechanism as yet unknown. One hypothesis is that central nervous system lipoproteins containing apoE and/or apoJ provide a vehicle for clearing A␤ via lipoprotein receptors (6,17,28).
Increasing evidence suggests that apoE receptors may be involved in neural cell processes in general and in the pathophysiology of AD in particular. First, neural cells express a variety of endocytic receptors in the LDLR family, with the LDLR expressed by glia (5,6), LDLR-associated protein (LRP) associated primarily with neurons and activated astrocytes (6,29,30), apoE receptor 2 (ER2) immunostaining neurons (7,31), and very LDLR immunostaining primarily neurons and activated microglia (4). Second, apoE3 enhances neurite outgrowth in vitro by a mechanism requiring LRP (32,33). Third, LRP may play a role in the metabolism of amyloid precursor protein, because LRP has been shown to mediate the endocytosis of a secreted form of amyloid precursor protein (34). Fourth, immunoreactivity for LRP and a number of its ligands including apoE and ␣ 2 -macroglobulin is found associated with senile plaques (30). Finally, genetic evidence suggests that polymorphisms in either LRP or ␣ 2 -macroglobulin increase the risk of late onset familial AD (35,36).
We have previously demonstrated (37) that aged preparations of A␤ 1-42 induce activation of primary rat astrocyte cultures, as measured by changes in morphology and an increase in IL-1␤ mRNA. This activation is inhibited by the addition of exogenous apoE-containing particles (38). However, the effects of A␤-induced astrocyte activation on endogenous apolipoproteins have not been reported. The current experiments were designed to determine whether astrocyte activation altered the expression of endogenous apoE and apoJ. In addition, the role of apoE receptors in mediating A␤-induced changes in astrocytes is unknown. By immunostaining, the activated astrocytes used for the present experiments express both the LDLR and LRP, consistent with previous observations (6,29,30). To distinguish between the effects mediated by the LDLR and LRP, we utilized receptor-associated protein (RAP), an antagonist with different binding affinities for these two receptors (39,40). We report here that A␤ induced a dramatic increase in endogenous apoE and apoJ levels in activated astrocytes and that RAP abolished the A␤-induced changes in astrocyte activation and apoE levels but had no effect on changes in apoJ. Our data suggest that LRP mediates A␤induced astrocyte activation, whereas the LDLR mediates the A␤-induced changes in apoE levels.

EXPERIMENTAL PROCEDURES
Materials-The A␤ 1-42 peptide was obtained from California Peptide Research, Inc. (Napa, CA) or from Dr. Charles Glabe (University of California at Irvine). The peptide was dissolved in 10 mM HCl to make a 2 mM stock solution and then diluted 1:20 into PBS. This 0.1 mM A␤ solution was stored at room temperature for 48 h before being added to the cells at a final A␤ concentration of 10 M. A␤ 1-42 aged according to this protocol is a mixture of fibrils and soluble globular aggregates, and is active to astrocytes (37,38) and toxic to neurons (26,27). Lipopolysaccharide (LPS), dibutyryl cyclic AMP (dbcAMP), heparin, and heparan sulfate were purchased from Sigma.
Astrocyte Cultures and Cell Treatment-Astrocyte cultures were prepared from the cerebral cortex of 1-2-day-old neonatal rats (Harlan Sprague-Dawley) as described previously (41). After 11 days in culture, cells were trypsinized and replated into 100-mm tissue culture plates at a density of ϳ6 ϫ 10 5 cells/plate. After growing to confluency, cells were trypsinized and seeded into 12-well tissue culture plates at a density of 1 ϫ 10 5 cells/well. After 24 h, cells were washed twice with PBS to remove serum and then incubated in ␣-minimum essential medium containing N2 supplements (Life Technologies, Inc.) for an additional 24 -48 h before treatment. As previously reported (37,41) these tertiary cultures are ϳ95-98% astrocytes, with only ϳ2-5% microglial cells. Cells were treated with aged A␤ 1-42 (10 M), dbcAMP (1 mM), LPS (10 g/ml), or control buffer for the desired length of time.
RAP was generated as a recombinant glutathione S-transferase fusion protein and purified as described previously (42). Before use, RAP was dialyzed against culture medium, as was a comparable molar concentration of bovine serum albumin to serve as a control. For experiments utilizing RAP, it was added to the cells 1 h before A␤ treatment. The bovine serum albumin-only control had no effect on the activation state of the cells in either the presence or the absence of A␤ (data not shown).
For experiments utilizing heparin or heparan sulfate, it was added to cells 1 h before A␤ treatment. Heparin and heparan sulfate were used at a final concentration of 100 g/ml and 1 M, respectively (43).
Examination of Astrocyte Morphology-Cell morphology was examined and morphological activation quantified as described previously (37). Briefly, cells were considered activated if they had a process greater in length than the diameter of the cell body.
Slot Blots-Total RNA was isolated, and slot blots were run as described previously (37). The rat apoE probe corresponds to nucleotides 396 -724 and was obtained from pALE (45).
Data Analysis-The difference between two groups of data was analyzed with variance F tests, and the means were analyzed with Student's t test. Statistical significance was established at a level of p Ͻ 0.05.

RESULTS
Astrocyte Activation-To determine the relationship between astrocyte activation and apolipoprotein levels, we first examined three standard activating stimuli for their ability to activate cultured rat astrocytes. Astrocytes were treated with A␤ 1-42 (10 M), dbcAMP (1 mM), LPS (10 g/ml), or PBS control buffer, and activation was assessed by morphology. As shown in Fig. 1A, control cells showed the typical morphological features of quiescent astrocytes in culture, being a monolayer of flat and polygonal-shaped cells. In contrast, incubation of cells with A␤, dbcAMP, or LPS for 12 h induced a marked change in cell morphology. Astrocytes became stellate-shaped, with a more spherical and phase bright cell soma and two or more processes. Quantitation of the morphological activation ( Fig.  1B) showed a significant activation by all three stimulating agents. This morphological alteration was time-dependent, with activation evident by 6 h and reaching a peak at 12 h after addition of the stimuli (data not shown), as we previously reported for A␤ 1-42 (37,38).
In addition to changes in morphology, glial activation was assessed by induction of pro-inflammatory cytokines. We have demonstrated previously that treatment of astrocyte cultures with A␤ 1-42 induces an increase in IL-1␤ mRNA levels (37). For the present study, we examined the levels of cell-associated IL-1␤ (the proIL-1␤ form of the protein can be detected in cell lysates). As shown in Fig. 1C, cells treated with A␤, dbcAMP, or LPS exhibited an increase in proIL-1␤ levels compared with PBS-treated cells. In agreement with our previous report (37), the GFAP levels did not change upon activation (Fig. 1C). The stimulation of IL-1␤ protein levels peaked at 12 h after treatment and gradually decreased by 48 h (data not shown). These data demonstrate that cultured rat astrocytes can be activated by A␤, dbcAMP, and LPS.
Effects of Astrocyte Activation on apoE and apoJ-To assess the effects of astrocyte activation on endogenous apoE and apoJ, cells were treated with the three activating stimuli or control buffer for 12 h, and the levels of apoE and apoJ protein in conditioned medium and cell lysates were analyzed by Western blots. Under reducing conditions, apoE appears as ϳ35-kDa monomer. ApoJ is synthesized as an ϳ80-kDa holo-protein that is cleaved during processing to two 40-kDa subunits that reassociate via disulfide bonding. Thus, the 80-kDa apoJ-immunoreactive band seen on gels represents primarily uncleaved holo-protein, because the cysteine-linked subunits are resolved to the 40-kDa species under the reducing conditions of the gel. It is interesting to note the presence of the 80-kDa species in the medium, suggesting that at least a portion of the apoJ is secreted in an uncleaved form. For the purposes of this study, however, the reported changes in apoJ protein refer to the amount of the 40-kDa subunit.
A␤ induced a robust increase in the levels of cell-associated apoE and apoJ (Fig. 2, A and B). The mean levels of cellassociated apoE and apoJ from four independent experiments were approximately 8-fold higher in A␤-treated cells relative to control cells. In contrast to the large A␤-induced increase in cell-associated apoE and apoJ, the levels of apoE and apoJ in conditioned medium decreased after treatment of cells with A␤. As shown in Fig. 2 (C and D), the levels of apoE and apoJ in conditioned medium from A␤-treated cells decreased by a mean of 35% (apoE) and 60% (apoJ) relative to control conditioned medium. Neither dbcAMP nor LPS altered the levels of cellassociated apoE and apoJ (Fig. 2, A and B) or the levels of apoE in the conditioned medium (Fig. 2C). However, there was an increase in apoJ levels in conditioned medium from LPStreated cells (Fig. 2D). As can be seen in the inset of Fig. 2D, LPS treatment consistently resulted in a number of immunoreactive bands in addition to the apoJ subunits and holo-protein. This pattern was not seen with other activating stimuli, and this observation was not pursued further as part of this study.
To begin to address the mechanism by which A␤ leads to increased levels of cell-associated apoE and apoJ, we measured mRNA levels for apoE and apoJ in control and A␤-treated astrocytes. As shown in Fig. 3, there was no difference in apoE or apoJ mRNA levels, expressed relative to glyceraldehyde-3phosphate dehydrogenase mRNA. This suggests that A␤ does not induce the accumulation of cell-associated apoE via an increase in transcription. However, we did observe the expected A␤-induced increase in IL-1␤ mRNA levels, as we previously reported (37).
RAP Blocks A␤-induced Activation and Changes in apoE but Not apoJ-By immunostaining, these activated astrocytes express both the LDLR and LRP (data not shown), consistent with the previous observations summarized above that these are the two primary members of the LDLR family that are expressed by activated astrocytes (4 -7, 29 -31). We explored further the mechanisms by which A␤ activates astrocytes and stimulates apoE and apoJ levels by testing the effect of the apoE receptor antagonist RAP. The binding affinity of RAP for LRP is 3.3 nM (46), whereas the K d of RAP for the LDLR is 250 nM (47), thus making it possible to distinguish effects mediated by these two apoE receptors. Two concentrations of RAP were tested: a high concentration (1 M) to inhibit both LRP and LDLR and a low concentration (70 nM) to inhibit LRP but not LDLR. We found that both concentrations of RAP blocked A␤-induced morphological activation (Fig. 4, A and B) and attenuated the A␤-induced increase in proIL-1␤ levels (Fig. 4,  C and D). This evidence suggests that LRP mediates A␤-induced astrocyte activation. Treatment of cells with 1 M RAP did not inhibit dbcAMP-or LPS-induced activation (data not shown), and RAP had no effect on the activation state of the cells in the absence of A␤ (Fig. 4). As illustrated in Fig. 5, we observed a differential effect of RAP on A␤-induced changes in apoJ and apoE levels. RAP at 1 M concentration had no effect on either the A␤-induced increase in cell-associated apoJ (Fig.   FIG. 1. Activation of cultured astrocytes by A␤ 1-42 5A) or the decrease in released apoJ (Fig. 5D). In contrast, 1 M RAP blocked both the increase in cell-associated apoE (Fig. 5B) and the decrease in apoE in the conditioned medium (Fig. 5E). However, 70 nM RAP did not block either of these A␤-induced changes in apoE (Fig. 5, C and F), suggesting that the LDLR is mediating these effects.
In addition to apoE receptors, RAP also binds directly to heparan sulfate proteoglycans, as does apoE. Thus, RAP inhibition of the A␤-induced increase in cell-associated apoE could be due to competition with apoE for binding to heparan sulfate proteoglycans not apoE receptors. To address this issue, we incubated cells with A␤ in the presence and absence of either heparin or heparan sulfate (Fig. 6). Neither compound blocked the A␤-induced increase in cell-associated apoE, suggesting that heparan sulfate proteoglycans are not involved in this activity and supporting our conclusion that apoE receptors, specifically the LDLR, mediates this effect. DISCUSSION We have demonstrated here that exposure of rat astrocyte cultures to A␤ 1-42 results in three phenomena: a morphological activation also monitored by increased IL-1␤ levels, an increase in cell-associated apoE, and an increase in cell-associated apoJ. Because all three phenomena are induced by the same concentration and preparation of peptide, it would seem natural to conclude that the three events are related. However, the use of RAP as a probe at two different concentrations, chosen based on its affinity for the two LDLR family members known to be present in these cells, suggests that these three phenomena can be at least partially dissociated from one another. A␤-induced astrocyte activation appears to be mediated by LRP based on its inhibition by a low concentration of RAP. A␤-induced accumulation of cell-associated apoE appears to be mediated by the LDLR, based on its inhibition only by a high concentration of RAP. In contrast, the A␤-induced accumulation of cell-associated apoJ does not appear to involve RAPinhibitable apoE receptors. These data suggest that both LRP and the LDLR can translate the presence of extracellular A␤ into cellular responses (Fig. 7). Thus, in addition to the receptor for advanced glycation end products and the scavenger receptor (48,49), we propose that apoE receptors mediate certain of the glial cell changes induced by A␤, whether directly or indirectly. Because a number of receptors in the LDLR family are expressed in the brain (4 -7, 29 -31), it is possible that receptors other than the LDLR and LRP are also involved in mediating the effects of A␤ on neural cells.
A␤-induced changes in cultured astrocytes appear to involve both a novel mechanism of action and a unique set of responses. Activation induced by dbcAMP and LPS is independent of apoE receptors, whereas A␤-induced activation appears to require LRP. This suggests that LRP is linked to a signal transduction pathway that ultimately leads to activation. The data presented here do not address whether the A␤/LRP pathway utilizes the same downstream activation signaling events as LPS and dbcAMP. In terms of novel responses, only A␤ induced alterations in apoE and apoJ, changes that appear to be independent of activation. The A␤-stimulated changes in apoJ persist in the presence of 1 M RAP and the changes in apoE persist in the presence of 70 nM RAP, both treatments that block morphological activation. These data also suggest that, whereas the A␤-induced changes in apoE appear to require the LDLR, the changes in apoJ are independent of RAP-inhibitable apoE receptors.
The LDLR may be involved in the A␤-induced changes in apoE in two distinct capacities: directly via an increased uptake of released apoE and indirectly via signal transduction that leads to an increase in intracellular apoE levels. An increase in apoE reuptake may be the result of an increase in the number of apoE receptors. Alternatively, an increase in the receptor binding affinity of apoE in the presence of A␤ above that of apoE alone may facilitate reuptake (50). However, an increased reuptake of apoE by the LDLR accounts for only a portion of the accumulation of cell-associated apoE. We show a ϳ50% decrease in apoE in the conditioned media and an ϳ8fold increase in cell-associated apoE. Normalizing for the different volumes of media versus cell lysates used for analysis, there is a 2-3-fold net increase in the total amount of apoE in activated astrocytes. This suggests that the actual amount of apoE increases, possibly via an A␤-activated signaling mechanism linked to the LDLR. A number of cellular and molecular events may contribute to the increase in apoE, including alterations in post-transcriptional mechanisms, such as the apoE turnover rate. However, the steady-state levels of apoE mRNA did not change, suggesting that transcriptional regulation does not play a major role (Fig. 3).
The hypothesized signaling mechanisms involved in the A␤induced changes in both astrocyte activation mediated by LRP and apoE levels mediated by the LDLR may involve A␤ interacting directly with apoE receptors or indirectly via an association with an apoE receptor ligand. ApoE, apoJ, and most recently ␣ 2 -macroglobulin have been shown to form a complex with A␤ that may facilitate clearance of the peptide (25-27, 50, 51). It is also possible that a complex between A␤ and an apoE receptor ligand triggers a unique intracellular signaling event that produces the change in apoE.
Both the mechanism for and function of the A␤-induced accumulation of cell-associated apoJ are unclear. As a receptor ligand, apoJ is probably not involved in the present in vitro system because apoJ is not a ligand for apoE receptors, and megalin, the only identified apoJ receptor, does not appear to be expressed by either glial cells or neurons (8,9). Although we did not detect the presence of megalin by immunostaining, it is possible that glial cells, particularly activated astrocytes, express an as yet unidentified apoJ receptor. Alternatively, in the presence of A␤, apoJ may function as a ligand for other apoE receptors that are known to be expressed by glial cells. ApoJ has been shown to potentiate the formation of a neurotoxic species of A␤ (26,27). Thus, the intracellular sequestration of apoJ may be an adaptive function that limits the activity of A␤. In addition, apoJ may be involved in the transport of A␤ at the blood brain barrier, because megalin is expressed by ependymal and epithelial cells (9). Finally, the function of apoJ in astrocytes may be independent of its role as an apoE receptor ligand. For example, apoJ, also known as clusterin and SP40 -40, may be acting in its role as a complement inhibitor.
Several reports suggest that apoE receptors may be involved in modifying the activity of A␤ in neural cells. The addition of exogenous apoE protects against A␤-induced toxicity in neuronal cell cultures, as well as A␤-induced activation of astrocyte cultures (24,25,38,52,53). For example, apoE3 but not apoE4 protects against A␤-induced neurotoxicity of rat hippocampal neurons, a process inhibited by 1 M RAP (25). As discussed above, apoE receptors may be involved in the actual uptake of any A␤ associated with apoE-containing particles, thus provid- FIG. 7. Model for the proposed role of apoE receptors in A␤induced astrocyte activation. LRP mediates A␤-induced changes in astrocyte morphology and increased expression of IL-1␤. LPS and db-cAMP produce similar effects independent of apoE receptors. The LDLR mediates the A␤-induced increase in cell-associated apoE and the decrease in apoE in the conditioned medium. Although a portion of these changes may be the result of increased reuptake of apoE by the LDLR, the total amount of apoE also increases. This suggests that the LDLR may be linked to an intracellular pathway that signals an increase in apoE translation and/or a decrease in degradation (steady-state apoE mRNA levels do not change). To effect these changes in astrocyte activation and apoE levels, A␤ may be interacting directly with apoE receptors or via an interaction with an apoE receptor ligand. A␤-induced changes in apoJ appear to be independent of apoE receptors. ing a mechanism to clear the extracellular space of both apoE and A␤. Alternatively, apoE receptors may be coupled to an intracellular signaling cascade. Although the LDLR family of receptors was previously thought to be responsible only for the endocytosis of lipoprotein particles, or the clearance of cell debris in the case of LRP, recent evidence suggests that apoE receptors may have a signal transduction capacity as well. For example, a recent report linked two members of the LDLR receptor family, the very LDLR and ER2, to signal transduction pathways in the central nervous system (54). In addition, Herz and co-workers (55) have demonstrated that the C terminus of LRP binds to two cytosolic adapter proteins, FE65 and mammalian Disabled, an observation consistent with a signaling function for the cytosolic portion of the receptor. Furthermore, Goretzki and Mueller (56) have shown that LRP tail interacts with a GTP-binding protein and that binding of at least two LRP ligands increases intracellular cAMP levels and the activity of cAMP-dependent protein kinase. These studies together suggest that LRP may well mediate signal transduction that leads to various cellular responses.
In terms of a general mechanism, we propose that astrocytes respond to A␤ by increasing apoE, a reaction that serves to limit the inflammatory response. Unregulated glial activation could potentially compromise neuronal health via the sustained secretion of pro-inflammatory cytokines and oxidative stress molecules (57,58). The secretion of apoE into the extracellular space appears to reduce the functional activity of the A␤ peptide, possibly via the formation of an apoE⅐A␤ complex that is avidly cleared by apoE receptors, resulting in an overall increase in cell-associated apoE. The functional relevance of the accumulation of cell-associated apoE is unclear. Alternatively, the interaction of apoE or an apoE⅐A␤ complex with its receptor may result in the activation or inactivation of an intracellular signaling pathway that limits the A␤-induced inflammatory response. In either case, one testable prediction of this hypothesis would be that astrocytes cultured from apoE knockout mice would exhibit a greater inflammatory response to A␤ than wild type mice. Indeed, we have observed that the levels of A␤-induced pro-IL-1␤ in the knockout cultures are severalfold greater than in the wild type cultures. 2 This hypothesis is also consistent with our previous results demonstrating that exogenous apoE attenuates the activity of A␤ in both astrocytes and neurons (25,38). Whereas our experiments have focused on the role of astrocytes in the inflammatory response, it should be emphasized that microglia may also play a similar role in this process because they are abundant around amyloid deposits, express apoE receptors, and secrete apoE and a variety of inflammatory agents (59 -61). Altogether, our results reveal new insight into molecular mechanisms of A␤induced astrocyte activation and provide a strong foundation for future research into the mechanisms by which apoE and apoE receptors mediate specific responses of activated glia.