α2-Macroglobulin Exposure Reduces Calcium Responses to N-Methyl-d-Aspartate via Low Density Lipoprotein Receptor-related Protein in Cultured Hippocampal Neurons*

There is increasing evidence that the low-density lipoprotein receptor-related protein (LRP) can function as a signaling link in the central nervous system. To investigate the pathophysiological role of LRP in the central nervous system, we examined the effects of activated α2-macroglobulin (α2M*), a ligand of LRP, on intracellular calcium signaling in cultured rat hippocampal neurons. Neuronal effects of α2M* (50 nm) were assessed by a comparison of calcium signals produced in control and α2M*-pretreated neurons byN-methyl-d-aspartate (NMDA) and α-amino-3-hydroxy-5-methyl-4-isoxazole propionic acid. α2M* pretreatment significantly decreased the calcium signals to NMDA, whereas little change was observed for the signals to α-amino-3-hydroxy-5-methyl-4-isoxazole propionic acid. Native α2M, which is not a ligand for LRP, did not affect signals to NMDA. The receptor-associated protein prevented α2M*-induced decrease of calcium responses to NMDA, suggesting that α2M* exerted its effects through an LRP-mediated pathway. Experiments changing calcium sources demonstrated that α2M* pretreatment altered calcium responses to NMDA by primarily changing extracellular calcium influx and subsequently affecting calcium release from intracellular calcium stores. Immunoblot analysis demonstrated that α2M* caused a reduction in the levels of the NMDA receptor subunit, NMDAR1. These results suggest that α2M* can alter the neuronal response to excitatory neurotransmitters and that α2M* pretreatment selectively reduced the calcium responses to NMDA by down-regulating the NMDA receptor.

Among the diverse ligands for LRP, ␣2M* is of particular interest because of its robust association with cytokines (11,12) and neurodegeneration (13). ␣2M* is a large tetrameric protein that has established roles as a pan-proteinase inhibitor and in binding and clearance of a variety of small molecules, including cytokines (14), growth factors (15), and endogenous soluble ␤-amyloid peptide (16,17). When ␣2M has been "activated" by protease or chemical modification, it becomes a competent ligand for binding and clearance by LRP. Although ␣2M* can be a neurotrophic factor, little is known about the neuronal function of ␣2M* on intracellular calcium signaling. To assess ␣2M* potential in the CNS, we examined the chronic and acute effects of ␣2M* exposure on an important neuronal function in hippocampus of the brain: calcium signaling in response to stimulation of glutamate receptors. Calcium plays an important role in both physiological and pathological processes. Physiologically, calcium is an important intracellular second messenger and controls a variety of neuronal functions. However, excessive intracellular calcium is associated with neurotoxicity (18 -21). ␣2M*-induced changes in calcium signaling could profoundly affect neuronal function in the CNS.
Here, the effects of ␣2M* were examined in cultured hippocampal neurons, which were grown in relative isolation from other cell types. The hippocampal neurons express N-methyl-D-aspartate (NMDA), ␣-amino-3-hydroxy-5-methyl-4-isoxazole propionic acid (AMPA) and kainate subtypes of glutamate receptors. Glutamate is the main excitatory transmitter in the CNS and is known to play an important role in neuronal degeneration (18,22,23). Calcium signaling elicited by activation of NMDA and AMPA receptors were assessed using selective agonists. ␣2M* pretreatment substantially decreased the calcium signals produced by NMDA stimulation but not by AMPA stimulation or K ϩ depolarization. The decline of calcium signals was due primarily to a reduction in the influx of extracellular calcium. The effects of ␣2M* were mediated by LRP and were associated with reduced levels of NMDAR1. These results suggest that the elevated levels of ␣2M*, for example around plaques in the Alzheimer's disease (AD) brain, could significantly influence NMDA receptor-mediated processes and neuronal function in the CNS. the density of 1.5 ϫ 10 6 cell/ml. The neurons were grown in minimum Eagle's medium plus 10% horse serum and 10% fetal bovine serum supplemented with 30 mM glucose and 25 M penicillin-streptomycin. Medium with 10% horse serum was replaced every 3 days. Treatment with 5-fluoro-2Ј-deoxyuridine (20 g/ml) on the third day after plating minimized non-neuronal cell proliferation. The cultures survive for about 20 days in a standard CO 2

incubator.
Pretreatment with ␣2M*-50 nM ␣2M* was added to primary neuronal cultures 6 or 8 days after plating, in which half the culture media was replaced with medium containing 2% horse serum during ␣2M* pretreatment. Control cultures (sister cultures) were not pretreated with ␣2M* or other LRP ligands. The neurons were observed throughout the first 10 days in culture. The ␣2M*-containing medium was replaced by physiological saline before calcium measurements were made. The composition of the physiological saline was 140 mM NaCl, 3.5 mM KCl, 0.4 mM KH 2 PO 4 , 0.33 mM Na 2 HPO 4 , 2 mM MgSO 4 , 2.2 mM CaCl 2 , 10 mM glucose, and 10 mM HEPES-NaOH, pH 7.3. Similar pretreatment protocol was applied to compounds native ␣2M (50 nM), lactoferrin (500 nM), and receptor-associated protein (RAP) (500 nM).
Intracellular Calcium Measurement-Intracellular calcium was determined for individual cells using standard microscopic fluo-3 digital imaging (25,26). In this study, hippocampal neurons were loaded with 1 M fluo-3/acetoxymethyl for 30 min and incubated in dye-free saline solution for an additional 45 min at room temperature to allow cleavage of acetoxymethyl ester. The culture dish was then mounted on the stage of an inverted microscope equipped for fluo-3 video imaging. Live video imagines of selected microscopic fields were recorded with photomultiplier (Hamamatsu Photonics, Hamamatsu City, Japan) and digitized by computer. Images were formed by a pixel-by-pixel of the 488-nm wavelength-excited imaging. Real time digitized display, image acquisition, and calcium measurements are made with Bio-Rad imaging time course software (Imaging Research Inc.). The somata of ϳ5-10 cells in each microscopic field were individually measured. Intracellular calcium levels were estimated by converting fluorescent intensity to intracellular calcium concentration using the following formula: where F ϭ measured signal, F max ϭ measured signal for a saturated calcium solution, and F min ϭ measured signal for a calcium free solution. Calibration was done in vitro using fluo-3 salt (100 M) in solutions of known calcium concentration (Molecular Probes, Eugene, OR). Calibration of the fluo-3 signal in neurons required making measurements under saturating calcium concentrations and was facilitated by introducing extracellular calcium into cells with the calcium ionophore A23187 (Molecular Probes). The calcium calibration in vivo was consistent with calcium calibration in vitro. To avoid photo-bleaching and cell damages of fluo-3, 1 s of optical recording was followed by a 3-s break. Analysis of base-line data indicated that little or no bleaching of fluo-3 fluorescence occurred. The calcium calibration in vivo was applied to all pilot experiments that showed the alteration of calcium levels in order to demonstrate that the change of calcium signals was not the consequence of a change in fluo-3 loading. All experiments were performed at room temperature (ϳ23°C).
Drug Application-Cells were stimulated with micropressure application of either NMDA (100 M) or AMPA (50 M), selective agonists for the NMDA and AMPA subtypes of glutamate receptors, respectively. The agonists were dissolved in bath saline and applied by a brief (1 s) micropressure pulse from drug pipettes (1-3-m tip) placed under visual control near target neurons. 150 mM K ϩ (K ϩ substituted for Na ϩ in physiological saline), which causes membrane depolarization, was applied in the same manner. A dye was included in the treatment solutions to monitor neuronal exposure, demonstrating that the agonist or K ϩ was rapidly distributed over an area sufficient to expose the target neurons. For NMDA stimulation the cell bath and agonist solutions were magnesium-free physiological saline containing 5 M glycine. For the stimulation of AMPA and K ϩ depolarization, the bath contained normal physiological saline. In some experiments the neurons were exposed to antagonists or other drugs by bath exchange. These drugs included D(Ϫ)-2-amino-5-phosphonopentanoic acid (APV), nimodipine, EGTA, caffeine, and ifenprodil.
NMDA Receptor Analysis-For analysis of the NMDA receptor subunits, cell lysates were prepared from the primary rat hippocampal neuronal cultures. Cells were lysed in 50 mM Tris-HCl, pH 8.0, containing 0.5 M NaCl, 4 M leupeptin, 2 M pepstatin, 1.5 M aprotinin, 400 M phenylmethylsulfonyl fluoride, and 0.3% Triton-X. The total cell lysates were centrifuged at 2000 rpm for 30 s, and the supernatant was analyzed by immunoblot as described (17,27). Briefly, equal amounts of protein (ϳ5 g) were separate by SDS-PAGE (Novex) and transferred to nitrocellulose membrane. The monoclonal antibody for NMDA receptor subunit, NMDAR1, was a generous gift from Dr. Anthone Dunah, Massachusetts General Hospital. The concentration of antibody NMDAR1 used for immunoblot in this study was 1 g/ml. From the same blots, actin was detected by specific monoclonal antibody (AC-40; Sigma) to assure that equal protein was present in each lane. Immunoreactivity was detected using horseradish peroxidase-linked antimouse IgG developed with chemiluminescent reagent and exposed to film. Film were scanned with a Bio-Rad GS-700 imaging densitometer, and relative levels of NMDAR1 were determined for ␣2M*-pretreated and untreated cultures.
Chemicals-Human recombinant ␣2M (Sigma) was activated in methylamine (100 mM) at 1 mg/ml as stock solution and stored at Ϫ20°C for no more than 2-3 weeks. NMDA and AMPA were obtained from Tocris Neuramin, and stored as stock solutions at concentrations of 50 and 1 mM, respectively. APV (Sigma) was dissolved in dimethyl sulfoxide (Me 2 SO) at a concentration of 10 mM as stock solution. Stock solutions of nimodipine (20 mM; Sigma) were also prepared in Me 2 SO. The final concentration of Me 2 SO was not more than 0.05% in cell bath solution. In control experiments 0.05% Me 2 SO alone had no effects on calcium levels. EGTA (Sigma) was applied at a concentration of 1 mM to titrate the extracellular calcium. Caffeine, ifenprodil, and lactoferrin were purchased from Sigma. Human recombinant RAP was prepared from a glutathione S-transferase fusion protein as described (28).
Data Analysis-Intracellular calcium signals were quantified by measurement of peak amplitude. The decay phase of intracellular calcium signaling, Ca30, was also analyzed in some experiments. Resting calcium level was subtracted from all peak amplitude values and Ca30 amplitude values on an individual cell basis. Each protocol consists of two or three culture sets of hippocampal neurons in which 5-15 neuronal somata in each field were measured. Data from several cultures were pooled for statistical analyses. Values are expressed as the means Ϯ S.E. One-way analysis of variance followed by the Fisher post-hoc test for multiple comparisons determined statistical significance. p Ͻ 0.05 was considered indicative of a statistically significant difference.

NMDA-induced Intracellular Calcium Signals-Glutamate
receptor-mediated calcium influx has been suggested to be an important factor in neurodegenerative processes, with the NMDA receptor as the subtype primarily associated (18,21). In this study, we examined whether ␣2M* could alter calcium signaling after stimulation with NMDA. Fig. 1A shows typical intracellular calcium recordings from the hippocampal neurons in control cultures and in cultures chronically exposed to ␣2M* (50 nM) for 48 h. In both control and ␣2M*-pretreated neurons, the calcium response evoked by NMDA was characterized by a rapid initial peak and a slower recovery phase. The amplitudes of the calcium signals were significantly smaller in ␣2M*pretreated neurons compared with controls. Mean values for the populations of neurons studied are shown in Fig. 1B. ␣2M* pretreatment substantially decreased the peak calcium amplitudes to NMDA stimulation by 56% in the cultured hippocampal neurons. In addition to its effects on the NMDA response, ␣2M* also decreased resting calcium levels by 17% (Fig. 1, A  and B).
To control for nonspecific effects of ␣2M*, we examined the influence of native ␣2M (which is not an LRP ligand) on the calcium responses to NMDA. In contrast to activated ␣2M, native ␣2M did not influence the calcium response to NMDA ( Fig. 1, C and D). This result indicates that ␣2M must be activated to decrease the calcium response to NMDA.
The effect of APV, an NMDA receptor competitive antagonist, was tested to determine whether the intracellular calcium responses to NMDA in both control and ␣2M*-pretreated neurons were entirely induced by activation of NMDA receptors. After the neurons were incubated in saline containing APV (10 M) for 15 min, NMDA did not induce a visible increase of the intracellular calcium in either control (Fig. 1E) or ␣2M*-pretreated neurons (Fig. 1F). These results indicate that the intracellular calcium changes produced by NMDA stimulation were specifically induced by NMDA receptor activation in both control and ␣2M*-pretreated neurons.
The Acute Effects of ␣2M* Treatment on Cultured Hippocampal Neurons-To test whether an acute treatment with ␣2M* produced effects similar to the chronic pretreatment, ␣2M* was applied by micropressure or bath application at a standard dose (50 nM) and a larger dose (500 nM) to neurons in hippocampal cultures. In addition, the response to NMDA stimulation was monitored during acute bath addition of ␣2M* (Fig. 2). There was no visible intracellular calcium response to ␣2M* stimulation when applied by micropressure pulse at the concentration of 50 nM ( Fig. 2A). Panels C and D in Fig. 2 are representative recordings of intracellular calcium signals to NMDA stimulation in the absence (panel C) and in the presence (panel D) of 50 nM ␣2M*; panel E shows averaged data of both resting levels and peak calcium amplitude. Acute ␣2M* treatment at the concentration of 50 nM also did not reduce the calcium responses to NMDA as seen for chronic ␣2M* pretreatment.
However, at the higher concentration (500 nM), ␣2M* dramatically reduced the amplitude of spontaneous intracellular calcium oscillations in hippocampal neurons when applied by micropressure pulse (Fig. 2B). The spontaneous intracellular calcium oscillations in neurons were characterized with synchronization of the calcium oscillations across neurons in a given field. Panel F and G in Fig. 2 are representative recordings of intracellular calcium responses to NMDA stimulation in the absence (panel F) and in the presence (panel G) of 500 nM ␣2M*. Besides its dramatic reduction of the amplitude of spontaneous intracellular calcium oscillations, acute application of ␣2M* at 500 nM significantly decreased the intracellular calcium responses to NMDA (by 33%) and resting calcium levels (by 14%) (Fig. 2H). The levels of resting calcium in both control and chronic ␣2M*-treated neurons were determined by averaging the mean of intracellular calcium within the individual oscillations from the same field. These data indicate that a high concentration of ␣2M* produced acute effects on neuronal calcium signaling that were not seen at lower concentrations.
␣2M* Alters Calcium Responses to NMDA via LRP-RAP is a protein that facilitates the proper folding and trafficking of LRP within the early secretory pathway (29). RAP antagonizes the binding of all known LRP ligands to this receptor in vitro (30). To test if the tremendous decline in calcium responses to NMDA by ␣2M* was due to an LRP-mediated pathway, we co-incubated cultures with RAP (500 nM) to block the interactions between LRP and ␣2M*. Fig. 3A shows a representative recording and the averaged data of the effects of RAP on the ␣2M*-induced alteration of the calcium responses to NMDA. ␣2M* substantially decreased the calcium responses to NMDA, and this change was eliminated by co-culture with RAP. RAP alone caused no change in the neuronal calcium responses to NMDA (Fig. 3B).
Although ␣2M* and RAP both bind LRP, only ␣2M* affected calcium responses to NMDA. We therefore tested another LRP ligand, lactoferrin, which is of particular interest due to its association with calcium signals in blood monocytes (31). Fig.  3C, with the representative recordings and averaged data, shows that lactoferrin pretreatment substantially decreased calcium responses to NMDA. Averaged data shows a 17% decline in resting levels and 67% decline in peak amplitude, which is similar to the ␣2M*-induced reduction in calcium signals. To test if the reduced calcium signaling by lactoferrin was due to enhanced LRP-mediated pathway, we co-incubated cultures with 500 nM RAP. RAP treatment abolished the effects of lactoferrin on calcium responses to NMDA (Fig. 3C). This result suggests that lactoferrin and ␣2M* share the common pathway, LRP, for altering the calcium responses to NMDA.
Although RAP inhibited the effects of chronic treatment with 50 nM ␣2M* (Fig. 2), it had no effects on the intracellular calcium signaling produced by acute treatment with 500 nM ␣2M* (data not shown). Thus, LRP is involved in the chronic, but not acute, effects of ␣2M* under these conditions.
Effects of ␣2M* on Calcium Responses to AMPA and KCl-Hippocampal neurons express multiple subtypes of glutamate receptors. Therefore, it was of interest to know if the response evoked by another glutamate receptor agonist was affected by the chronic ␣2M* pretreatment. We examined the effects of ␣2M* on the intracellular calcium responses to AMPA, shown in Fig. 4 A, B (representative recordings), and C (mean values). Similar to NMDA, AMPA increased intracellular calcium, which was characterized by an initial peak but quick recovery phase in both control and ␣2M*-pretreated neurons. However, ␣2M* pretreatment did not change the calcium signaling of neurons to AMPA. This result indicates that ␣2M* alters the intracellular calcium responses to NMDA but not to AMPA.
Activation of NMDA receptors can cause a membrane depolarization that activates voltage-sensitive calcium channels, resulting in an enhanced calcium signal. An alteration of this signal by ␣2M* pretreatment could contribute to the reduced calcium response to NMDA in the ␣2M*-pretreated neurons. To test this possibility, the effect of ␣2M* pretreatment on calcium signals produced by K ϩ depolarization in hippocampal neurons was investigated. Representative recordings are shown in Fig. 4, D and E, and mean values are shown in Fig.  4F. Similar to AMPA, both control and ␣2M*-pretreated neurons show an initial peak and quick recovery phase in the intracellular calcium signaling in response to K ϩ depolarization. ␣2M* pretreatment had no effect on the K ϩ -evoked calcium signaling, suggesting that activation of only voltage-sensitive calcium channels did not contribute to the altered intracellular calcium responses to NMDA in ␣2M*-pretreated neurons. The results summarized in Fig. 4 demonstrate that ␣2M* did not affect calcium entry after stimulation of AMPA receptors or voltage-sensitive calcium channels.
Influence of Extracellular Calcium Influx on NMDA-induced Intracellular Calcium Signaling-LRP has a binding site for calcium, which may allow LRP to act as a sensor of extracellular calcium levels (32). We hypothesized that extracellular calcium levels may alter the calcium binding to LRP and subsequently affect calcium influx through NMDA-gated calcium channels. Therefore, we examined the effect of lowering physiological extracellular calcium levels on the response to NMDA by EGTA titration. Results are summarized in Fig. 5. Lowering extracellular calcium from 2.2 to 0.22 mM reduced the response to NMDA in both control and ␣2M*-pretreated neurons. Similar effects were observed in the calcium response to NMDA (Fig. 5, A and B) in both control and ␣2M*-pretreated neurons. The peak calcium amplitude was decreased by 79% in control and by 85% in ␣2M*-pretreated neurons (Fig. 5C). Similar reductions were observed in the decay phase, such as 30 s after stimulation (Ca30), in both control and ␣2M*-pretreated neurons (75 versus 85%) (Fig. 5D). These results indicate that extracellular calcium is an important component of the NMDA response, and the effects of ␣2M* on the calcium responses to NMDA depend on the activation of NMDA receptors (Figs. 1, E and F) rather than on the concentration of extracellular calcium. However, we found no indication that altered extracellular calcium levels reduced the calcium signaling of neurons to ␣2M* pretreatment, which suggested that the binding site for calcium of LRP may not play an important role in the different calcium responses to NMDA induced by ␣2M*.
␣2M* did not affect the calcium response to KCl (Fig. 4), indicating that calcium signals from only voltage-sensitive calcium channels did not contribute to the altered intracellular calcium responses to NMDA in ␣2M*-pretreated neurons. However, activation of NMDA receptors can cause a membrane depolarization that activates voltage-sensitive calcium channels, contributing to the calcium signal. To determine whether ␣2M* affected calcium influx through voltage-sensitive calcium channels subsequent to the activation of NMDA receptor, we used an L-type voltage-sensitive calcium channel blocker, nimodipine (5 M), in control and ␣2M*-pretreated neurons. Similar reductions were observed in average peak calcium amplitude in control and ␣2M*-pretreated neurons (48 versus 49%, p Ͼ 0.05), whereas the difference between control and ␣2M*pretreated neurons was eliminated by nimodipine in Ca30 calcium amplitude (88 versus 57%, p Ͻ 0.05) (Fig. 5, E and F). These results suggested that the L-type voltage-sensitive calcium channels contribute not only to the intracellular calcium responses to NMDA in both groups but also partly to the reduced intracellular calcium response to NMDA in the neurons chronically treated with ␣2M*.
Involvement of Intracellular Calcium Stores in NMDA-induced Intracellular Calcium Signaling-Calcium release from intracellular calcium stores could also contribute to NMDAinduced intracellular calcium response in hippocampal neurons. Such release could be trigged by extracellular calcium influx via either NMDA receptor-gated calcium channels or voltage-sensitive calcium channels activated by membrane depolarization. This release is called calcium-induced calcium release and is controlled by the ryanodine receptor located on the endoplasmic reticulum. To examine the role of these stores in NMDA-induced intracellular calcium signaling, both control and ␣2M*-pretreated rat hippocampal neurons were tested 1) in the presence of 20 mM caffeine, which depletes intracellular calcium stores (33,34), and 2) in the presence of 10 M dantrolene, an antagonist of the ryanodine receptor that blocks calcium-induced calcium release (33)(34)(35).
Panels A-C in Fig. 6 show the effect of caffeine on NMDA-induced intracellular calcium responses of hippocampal neurons in control and ␣2M*-treated neurons. 20 mM caffeine treatment reduced the response to NMDA in both control and ␣2M*-pretreated neurons (Fig. 6, A and B). There was an 80% reduction of peak calcium amplitude in control neurons and a 63% reduction in ␣2M*-pretreated neurons (Fig. 6C). The difference between control and ␣2M*-pretreated neurons was greatly reduced by caffeine but not abolished. Caffeine treatment at 20 mM also altered resting calcium levels. Mean resting calcium levels were increased by caffeine in both control and ␣2M*-pretreated neurons from 41.6 Ϯ 2.0 to 50.9 Ϯ 3.8 nM (n ϭ 40) in control neurons and from 28.3 Ϯ 2. 1   FIG. 3. ␣2M* alters the calcium responses to NMDA via LRP. Neurons were co-incubated ␣2M* with RAP at the concentration of 500 nM for 48 h. Panel A shows the representative recording and averaged data (mean Ϯ S.E.) of the effects of RAP on ␣2M*-induced alteration of the calcium responses to NMDA. ␣2M* substantially decreased the calcium responses to NMDA, and this change was eliminated by co-culture with RAP. Panel B shows the representative recordings and averaged data (mean Ϯ S.E.) of the effects of RAP. RAP alone did not alter the calcium responses to NMDA in both resting and peak levels. Panel C shows the representative recordings and average data (mean Ϯ S.E.) of the effects of lactoferrin. LF in panel C represents lactoferrin. In these figures, the average data include resting calcium levels (panels B and C) and peak calcium amplitude (panels A-C), in which the resting levels have been subtracted. *, p Ͻ 0.05. The effects of dantrolene on NMDA-induced intracellular calcium responses in control and ␣2M*-pretreated neurons are shown in Fig. 6, D-F. Dantrolene reduced the calcium responses to NMDA stimulation in both control and ␣2M*-pretreated neurons (by 54% in controls and 43% in ␣2M*-pretreated neurons). These results show that the control neurons were more sensitive to dantrolene. In addition, the ryanodine receptor immunoreactivity (relative to immunostaining in control neurons) was similar in control and ␣2M*-pretreated neurons (data not shown), which further demonstrated that the ␣2M* effects on the calcium response to NMDA are not primarily due to the alterations in intracellular calcium stores.
Together, Figs. 5 and 6 demonstrate that ␣2M* pretreatment altered calcium responses to NMDA by primarily changing extracellular calcium influx and, subsequently, affecting calcium release from intracellular calcium stores. Thus, we hypothesize that extracellular calcium influx through NMDA receptors plays an important role in the ␣2M*-induced alteration of intracellular calcium signaling to NMDA.
␣2M Alters NMDA Receptor Subunit Function and Immunoreactivity-NMDA receptors in rat hippocampal neurons are comprised of two distinct types of subunits, NMDAR1 and NMDAR2. NMDAR1 is the key subunit that possesses all properties characteristic of the NMDA receptor. NMDAR2 subunits modulate the properties of NMDAR1. The native receptor in hippocampal neurons is formed by a combination of NMDAR1 with at least one of four NMDAR2 subunits (referred to as NMDAR2A, -B, -C, and -D) (36 -38). The intricacy of NMDA receptor subunits is due to the variety of combinations of NMDAR1 with the physiological and pharmacological regulatory subunits, NMDAR2 (38). Therefore, in this study we only focused on the chronic effects of ␣2M* on 1) the intracellular calcium responses to NMDA in the presence of 0.5 M ifenprodil, an antagonist of NMDAR2B, and 2) the expression of NMDAR1.
NMDA receptors show differential pharmacological and functional properties depending on their subunit composition (39). Low micromolar concentrations of ifenprodil acts selectively to block receptors containing the NMDA2B subunit (40,41). We examined the effects of ifenprodil on the calcium responses to NMDA to determine whether receptor properties were affected by ␣2M* treatment (Fig. 7, A-C). 0.5 M ifenprodil reduced the amplitude of the calcium responses to NMDA in control neurons, whereas no significant effects were observed in ␣2M*-pretreated neurons (Fig. 7C). Ifenprodil eliminated the difference of calcium responses to NMDA between control and ␣2M*-pretreated groups. Thus, the effects of ␣2M* on NMDA receptor-gated calcium channels is limited to the NMDA receptor subtype NMDAR2B. In addition, we examined the effects of ␣2M* on the expression of NMDAR1 using immunohistochemical techniques. Fig. 7D shows that NMDAR1 immunoreactivity in ␣2M*-pretreated neurons declined by 38% relative to control neurons (n ϭ 7) (62.3 Ϯ 12.1 versus 100.5 Ϯ 1.7, p Ͻ 0.02). There were no visible changes in the levels of control protein, actin, when the same blots were probed with monoclonal antibody against actin. These data suggest that the reduced calcium responses to NMDA by ␣2M* pretreatment is due to a reduced NMDA receptor number and that a smaller proportion of NMDA receptors in the ␣2M*pretreated neurons contain the combination of NMDAR1 with NMDAR2B subunits than that in control neurons. DISCUSSION In this study, we found that exposure of cultured rat hippocampal neurons to ␣2M* alters the physiological characteristics of these neurons by decreasing the resting intracellular calcium levels and the calcium responses to NMDA. This effect of ␣2M* appears to be selective for the NMDA receptor since the response to AMPA and K ϩ depolarization were not altered. In addition, we found that an acute treatment of ␣2M* did not affect either the resting calcium level or the NMDA response in control neurons. However, acute application of a higher concentration of ␣2M* did affect the calcium response to NMDA and dramatically reduced the amplitude of spontaneous intracellular calcium oscillations. RAP prevented the ␣2M*-induced decrease of responses to NMDA, and another LRP ligand, lactoferrin, showed a similar declined response to NMDA as ␣2M*, suggesting that ␣2M* altered calcium signals to NMDA via an LRP-mediated pathway. Calcium release through intracellular calcium stores subsequent to the NMDA receptor activation contributed to the decreased intracellular calcium responses to NMDA in ␣2M*-pretreated neurons. Initial analysis of the molecular mechanism for these effects showed that the reduction of calcium signaling was limited to the NMDAR2B-contaning receptor, and the NMDAR1 levels were significantly reduced by ␣2M* pretreatment. These results suggest that The effects of 50 nM ␣2M* on resting calcium levels and the calcium response to NMDA required chronic exposure, and were not reproduced during acute treatment. Thus, 50 nM ␣2M* appears to exert its effects through a slow regulatory process, such as receptor expression, rather than by fast effects such as changes in enzyme activities. However, 500 nM ␣2M* had acute effects, reducing both the spontaneous intracellular calcium oscillations in hippocampal neurons and the calcium response to NMDA (Fig. 2, F-H). The synchronized calcium oscillations across neurons in a microscopic field depend on NMDA receptor activation, as indicated by their reduction by APV (Fig. 1, E and F) and depend on calcium release from intracellular calcium stores (data not shown). Neither the alteration in synchronized calcium oscillations nor the decreased intracellular calcium responses to NMDA and resting calcium levels were observed when neurons were co-cultured with ␣2M* (500 nM) and RAP. Thus, 500 nM ␣2M* appears to exert several effects on NMDA receptors through a fast regulatory process independent of LRP. The alterations of ␣2M* in calcium signaling are consistent with its reported neuronal trophic (4,6,42), protective (10), and electrophysiological effects (43,44).
The current result that chronic ␣2M* induces a reduced intracellular calcium response to NMDA complements the results previously reported by Bacskai et al. (45), in which the intracellular calcium signaling mediated by LRP was associated with NMDA receptor activation. However, in that study, an elevation of resting calcium levels through NMDA receptor was produced by acute ␣2M* stimulation (45); in this study we reported that ␣2M* produced a chronic reduction in resting calcium levels. After side-by-side comparisons of the experi-ments were performed, we found that different primary neuronal culture conditions account for these differences (data not shown). These comparisons revealed that the cultured neurons in the current study had lower intracellular resting calcium levels (by 27%, p Ͻ 0.0002) and were significant smaller in size (by 179% p Ͻ 0.0001) than those in the studies of Bacskai et al. (45). Furthermore, monomeric LRP ligands, such as lactoferrin, had no effect on that system, suggesting that the formation of LRP-LRP dimers was necessary for the calcium response. In the present system, lactoferrin behaved similarly to the tetrameric ␣2M* molecules. These differences raise interesting questions with respect to the influence of the culture condition of neurons on the sensitivity and response to ␣2M* pretreatment and suggest that neuronal responses in vivo may be heterogeneous, particularly in neurodegeneration.
Activation of NMDA receptors increases intracellular calcium via several pathways: (a) calcium influx through receptorgated channels, (b) calcium influx through voltage-sensitive calcium channels activated by membrane depolarization, and (c) release of calcium from intracellular calcium stores. Therefore, ␣2M* pretreatment could decrease the response to NMDA by altering one or more of these pathways. Extracellular calcium influx depends not only on the calcium gradient across the cell membrane but also on the activity of calcium channels on the cell membrane. ␣2M* did not alter the responses to AMPA and K ϩ depolarization (Fig. 3), although L-type voltage-sensitive calcium channels to some extent were involved in the calcium responses to NMDA (Fig. 5, E and F). It is unlikely that the effects of ␣2M* on the NMDA response are primarily due to a change in calcium influx via voltage-sensitive calcium channels. Studies with depletion of intracellular calcium stores showed that calcium release from intracellular stores plays a prominent role in the reduced calcium response to NMDA in the ␣2M*-pretreated neurons (Fig. 6). ␣2M* pretreatment also influenced the NMDA receptor subunit composition, as evidenced by sensitivity of the calcium responses to NMDAR2B subunit antagonist ifenprodil and NMDAR1 expression (Fig.  7). With the ␣2M* pretreatment, neurons exhibit a relatively smaller proportion of NMDA receptors containing the NMDAR2B and NMDAR1 subunit than control neurons. We conclude that ␣2M* alters the initial influx of calcium through NMDA receptor-gated channels and, thus, alters all subsequent effects on intracellular calcium levels. The subunit composition of the NMDA receptors in hippocampal neurons is known to be deregulated in neurodegeneration (46) and could be influenced by ␣2M* pretreatment.
Our cell culture data support the hypothesis that the effects of ␣2M* on calcium signaling are through binding to LRP rather than directly interacting with NMDA receptor. Calcium responses to NMDA can be blocked by magnesium and APV (Fig. 1, E and F) and potentiated by 5 M glycine in our study, demonstrating that the effects of ␣2M* on calcium signaling are NMDA receptor-specific. Our data support the model of ␣2M* directly binding to LRP and indirectly affecting the activation of NMDA receptor by modifying the receptor levels (panel D in Fig. 7) then altering intracellular calcium homeostasis. First, native ␣2M, which does not interact with LRP, had no effects on calcium responses to NMDA (Fig. 1, C and D). Second, another LRP ligand, lactoferrin, had similar effects as ␣2M* (Fig. 3C). Finally, the LRP blocker, RAP, blocked the effects of ␣2M* (Fig. 3A). The mechanism connecting LRP to the NMDA receptor remains unknown. In vitro experiments demonstrate binding of several cytoplasm proteins to LRP, including the Disabled-1 and the scaffold protein, postsynaptic density 95 (47,48). In particular, postsynaptic density 95 is of interest because of its binding to the NMDA receptor. These proteins may help in signaling transduction from LRP to the NMDA receptor (49,50). Other studies found that cytoplasmic residues of LRP can be phosphorylated (51), potentially affecting the association of LRP with those cytoplasmic proteins. The molecular connection between LRP and NMDA receptors will be pursued in the future.
NMDA receptors are known to play an important role in development (52), synaptic plasticity (53), and the generation of long term potentiation (54). They also may be involved in the pathological processes accompanying some neurological diseases. For example, in AD, the excessive calcium influx via NMDA-gated calcium channels and voltage-sensitive calcium channels might contribute to neuronal injury. NMDA receptors have been proposed to be involved in neurotoxicity caused by ␤-amyloid peptide, in which the neurotoxicity could be prevented by co-treatment with antagonists to NMDA (21) or calcium channel blockers (55). The current demonstration of the reduction of the intracellular calcium signaling in response to NMDA stimulation in ␣2M*-pretreated neurons, thus, could have important functional implications under conditions of elevated ␣2M* around plaques in AD brain. The functional alterations at the cellular level in AD brain could lead to reduced calcium toxicity or excitotoxicity, which could be a contributing factor in the neuronal damage observed in vivo.
LRP is associated with late-onset AD from several perspectives. Genetic studies have implicated two ligands of LRP as risk factors in AD; they are apolipoprotein E (56) and, more recently, ␣ 2 -macroglobulin (␣2M) (57). Immunohistochemical studies have found ␣2M* prominently associated with plaques (13). We focused on the chronic effects of ␣2M* because the presence of ␣2M* on plaques in the AD brain could expose surrounding neurons to plaque-associated protein for the life of . Averaged data (mean Ϯ S.E.) of peak amplitude (panels C) show the neuronal sensitivity to ifenprodil in control (open bars) and ␣2M* pretreatment neurons (filled bars) on the intracellular calcium responses to NMDA, in which the resting levels have been subtracted. Ifen in panels A and B represents ifenprodil. *, p Ͻ 0.05. The effects of ␣2M* on the expression of NMDA receptor subunit was also examined using immunohistochemical techniques in control and ␣2M*-pretreated neurons. Panel D shows a representative immunoblot of NMDA receptor subunit, NMDAR1, from cell lysates in both control (column 1) and ␣2M*-pretreated neurons (column 2). The concentration of monoclonal antibody for NMDA receptor subunit, NMDAR1, used for immunoblot in this study was 1 g/ml. From the same blots, actin was detected by specific monoclonal antibody. We found that NMDAR1 immunoreactivity in ␣2M*-pretreated neurons (n ϭ 7) declined by 38% relative to control neurons (n ϭ 7) (62.3 Ϯ 12.1 versus 100.5 Ϯ 1.7, p Ͻ 0.02), whereas no detectable changes in the levels of actin were observed. the plaque. Our studies of cultured hippocampal neurons examining the relationship between the NMDA receptor and LRP in neuronal degeneration demonstrate the potential for LRP to alter important aspects of neuronal function in the AD brain. Our study provides a context to interpret the genetic associations of apoE, ␣2M*, and LRP with AD. A deficiency of LRPmediated calcium responses to NMDA in the central nervous system may predispose toward AD by causing a relative increase in neuronal degeneration caused by intracellular calcium overload. Further evidence of how each of these genetic factors alters neuronal function may elucidate a common mechanism underlying disruption of intracellular calcium homeostasis in AD. Moreover, enhancing LRP-mediated calcium signaling or LRP-mediated neurotrophic effects may be a pharmacological approach to decreasing calcium response to NMDA and its associated pathological changes in vivo.