Nicotine Preconditioning Antagonizes Activity-dependent Caspase Proteolysis of a Glutamate Receptor*

Neuronal excitation is required for normal brain function including processes of learning and memory, yet if this process becomes dysregulated there is reduced neurotransmission and possibly death through excitotoxicity. Nicotine, through interaction with neuronal nicotinic acetylcholine receptors, possesses the ability to modulate neurotransmitter systems through numerous mechanisms that define this critical balance. We examined the modulatory role of nicotine in primary mixed cortical neuronal-glial cultures on activity-dependent caspase cleavage of a glutamate receptor, GluR1. We find that GluR1, but not GluR2 or GluR3, is a substrate for agonist (α-amino-3-hydroxy-5-methyl-isoxazole-4-propionic acid)-initiated rapid proteolytic cleavage at aspartic acid 865 through the activation of caspase 8-like activity that is independent of membrane fusion and is not coincident with apoptosis. Dose-dependent nicotine preconditioning for 24 h antagonizes agonist-initiated caspase cleavage of GluR1 through a mechanism that is coincident with desensitization of both nAChRα4β2 and nAChRα7 receptors and the delayed activation of a caspase 8-like activity. The modulation of GluR1 agonist-initiated caspase-mediated cleavage by nicotine preconditioning offers a novel insight into how this agent can impart its numerous effects on the nervous system.

The dynamic regulation of the expression of ionotropic glutamate-activated receptors (GluR), 1 the principal fast excitatory neurotransmitter receptors in the brain, is vital to the maintenance of effective neurotransmission. Disruption of this regulation can have severe pathophysiological consequences including neuronal death through excitotoxicity, as associated with stroke, trauma, and numerous severe neurodegenerative disorders (1,2). GluRs are assembled from multiple cDNA products to form receptors of distinct function, which fall into three pharmacologically defined groups (2). Although the activation of the N-methyl-D-aspartic acid (NMDA) receptor sub-class is central to the establishment of long-term potentiation, and its sustained activation appears responsible for imparting the majority of excitotoxicity (2,3), NMDA channel opening also requires local depolarization, which is often provided through coincident activation of the non-NMDA receptors, ␣amino-3-hydroxy-5-methyl-isoxazole-4-propionic acid (AMPA) receptors (GluR1-4), and kainic acid receptors (GluR5-7). This study examines AMPA-GluRs, in which channel permeability to calcium is dependent upon subunit composition. For example, inclusion of GluR2 imparts calcium impermeability, whereas receptors composed of GluR 1, 3, and/or 4 permit calcium entry. Consequently, altering the ratio of GluRs with varied subunit composition or modifying their subcellular location contributes significantly to the temporal and spatial regulation of NMDA receptors and glutamate neurotransmission.
Another ligand-activated ionotropic neurotransmitter system that can contribute to modifying the function of GluRs is the neuronal nicotinic acetylcholine receptors (nAChR). These fastionotropic ligand-activated channels (4 -6) are assembled from the products of cDNAs encoding at least six ␣-like subunits (␣2-␣7) and three ␤-like subunits (␤2-␤4). The expression of two nAChR subtypes is prevalent in the mammalian brain. First, receptors composed of nAChR␣4 and nAChR␤2 subunits are distinguished by high-affinity [ 3 H]nicotine or [ 3 H]cytisine binding (7). It is this receptor subclass in which up-regulation occurs in response to chronic nicotine administration (8), as in smoking, and in which expression correlates with the establishment of tolerance, a correlate of addiction (9). A second nAChR subtype composed of nAChR␣7 subunits binds ␣-bungarotoxin with high affinity and rapidly desensitizes to nicotine. However, while open, this receptor exhibits an unusually large Ca ϩ2 : Na ϩ permeability ratio of ϳ10:1 by comparison with the 3:1 ratio for NMDA receptors (4,6). This can have significant consequences on subcellular processes including signal transduction and the release of signaling molecules including arachidonic acid (10). Further, nicotine has been suggested to modulate the regulation of caspase (Csp) activation, including Csp8 function (11). However, due to the mechanisms of desensitization and inactivation of nAChRs, even if the receptor number is increased and there is abundant agonist (e.g. nicotine) present, signaling through this receptor system may actually be decreased. Therefore nicotine imparts multiple effects on neuronal function ranging from cognitive enhancement to neuroprotection against toxins (12)(13)(14)(15)(16) to its well known attribute, craving and addiction, through affecting multiple mechanisms that combine properties of both imparting receptor activation and enhancing long-term desensitization.
Neurons use a variety of mechanisms to control AMPA-class GluRs expression. For example, during periods of sustained synaptic activity GluRs, in particular GluR1, are observed to change in subcellular distribution as reflected by the accumulation of this subunit at some sites of activity and by the * This work was supported by National Institutes of Health Grant NIH-NS35181 and the Val A. Browning Foundation. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The most rapid way to regulate the concentration of cellular proteins is through selective proteolysis (26). AMPA-GluRs are subject to this mechanism, particularly in the presence of elevated free intracellular calcium, which can activate calpains, which have numerous cellular substrates including GluR1 (25). However, the limited proteolytic cleavage of GluR1 may include other proteases. Recently, members of the cysteine protease family, caspases, have been reported (27) to cleave certain GluR subunits, especially GluR4, following trophic factor withdrawal or upon induction of cell death through apoptosis. Although the activation of caspase cascades is most commonly associated with cell death, under more restricted conditions certain caspases might also perform limited substrate cleavage, which has the regulatory importance of contributing to proper cellular function. For example, the conditional activation of Csp1 converts pro-interleukin 1␤ to the active cytokine that is released from the cell (28). Therefore, a less recognized role for caspases might be in regulating processes related to normal neuronal responses and function.
In this study we report that GluR1 is a substrate for AMPAinitiated activity-dependent limited caspase cleavage and demonstrate that nicotine preconditioning of primary mixed cortical neuronal-glial cultures modulates this proteolytic activation. Site-directed mutagenesis and in vitro assays demonstrate that GluR1 is cleaved through a Csp8-like protease at residue 865 in the distal C terminus. Preconditioning of cultures with nicotine alters the activation of the Csp8-like proteolytic cascade as measured using cell-permeable fluorescent substrates, and this correlates with reduced AMPA-initiated caspase cleavage of GluR1. The use of various agonists and antagonists of nAChRs suggest that this mechanism of reducing Csp8 activation requires the coincident desensitization and/or inactivation of both nAChR␣7 and nAChR␣4␤2 receptor subtypes. The modulation by nicotine of activity-dependent cleavage of GluR1 has several implications regarding the action of this compound on the molecular processes underlying GluR excitatory neurotransmission and the modulation of neurological disease.

MATERIALS AND METHODS
Reagents-All reagents were obtained from RBI/Sigma. Drugs were dissolved in minimum Eagle's medium (MEM) or Me 2 SO at 100 -1000ϫ. Caspase inhibitors and substrates were obtained from Calbiochem or Enzyme Systems Products. Recombinant caspases 3 and 8 and the calpain inhibitor PD150-606 were obtained from Alexis Biochemicals or Calbiochem.
Neuronal Tissue Culture-Murine primary cortical cultures composed of mixed neuronal and glial cells were prepared from E15 CD1 mice (Charles River) and maintained for 13-15 days as described previously (15,16). All experiments were conducted at least 24 h after cells were fed.
Temperature Incubations-For experiments using varied temperature, the medium was buffered with 10 mM HEPES, the culture dishes wrapped in Parafilm, and the dishes placed in contact with water heated to the indicated temperature using circulating water baths. Cultures were equilibrated for 10 -15 min before the addition of drugs. Degradation and Arrhenius activation were calculated as described elsewhere (29).
Immunoblots and Immunocytochemistry-Western blot analyses were done as described previously (30). Briefly, neurons were washed with phosphate-buffered saline and then lysed and harvested by scrap-ing in a cold protease mixture (4 mM phenylmethylsulfonyl fluoride and 10 mM EDTA and 10 mM benzamidine) prepared in water. SDS-PAGE loading buffer containing dithiothreitol was added to this mixture immediately, and the cells were further scraped into a microcentrifuge tube and boiled in a dry heating block for 10 min. Upon gel fractionation by SDS-PAGE, proteins were transferred to nitrocellulose utilizing a semi-dry blot apparatus, blocked in phosphate-buffered saline with 5% dry milk and 0.05% Tween 20, and incubated in primary antibodies overnight at 4°C. Antibodies were to the C terminus of GluR1 (Chemicon), E11 (GluR1 (31), 3A11 (GluR2; Chemicon), and 2F5 (GluR3 (30)). The blots were washed in phosphate-buffered saline/Tween 20 and incubated in the presence of peroxidase-conjugated secondary antibodies (Jackson ImmunoResearch) and detected on film using the Enhanced Chemiluminescence Kit (Amersham Life Sciences, Inc.). Films were scanned, and relative band intensities were determined by densitometry. Immunocytochemistry was performed as described previously (31).
Site-directed Mutagenesis-Mutagenesis to convert GluR1 aspartic acid 865 to alanine (D865A) was performed using PCR site-directed mutagenesis as described elsewhere (30) with modifications appropriate to GluR1. Constructs were confirmed by automated sequencing (University of Utah DNA sequencing core facility).
Transfection-The CalPhos transfection kit from CLONTECH was used according to manufacturer's instructions to introduce expression plasmid cDNAs (2 g of GluR (GluR1:pcDNA1/AMP; see Ref. 31) combined with 1 g of green fluorescent protein cDNA (GFP:pcDNA1/ AMP)) prepared using the Qiagen Maxi-kit into HEK293 cells (ATCC). Cells were used at 24 -48 h post-transfection.
Caspase Assays-To measure caspase cleavage of GluR1 or GluR3 in vitro, HEK cells were transfected with the respective GluR cDNA and 48 h later were washed with phosphate-buffered saline and lysed in cold buffer composed of 50 mM HEPES, pH 7.3, 50 mM NaCl, 10 mM EDTA, 10 mM dithiothreitol, 0.1% Chaps, 1 mM phenylmethylsulfonyl fluoride. The cells were subjected to Dounce homogenization and cleared by centrifugation. To each replicate 100-l sample of cell lysate was added a 25-l aliquot of either recombinant Csp8 or Csp3 (25 units/assay), respectively. The tube was sealed and placed at 37°C for the period indicated, whereupon a 10-l sample was removed and mixed with 10 l of SDS-sample buffer, boiled for 5 min, and fractionated by SDS-PAGE. For microscopic visualization of caspase activation in cultured neurons, cells were loaded with cell-permeable Csp8 peptide substrate (Z-IETD-AFC), which fluoresces blue upon cleavage. Cultures were washed with Hanks' solution, incubated in the presence of Hanks' solution supplemented with 10 mM HEPES, and supplemented to 10 M with the cell-permeable caspase-substrate peptides indicated (stock solution of 1:1000 dissolved in Me 2 SO) for 30 min at 37°C. Cells were washed again, AMPA (100 M) or kainic acid (100 M) was added, and at the desired times thereafter (optimized in trial experiments) fluorescence was automatically recorded at 1-min intervals. The appearance of fluorescence in individually identified neuronal cells was quantitated using Image Pro-Plus software.

GluR1 Is Subject to Limited Proteolysis upon Receptor
Activation-Activity-dependent subcellular redistribution of GluR1 occurs following the exposure of neurons to glutamate receptor agonists and antagonists (17,18). Consistent with these findings, the addition of a nonlethal, nonapoptotic (15,16), dose of AMPA (100 M) or kainic acid (100 M, not shown) for 1 h to our murine primary cortical neuronal-glial cultures was accompanied by the rapid loss of GluR1 C-terminal immunoreactivity as measured by Western blot analysis (Fig. 1A). Immunoreactivity to GluR2 or GluR3 from the same culture samples was unaffected (Fig. 1A). GluR1 degradation was inhibited by coapplication of CNQX but was unaffected by the presence of tetrodotoxin, nifedipine, or the calpain inhibitor, PD 150 -606 (not shown). Although degradation was complete in all experiments by ϳ90 min, some GluR1 remained suggesting that a pool of this subunit protein was unaffected. This pool is likely to be intracellular because parallel immunocytochemical examination (using the same anti-C-terminal GluR1 antibody) of cultures treated with AMPA for 1 h revealed a change from normally diffuse immunoreactivity to a punctate distribution suggestive of an intracellular-endosomal pattern (Fig. 1B). Ex-amination of GluR1 degradation using an antibody prepared to the extracellular domain (24) failed to reveal an equivalent change in GluR1 expression over the same time period, albeit a change of ϳ1 kDa in migration on gels was observed consistent with removal of the C-terminal region (not shown). Because the C terminus of GluR1 harbors numerous sequences important to the subcellular distribution and function of this receptor (17,19,32), we examined further the nature of this presumed proteolytic event.
Activity-dependent GluR1 Degradation Does Not Require Endocytosis and the Rate-limiting Step Is ATP-independent-Varying temperature can be used to measure several parameters pertaining to cellular and proteolytic mechanisms. Endocytosis and lysosomal degradation of proteins are particularly sensitive to reduced temperatures because membranes fail to fuse as temperatures go below 16°C (e.g. see Ref. 29). This has the effect of producing a dramatic and steep reduction in the degradation rate as delivery systems stop working at lower temperatures. Other proteolytic mechanisms that do not require membrane fusion, such as the ATP-ubiquitin-dependent proteasome and other cytoplasmic or extracellular proteases, are also affected by reduced temperature. In this case, however, the protease enzymatic rate is slowed according to its Q 10, a constant that reflects the effect of a 10°C change on the enzymatic function. We examined these parameters by measuring the rate of agonist-dependent proteolysis of GluR1 at 37, 26, 16, and 6°C, respectively (Fig. 1C). An average of three experiments revealed that the rate of AMPA-initiated GluR1 degradation decreased ϳ5-fold over the 31°C temperature range for an average Q 10 of 1.57 Ϯ 0.17. An additional benefit of making this measurement is that the energy of activation (Ea) for the rate-limiting step of degradation can be calculated through plotting the reaction Arrhenius plot (29). The Arrhenius plot derived from these data (Fig. 1C, insert) is best fit by a straight line (R 2 ϭ 0.95). This result is consistent with a proteolytic mechanism that does not require membrane fusion because no Arrhenius "break" in the plot was observed as for other transmembrane receptor proteins (see Ref. 29) including the muscle nicotinic acetylcholine receptor, which is degraded to completion in the lysosome following endocytosis (33,34). The apparent Ea for the rate-limiting step of AMPA-initiated GluR1 proteolysis as calculated from the Arrhenius plot is 8.4 kcal/mol. This is substantially less than the energy of activation for ATP-dependent ubiquitin-mediated degradation, which is 27 Ϯ 5 kcal/mol (29). Because most ATP-independent proteases exhibit an Ea of between 9 and 15 kcal/mol (29), the latter possibility is favored, although the possibility that an AMPA-initiated step prior to proteolysis is rate-limiting to GluR1 cleavage cannot be ruled out. Nevertheless, these results suggest that the mechanism of AMPA-initiated proteolysis of GluR1 is independent of endosome formation and lysosomal proteolysis and probably does not involve an ATPubiquitin proteasome. This result also further supports the possibility that the GluR1 punctate endosome-lysosome-like pattern of immunostaining revealed in cultured neurons following AMPA addition reflects a pre-existing internal protein pool, which is not necessarily degraded by the AMPA-initiated protease system measured in these experiments.
GluR1 Is a Substrate for a Caspase 8-like Proteolytic Activity-Although multiple proteases could act upon GluR1, including calpains (25,35), we explored the possibility that caspases could mediate AMPA-initiated limited cleavage. Inspection of the GluR1 cytoplasmic C-terminal sequence reveals a putative caspase 8 cleavage sequence (residues 862-865 (VSQD) (see Ref. 36) and Fig. 2A) that is 12 amino acids N-terminal to the C-terminal GluR1 immunogen sequence. To determine whether GluR1 is a substrate for cleavage by Csp8 at this site, crude membrane fractions were prepared (see "Materials and Methods") from HEK293 cells that were transfected for 48 h with the CMV-pcDNA1/Amp expression vector encoding either rat GluR1 or rat GluR3 (30,31) for use as substrates for proteolysis in vitro using recombinant Csp8 or Csp3. As shown in Fig. 2B, Csp8 cleaved GluR1 but not GluR3, and neither GluR was a substrate for recombinant Csp3.
To determine whether cleavage of GluR1 occurred at residues 862-865 (VSQD), this site was removed through conversion of the aspartic acid (Asp 865 ) to alanine (D865A) by site-directed mutagenesis (see "Materials and Methods" and Ref. 30), and the above experiment was repeated. As shown in Fig. 2C, the introduction of a single mutation in GluR1 at D865A reduced recombinant Csp8 cleavage in vitro from ϳ75% of the total wild-type GluR1 pool to less than 13% for the total GluR1D865A pool. The background degradation of GluR1D865A is likely related to either nonspecific proteolysis or Csp8 cleavage at another less susceptible cleavage site(s), or possibly to the activation of other proteolytic systems by recombinant Csp8 following its addition to this crude membrane extract. Notably, the presence of 1 mM phenylmethylsulfonyl fluoride and 10 mM EDTA had no effect on FIG. 1. GluR1 is selectively degraded following exposure to AMPA by a membrane fusion-independent process. A, Western blot analysis of GluR1, GluR2, and GluR3, respectively, from primary cultured mouse cortical neurons exposed to 100 M AMPA for 1 to 120 min (Veh, vehicle control). AMPA specifically decreased GluR1 as revealed by an antibody to the C terminus, but not GluR2 or -3, in a time-dependent manner. B, immunocytochemical analysis of GluR1 in these neurons shows that the diffuse cytoplasmic staining of neurons treated as described for panel A for 1 h with AMPA revealed a persistent endosome-like punctuate pattern. C, averaged results Ϯ standard deviation (n ϭ 3) of GluR1 degradation measured by Western blot analysis of neuronal cultures treated with 100 M AMPA and maintained at 37, 26, 16, and 6°C for the times indicated (points are off set for clarity). The regression lines were generated by "best fit" analysis; the R 2 values, ranging from 0.83 to 0.98, were all significant. The rate of degradation decreased consistent with a calculated Q 10 of 1.57 Ϯ 0.17. The Arrhenius plot (inset) calculated from these data is best fit by a straight line (R 2 ϭ 0.95) and exhibits no indication of a phase transition between 20 and 16°C. This suggests that this reaction proceeds independently of membrane fusion as would be required for lysosomal degradation. Calculation of the Ea for the rate-limiting step for this cleavage reaction is 8.4 kcal/M. Error bars for this plot are ϮS.E. the proteolytic rates (not shown), which precludes calpains and serine proteases, respectively, from contributing to the in vitro degradation measurement. Nevertheless, the majority of GluR1 degradation is attributable to the added recombinant Csp8 in vitro and cleavage by this protease is predominantly at GluR1-Asp 865 .
AMPA-mediated GluR1 Proteolysis Is Blocked by Caspase 8 Inhibitors-We next used cultured neurons to examine whether inhibitors of caspases could affect AMPA-initiated GluR1 degradation. For these experiments, cell-permeable inhibitors of various caspases were used as follows. Primary cortical neuronal cultures were incubated in the presence of cell-permeable caspase peptide inhibitors (e.g. Z-VAD-FMK, a broad-range inhibitor of caspase activity); Z-IETD-FMK, an inhibitor of Csp8; and others, as follows) before adding AMPA and harvesting neurons thereafter for Western blot analysis of GluR1 degradation. Because caspase inhibitors suffer from two experimental concerns, the relative intracellular concentration that can be achieved by incubation and the often relatively poorly characterized target specificity of the respective peptides (37,38), we also examined the dose response for each caspase inhibitor tested. As shown in Fig. 2D, Z-IETD-FMK showed potent inhibition of AMPA-initiated GluR1 proteolysis even at the lowest dose tested (1 M), whereas other inhibitors either failed to inhibit cleavage (Z-VDVAD-FMK, Csp2) or inhibited degradation only as higher concentrations were used (Z-DEVD-FMK, Csp3). Other caspase inhibitors with reported specificity to Csp6 (Z-VEID-FMK) and Csp1 (Z-YVAD-FMK) exhibited inhibition of degradation to a limited extent beginning at incubation concentrations of 10 and 100 M, respectively. Notably, these data are consistent with the possibility that GluR1 could serve as a substrate for other caspases. However, given the likely promiscuity of these inhibitors toward caspases other than their intended target (especially at high concentrations (28,37,39,40)) and the findings that GluR1 is cleaved by Csp8 at residue Asp 865 in vitro, these data are consistent with Csp8 or Csp8-like mediated cleavage of GluR1 in response to AMPA-initiated receptor activation in cultured neurons.
Nicotine Pretreatment of Neuronal Cultures Antagonizes Activity-dependent GluR1 Proteolysis-Chronic exposure of ionotropic cholinergic receptors to the extrinsic agonist, nicotine, has been implicated in numerous effects on glutamate receptor neurotransmission ranging from direct effects on neurotransmitter receptor function to conferring neuroprotection to glutamate receptor excitotoxic challenges (11,15,16,41). During the course of examining the mechanism of nicotine-mediated neuroprotection, we noted that cultures treated with nicotine retained GluR1 expression subsequent to AMPA challenge. To examine this further, nicotine at physiologically relevant doses (10 -1000 nM) was added to cultures for 24 h prior to addition of AMPA (100 M). Western blot analysis of GluR1 from cultures harvested at various times thereafter (Fig. 3A) revealed that nicotine decreased AMPA-initiated GluR1 cleavage in a dosedependent manner (Fig. 3, A and B). Nicotine had no detectable effect on AMPA-initiated GluR1 proteolysis at lower concentrations (i.e. Ͻ1 nM) and saturated at concentrations of Ն1 M (not shown). The dose-dependent inhibition of AMPA-initiated GluR1 proteolysis by nicotine also coincided with the retention of the diffuse GluR1 staining characteristic of control cultures (Fig. 3D). Similar to the data in Fig. 1, GluR2 and GluR3 expression were unaffected by nicotine administration over this time course (not shown).
We next examined the pharmacology of how nicotine imparts antagonism of AMPA-initiated GluR1 proteolysis. The nAChRs expressed in our primary neuronal cultures predominantly consist of the nAChR␣4␤2 and nAChR␣7 subtypes (see Refs. 15, 16, and 41; and data not shown). Very low levels of the nAChR␣3 and nAChR␤4 subunits are detectable, and nAChR␣2 has not been detected (immunocytochemistry data not shown). The block of activity-dependent degradation by nicotine in our cultures requires a pre-exposure of the neurons to nicotine of at least 24 h. Under these conditions nearly all nAChRs are in the high-affinity desensitized state (42). In this state, cytisine, at the concentrations we used (1-1000 nM), will bind mainly to the nAChR␣4␤2 subtype of nAChR. In fact, the nAChR␣4␤2 receptor is saturated at these concentrations (7). Cytisine is able to bind desensitized nAChR␣7 receptors at the higher concentrations that we used (100 nM to 1000 nM), but these are sub-saturating conditions. Binding of cytisine to desensitized nAChR␣3␤2 or nAChR␣3␤4 receptors (expression of these is rare in our system) requires concentrations greater than 10 M for saturation; this would not be expected to be a major contributor FIG. 2. GluR1 is a substrate for caspase 8 cleavage. A, a putative Csp8 site (Csp8) in the C terminus of GluR1 was identified at residues 862-865 (VSQD, asterisk) near the C-terminal epitope, as marked, which is recognized by the anti-GluR1 antibody used in this study. B, to determine whether GluR1 was a substrate of Csp8, HEK cells transfected with GluR1 or GluR3 were solubilized in caspase buffer and subjected to incubation at various times in the presence of recombinant Csp8 or Csp3 (see "Materials and Methods"). At the times noted, samples were removed and analyzed using Western blot analysis with antibodies directed to the respective GluR subunit. Csp8 cleaved GluR1, but no cleavage by Csp3 was detected. GluR3 was not a substrate of either caspase (Csp3 is not shown). C, in this experiment the putative Csp8 recognition sequence was mutated by conversion of aspartic acid 865 to alanine (D865A), and the above experiment was repeated. The plot shows the respective degradation rate between GluR1 wild-type (WT) and R1D865A indicating that removal of Asp 865 substantially diminished Csp8 cleavage of GluR1. Error bars are ϮS.E. D, neuronal cultures were treated for 1 h with different starting concentrations of inhibitors to Csp2 (z-VDVAD-FMK), Csp3 (z-DQMD-FMK), or Csp8 (Z-IETD-FMK) before adding 100 M AMPA for 1 h and assaying for GluR1 using Western blot analysis (Veh, vehicle control; no AMPA or inhibitors). Notably, in these experiments inhibitor alone was equivalent to the vehicle control, and AMPA decreased GluR1 signal consistent with the results shown for Csp2 inhibitor (not shown). The Csp8 inhibitor blocked AMPA-initiated GluR1 cleavage in a dose-dependent manner, but inhibitors to caspases 2 and 3 had no effect or only a small inhibitory effect at higher concentration, which could reflect the promiscuous nature of these inhibitors (62).
to the effects observed in the majority of our cultured neurons. As shown in Fig. 3C, cultures treated with cytisine 24 h prior to AMPA only partially blocked AMPA-initiated GluR1 proteolysis in a bell-shaped dose-response curve that peaked at ϳ10 nM. This dose-response curve suggests that initially some antagonism through nAChR␣4␤2 activation is observed (cytisine is a partial agonist at nAChR␣4␤2 subtypes (43)(44)(45)), but this compound loses efficacy at higher concentrations where receptor desensitization dominates (45). Because cytisine fails to activate (and binds poorly) nAChR␣7 at these concentrations (6), we determined whether desensitization/ inactivation of both receptor types (nAChR␣7 and nAChR␣4␤2) was required to reconstitute the nicotine effect. As shown in Fig. 3C, the coincident addition of ␣-bungarotoxin (an antagonist of nAChR␣7) and cytisine was able to completely block AMPA-dependent degradation of GluR1 in a dose-dependent relationship resembling inhibition by nicotine alone. The addition of ␣-bungarotoxin alone had no effect (not shown). Because the concentration and duration of nicotine required to inhibit AMPA-initiated GluR1 cleavage is likely to fully desensitize nAChR␣7 and nAChR␣4␤2 (4,6,46), our data are consistent with a mechanism that suggests that the coincident desensitization and/or inactivation of both nAChR␣4␤2 and nAChR␣7 acts to precondition the system in a manner that antagonizes the mechanism of AMPAinitiated proteolysis.
Nicotine Preconditioning of Neurons Alters AMPA-mediated Caspase 8-like Protease Activation-If a Csp8-like activity cleaves GluR1 upon neuronal exposure to agonist, does nicotine preadministration alter AMPA-initiated Csp8 activation? This possibility was tested by pretreating cultures with 1 M nicotine for 24 h and comparing the activation of Csp8-like activity relative to vehicle (control)-treated cultures. Treated cultures (24 h post nicotine) were incubated for 30 min at 37°C in Hanks' solution (see "Materials and Methods") supplemented with the cell-permeable synthetic Csp8 substrate peptide, Z-IETD-AFC. Upon cleavage this substrate fluoresces blue, which is observed in cultured neurons following the addition of AMPA or kainic acid, as shown in Fig. 4A. The average meas-urable fluorescence for neurons from four cultures revealed a rapid increase in Csp8 activity in control cultures in response to agonist; this increase is complete within 15-25 min after agonist exposure as quantitated in Fig. 4B. Monolayer cells in our culture failed to respond to AMPA, as they produced no fluorescence following this treatment (not shown). Cells pretreated with nicotine demonstrated a marked delay of 10 -20 min in the initial appearance of AMPA-initiated proteolysis. Although the accumulation of fluorescent product achieved a maximum by 40 min, overall cleavage of the Csp8 fluorescent substrate in nicotine-treated cultures failed on average to achieve the total fluorescent accumulation observed in control cells (Fig. 4B). However, it should be noted that determining the saturation of the reaction in nicotine-treated cultures was complicated by the variability in the individual cell responses observed 40 min after AMPA addition to these cultures (see Fig. 4A). This suggests that the response to Csp8 or Csp8-like proteolytic activation in nicotine-treated cultures is not as homogenous as the response observed in control cultures, a result that possibly reflects neuronal variability the expression of nicotinic and/or glutamate receptor subtypes. For example, a number of cells in the nicotine-pretreated cultures had a rate of initial activation and accumulation of Csp8 fluorescent substrate in response to AMPA that was unaltered relative to controls, suggesting that they were either unresponsive to nicotine or were possibly expressing receptor combinations other than nAChR␣7 and nAChR␣4␤2. Notably, this delay by nicotine in the accumulation of Csp8 fluorescent substrate appears to reflect a slowed or delayed activation of the Csp8 enzyme. Because Csp8 activation from the pro-form to the active form requires aggregation (e.g. Ref. 47), the possibility that nicotine disrupts pro-caspase8 aggregation or disrupts hypothesized complexes that place Csp8 in close proximity to the GluR1 target cleavage sequence is a topic of active investigation. However, the effect of nicotine on preconditioning the neuronal Csp8 response is indeed of interest and could reflect a novel mechanism through which this exogenous agent could impart long-term effects on neurotransmission through modifying normal neuronal degradative pathways. of GluR1 by nicotine (Nic) and the partial agonist to nAChR␣4␤2, cytisine, which was used either alone (Cyt) or in a 1-h pretreatment in the presence of 1 M ␣-bungarotoxin (Cyt ϩ ␣BgTx), a nAChR␣7 antagonist. The presence of ␣-bungarotoxin alone had no effect (not shown). These results are consistent with the conclusion that desensitization of nAChR␣4␤2 at higher concentrations of cytisine accompanied by antagonism of nAChR␣7 reconstitute nicotine inhibition of degradation. D, immunocytochemical analysis of GluR1 in neurons treated as in the experiment described in A and B, again showing a correspondence between retention of the diffuse cytoplasmic GluR1 immunoreactivity and protection of GluR1 immunoreactivity as measured on Western blots.

DISCUSSION
The effects of prolonged exposure of the nervous system to nicotine can range from neuroprotection to cognitive enhancement to addiction (4 -6). Given the diverse impacts of nicotine on neurological processes, these results offer a novel mechanism through which preconditioning by nicotine and the desensitization of the major subtypes of nAChRs in our system, nAChR␣7 and nAChR␣4␤2, can modify neurotransmission and neuronal function. Because nicotine is lipophilic and is slowly removed from the central nervous system (9), receptor desensitization is thought to be the predominant state of these receptors in the brain of regular smokers. The effect of these seemingly antagonistic responses by nAChRs to nicotine (activation and desensitization) can be rather complex because much of the influence of this receptors system is on modulating neurotransmitter release, which can also have seemingly contradictory effects if the neurotransmitter is excitatory or inhibitory. In our primary neuronal tissue culture system, we have demonstrated previously that nicotine imparts a neuroprotective effect to an NMDA-mediated excitotoxic challenge through activation of nAChR␣7 but not nAChR␣4␤2 (15,16,41). However, it is important to note that the expression of nAChR␣7 and nAChR␣4␤2 can occur in Ͼ70% of cultured neurons (not shown); this probably establishes the basis for such a robust detection of this effect of nicotine on GluR1 cleavage in this culture system. Consequently, in other neuronal culture systems or within the brain, the coincident expression of these three receptors (GluR1, nAChR␣7, and nAChR␣4␤2) would vary and probable be less likely to occur than in cultured cells. However, when these receptors do occur in combination, as in neurons in subregions of the hippocampus (48 -50), the mechanism of nicotine preconditioning may play an important role in the modulation of GluR1 expression and function and may contribute to the regional specificity of normal neuronal function and susceptibility to pathology through agents such as glutamate receptor excitotoxins.
We have demonstrated that the addition of AMPA to cultured neurons can initiate mechanisms that lead to the rapid and limited cleavage of GluR1 at residue Asp 865 in the Cterminal domain through a Csp8 or Csp8-like protease. Notably, this process is disrupted by prolonged nicotine exposure, which requires the coincident desensitization or inactivation of nAChR␣4␤2 and nAChR␣7, respectively. The possibility that internalization of GluR1 through an endocytotic pathway occurs prior to cleavage of GluR1 seems unlikely for several reasons. Most notable is the absence of a break in the Arrhenius plot at temperatures below 16°C where membrane fusion would cease. Such a break is notable in other systems where internalization precedes degradation such as the muscle nicotinic acetylcholine receptor (33,34). Further, immunoreactivity to GluR1 continues to diminish at lower temperatures, as revealed by Western blot analyses using the C-terminal antibody, and proceeds kinetically (Arrhenius plot data) in a manner most consistent with the proteolytic removal of just the GluR1 C-terminal region following agonist activation.
The observation that nicotine interferes with AMPA-initiated GluR1 degradation suggests a unique mechanism through which nAChRs can modulate neurotransmission by other receptor systems. GluR1 is a particularly interesting substrate because its C-terminal cytoplasmic domain contains numerous signals for both phosphorylation and binding to cytoskeletonlinking proteins. Compelling evidence (e.g. see Ref. 2) argues that regulation of GluR1 expression is particularly important in processes determining reinforcement of local synaptic mechanisms and synaptic plasticity, and mice in which this receptor has been disrupted by gene targeting fail to establish long-term potentiation (51), a cellular correlate of processes that correspond to establishing learning and memory. In terms of the present study, preconditioning neurons with nicotine results in an outcome that favors retention of GluR1. The delayed removal of these key regulatory sequences from this protein would alter both its concentration and/or function and in turn affect the overall GluR system response, which could have significant effects on the animal. Consistent with this possibility is the finding that when rats are chronically administered nicotine, the threshold required to initiate long-term potentiation is lowered; this is related to nAChR desensitization (52), although other possible mechanisms through which nicotine may precondition the neuronal GluR response have not been fully investigated. For example, nAChRs are important modulators of GABA release, and in our cultures nAChRs are expressed by glutamic acid decarboxylase positive neurons (not shown). Therefore, desensitization of nAChRs could reduce GABA function and enhance processes such as long-term potentiation, as seen when GABA receptor function is diminished by bicuculline (51). However, the results of preliminary experiments suggest that bicuculline has no effect on activity-dependent GluR1 degradation (data not shown) as would be expected if nAChR modulation of the GABA system were required. Another possibility is that decreased nAChR function would reduce free calcium, which could impact upon calpain activation of caspases (53,54) or the direct cleavage of GluR1 by this protease (25,35). Although this study does not directly address issues pertaining to calpain activation, we found no inhibition of activity-initiated nicotine-sensitive caspase cleavage by the presence of EDTA or the calpain inhibitor, PD150 - FIG. 4. Nicotine pretreatment alters AMPA-initiated caspase 8 activation in neurons. A, cultured cortical neurons were pretreated with 1 M nicotine for Ͼ24 h and then loaded with a fluorogenic substrate of Csp8 (Z-IETD-AFC; see "Materials and Methods"). AMPA (100 M) was added, and the subsequent activation of Csp8-like activity was recorded by visualizing the appearance of fluorescent product at 1-min intervals thereafter. These data show that nicotine did not completely inhibit Csp8-like activities, but nicotine pretreatment did delay the agonist-initiated accumulation of fluorescent product. Similar results were obtained when kainic acid was used in place of AMPA (not shown). The arrow identifies one cell in the field that exhibited no nicotine-mediated delay of Csp8-like activation. Quantitation of multiple experiments is shown in B. Nicotine delayed the activation of Csp8-like activity and reduced the amount of fluorescent product accumulated by approximately one-third for the time period tested. Error bars are ϮS.E. 606 (see "Materials and Methods"; not shown), which suggests that agonist-initiated caspase cleavage occurs independently of calcium-activated proteases. Finally an intriguing possibility to explain how the rapid AMPA-initiated caspase-like cleavage of GluR1 can be altered by nicotine pre-exposure is that chronic nicotine exposure disrupts the aggregation of Csp8 to the vicinity of the GluR1 C terminus. GluR1 is known to associate with multiple cytoplasmic proteins and to form multiprotein complexes through its C terminus (2,17,32). Notably pro-Csp8 is converted to the active enzyme upon aggregation (55), and an attractive hypothesis is that activity-dependent aggregation of this protease would initiate the highly specific cleavage of GluR1. If this two-step process (aggregation followed by activation) is disrupted (as by chronic desensitization), a delay in activation of Csp8, as suggested by the data in Fig. 4, would be expected. Although AMPA-initiated Csp8 activation does occur in nicotine pretreated neurons, the altered rate of initiation of proteolysis may directly impact upon substrate proteins targeted for proteolytic modification.
A final comment concerns the lack of increased apoptosis in our system (Refs. 15 and 16; data not shown) despite the activation of an initiator caspase in proteolytic cascades that lead to apoptosis by peripheral cells (28,55,56). Others (e.g. see Ref. 57) have also observed a limited role for caspase activation and cleavage without ensuing apoptosis. This possibility is particularly attractive for neurons because of the stringent partitioning of organelles and proteins to sub-compartments by this cell type. Therefore, if Csp8 localizes to the dendrite or possibly protein complexes in association with surface receptors where other components of the proteolytic cascade that lead to apoptosis are absent, the role of this protease may be modified to the performance of local regulatory functions. For example, recent studies have implicated the role for lysosomes in mediating the Csp9/Csp3 apoptotic cascade (58 -61). In neurons, this pathway would likely be curtailed because in healthy neurons lysosomes are rarely found in dendrites. In any case, because Csp8 is an initiator caspase, the modulation of its activity through cholinergic mechanisms reflects a novel way in which this neurotransmitter receptor system can impact upon the function of cellular proteolytic systems to influence normal neuronal function and survival in pathology.