Nicotinic Receptor Subunit α5 Modifies Assembly, Up-regulation, and Response to Pro-inflammatory Cytokines*

In the mammalian brain high affinity nicotine-binding sites are composed of at least the α4 and β2 subunits. Additional nicotinic acetylcholine receptor subunits that are often co-expressed with α4+β2 include α5. The introduction of α5 into 293 cells expressing α4+β2 strongly favors assembly of α4+α5+β2 receptors, increases constitutive ligand binding density as measured using [3H]epibatidine, but reduces the magnitude of up-regulation in response to chronic nicotine. In contrast, when β4 is substituted for β2, α5 interferes with the assembly of these receptors, demonstrating an important role for the β subunit in this process. When cells co-express α4+α5+β2+β4, over 50% of the subunit associations include all four subunits, but they fail to be detected using [3H]epibatidine binding. However, complexes of α4+α5+β2 do preferentially emerge from these subunit mixtures, and these mixtures bind ligand. In previous studies of α4+β2+β4 co-expression by 293 cells, the inflammatory cytokines IL-1β and TNFα influenced the outcome of receptor assembly (Gahring, L. C., Days, E. L., Kaasch, T., González de Mendoza, M., Owen, L., Persiyanov, K., and Rogers, S. W. (2005) J. Neuroimmunol. 166, 88–101). When α5 is included in this subunit mixture, and cells are exposed to either inflammatory cytokine, subunit association is no longer altered. These findings suggest that α5 is an influential modulator of α4+β2 nicotinic acetylcholine receptor assembly and stabilizes their expression in response to fluctuations in external conditions.

In the mammalian central nervous system a key nicotinic receptor (nAChR) 2 is the high affinity nicotine-binding site that undergoes up-regulation in response to chronic exposure to ligands such as nicotine (1)(2)(3)(4). Up-regulation can increase high affinity ligand binding densities 3-6-fold, and this change has been directly related to many unique features of nAChR biology including the addiction phenotype. The high affinity nicotinebinding site that is up-regulated by chronic ligand exposure was revealed through immunological methods to be composed almost entirely of receptors containing ␣4 and ␤2 subunits (4). This result was confirmed by mouse genetic approaches where ablation of either of these subunits effectively eliminates almost all high affinity binding sites in the brain (5,6). However, the mechanisms governing the up-regulation process without significant compromise to their normal physiological roles or in extreme environments as during exposure to chronic nicotine or at sites of inflammation remain to be clarified (for discussions see Refs. 7 and 8).
Heterologous expression systems, which closely parallel results from in vivo studies (8), have been employed to show that nAChRs from subunit combinations such as ␣4ϩ␤2, ␣4ϩ␤4, or ␣4ϩ␤2ϩ␤4 (9 -13) readily interact to form functional receptors. But in the brain, subunit expression and possible interactions are not this simple. For example, in the adult brain ␤4 expression is limited to only a few brain regions, whereas other subunits such as ␣5 are widely expressed. Although described as an ␣ subunit, ␣5 lacks key residues important for ligand binding and is therefore considered a structural subunit (8). Nevertheless, its inclusion impacts upon many receptor functions. For example, in the absence of ␣5 there is a shift in receptor efficacy for multiple pharmacological agents, changes in channel ion permeability, and alterations to neurotransmitter release (8,15,16). These differences extend to the mouse where the absence of ␣5 impacts upon sensitivity to nicotine-induced locomotor activity and seizure (17) as well as the susceptibility to inflammatory bowel disease (18). In humans genetic findings suggest that a polymorphism in the cytoplasmic domain of ␣5 corresponds with susceptibility to early life abuse of nicotine and subsequent addiction (19,20). Some of these effects may be explained by the finding that the majority of ␣4ϩ␤2 receptors in the rodent brain also contain ␣5, and these receptors exhibit limited up-regulation in response to nicotine (14). However, differing cell types and brain regions do differ quantitatively in ␣5 inclusion, indicating that regional heterogeneity in the effect by this subunit is likely to be present (8,14,(21)(22)(23)(24). Thus, understanding the impact of ␣5 on nAChR assembly and response to the environment is relevant to understanding basic mechanisms contributing to the receptor assembly process.
In this study we examined the role of ␣5 in modulating ␣4ϩ␤2 receptor assembly in the transiently transfected 293 cell model system as measured by the binding of [ 3 H]epibatidine ([ 3 H]EB). To summarize, ␣5 readily enters into stable interactions with ␣4ϩ␤2 as measured by [ 3 H]EB binding. Although ␣5 incorporation strongly favors binding sites composed of ␣4ϩ␤2ϩ␣5 subunits, it interferes with expression of ligandbinding sites if ␤4 is substituted for ␤2. In terms of up-regulation to nicotine, ␣5 lessens this response. Finally, the co-expres-sion of ␣5 with ␣4ϩ␤2, ␣4ϩ␤4, or ␣4ϩ␤2ϩ␤4 effectively abolished any impact by either TNF␣ or IL-1␤ on nAChR assembly, which is observed when ␣5 is not present (13). These findings reveal an additional and possibly modulatory role by ␣5 in stabilizing nAChR assembly to rapid shifts in external conditions such as those produced by nicotine or pro-inflammatory cytokines.
Expression Vectors-In all cases rat nAChR subunit cDNAs within the pcDNA3.1ϩ expression vector (Invitrogen) were as used and described previously (13). The epitope-tagged nAChR-␤2HA and nAChR-␤4HA expression vectors were generously provided by Dr. William Green (University of Chicago, Chicago, IL). During the course of these studies, we identified a problem in the sequence of the rat ␣5 cDNA expression construct PC989E(4) (provided by James Patrick, Baylor College of Medicine, Houston TX) (29) (Fig. 1). This expression construct deleted the initial 15 amino acids of the native protein, resulting in an incorrect start methionine. To correct this, standard RT-PCR-based methods were used to generate the original rat ␣5 5Ј-end from a rat brain cDNA pool, which was then subcloned into the pcDNA3.1 expression vector to generate a rat ␣5 expression cDNA construct with the sequence shown in Fig. 1 that is also consistent with the cDNA of other ␣5 subunits from mouse and human. When either of the ␣5 constructs is co-transfected with ␣4 and ␤2 into 293 cells, either methionine is competent as a translational start, and the product interacts with other nAChR subunits. However, translational efficiency of the repaired Met 16 construct, as determined using co-immunoprecipitation and measurement of association with ␣4, was on the order of 10-fold more than constructs with the second methionine (Met 16 ; Fig. 1b). All of the experiments reported in this study used the corrected ␣5 cDNA, whose translation is initiated from the first methionine.
Construction of the ␣5R3b epitope-tagged subunit was done as described for ␣4 (for details see Ref. 13), where the GluR3R3b sequence was introduced into the ␣5 subunit in place of the cysteine loop. These subunits are expressed, and they function similarly to the wild-type counterparts in these measurements (13). Because ␣5R3b is significantly smaller than ␣4R3b, both subunits can be detected simultaneously on Western blots using an anti-R3b epitope-defined monoclonal antibody (2F5).
Cell Culture and Transfection-Human embryonic kidney 293 cells were grown as before (13) under standard tissue culture conditions in DMEM (Cellgro, obtained from Fisher) supplemented with 10% heat-inactivated fetal bovine serum (Hyclone), 1% penicillin/streptomycin (Cellgro, obtained from Fisher), and 1% sodium pyruvate. The jet-PEI system (QBiogene, Irvine, CA) was used for all transfections of 293 cells. Treatment with nicotine and/or optimized pro-inflammatory cytokines (25 ng/ml IL-1␤ or 25 ng/ml TNF␣; both human recombinant) was initiated 2 h after transfection (13). All of the cells were harvested within 24 h after transfection.
Sucrose Gradients-The methods used for sucrose gradient analysis were adapted from other investigators and have been described previously (13,30,31). Briefly, 293 cells transiently transfected with various nAChR subunits were solubilized in Triton X-100 and cleared by centrifugation before being layered onto a 5-ml sucrose gradient (continuous 5-20% (w/w) in 20 mM Tris, pH 7.5) prepared using a Gradient Master 107ip (BioComp Instruments Inc., Fredericton, Canada). The gradi-  (4)), there is an omission of the cytosine base identified by an asterisk that lead to a subcloned variant that lacked the first methionine (circled), thus forcing sole translation of the ␣5 cDNA protein product to initiate from the second methionine (boxed and identified as Met 16 ). This sequence was corrected through RT-PCR methods of native ␣5 mRNA (see "Experimental Procedures") to generate a cDNA that includes the longer ␣5, including an initiating methionine at residue 1 (circled). This also brings the ␣5 cDNA into agreement with those from the mouse (GenBank TM DQ318788 and C57BL6/J) and human (GenBank TM NM_000745), respectively. (B) As shown using Western blot analysis of 293 cells transfected with ␣4ϩ␤2 and either the corrected ␣5 or the Met 16 start methionine, expression of ␣5 can proceed from either methionine, although the first initiating methionine is clearly favored.
ents were centrifuged at 54,000 rpm for 2 h (SW55Ti rotor; Beckman Instruments, Fullerton, CA) at 4°C. The fractions were collected using an FC203B fraction collector with Auto Densiflow (Labconco, Kansas City, KS). The Svedberg markers run in a separate gradient from the samples were bovine serum albumin, 4.2 S (Pierce); ␤-amylase, 8.9 S (Sigma); catalase, 11.2 S (Sigma); and purified 20 S proteosome subunit, 20 S (a gift from Dr. M. Rechsteiner, University of Utah, Salt Lake City, UT).
Two-dimensional Blue Native Gel Electrophoresis (BNG)-Samples were prepared from transfected 293 cells as described previously (13). Briefly, crude membranes were suspended in buffer (aminocaproic acid, 50 mM Bis-Tris, pH 7.0, 2% dodecylmaltoside, 0.5 mM EDTA, and 5% (w/v) Coomassie Blue G), and these samples were resolved using blue native gradient gel (4 -20% acrylamide) electrophoresis. This gel was soaked in a SDS ϩ dithiothreitol solution, rotated 90°C, and placed onto the second denaturing SDS-PAGE gel for size fractionation. Analyses thereafter used standard Western blot methods. Molecular weight markers were from the Amersham Biosciences high molecular weight calibration kit.
[ 3 H]Epibatidine Binding-Radioligand binding of transiently transfected 293 cells was as before (32). Crude membrane fractions were prepared from transfected cells following brief homogenization and collected by centrifugation for 20 min at 20,000 ϫ g at 40°C. Binding was for 4 h at room temperature of 5 nM of [ 3 H]epibatidine (PerkinElmer Life Sciences) to 20 g of protein/sample. All of the samples were measured in triplicate. Nonspecific binding included an additional sample set where the membranes were preincubated for 30 min with nicotine (300 M) before adding the radiolabeled compound (26). The samples were vacuum-filtered through GF/C glass filters (Whatman International), and standard methods of scintillation counting using a Beckman scintillation counter with internal standards followed. Specifically bound counts were calculated by subtracting counts remaining in the nicotine preincubated control sample from the total bound fraction.

RESULTS
The Association of ␣5 with ␣4 Varies According to Which ␤ Subunit Is Co-expressed-To determine whether ␣5 modifies nAChR assembly and influences the impact of nicotine or cytokines on this process, we began by examining how ␣5 interacts to form receptors. To measure this, we used [ 3 H]EB binding to membranes from cells transfected with either ␣4ϩ␤2 or ␣4ϩ␤4. In these experiments (Fig. 2), the cDNA concentrations of the ␣4 and relevant ␤ subunit were held constant (0.25 g/transfection, respectively), whereas the concentration of ␣5 input cDNA was varied from 0, 0.1, 0.25, to 0.5 g, respectively. After incubation, the cells were solubilized, divided into equal fractions, and subjected to immunoprecipitation with the specified anti-␤ or anti-␣5 subunit antibody (see "Experimental Procedures"). Immunoprecipitates were fractionated by SDS gel electrophoresis, and the association with ␣4 was revealed using Western blot analysis ( Fig. 2A). In starting mixtures of ␣4ϩ␤2, increasing the amount of ␣5 corresponds with more ␣5 in association with ␣4 as well as between ␣4 and ␤2. The increased association of ␣5 with ␣4 and ␤2 also corresponds with increased [ 3 H]EB binding, indicating that receptor assembly was promoted by ␣5 (Fig. 2B). When input cDNA concentrations were equivalent for all three subunits (25 g each), the amount of total binding detected was ϳ4-fold greater than transfections of ␣4ϩ␤2 not containing ␣5. To determine whether ␣5 altered agonist-mediated up-regulation, this experiment was also repeated in the presence of chronic nicotine (1 M) during the duration of the transfection. As shown in Fig.  2B, the relative proportion of increased [ 3 H]EB binding in cells treated with nicotine versus those not exposed to nicotine val-FIGURE 2. The ␣5 subunit enters into stable complexes with ␣4 and up-regulation to nicotine is reduced. 293 cells were co-transfected with varying amounts of ␣5 cDNA (as indicated) with constant cDNA amounts for either ␣4 and ␤2 or ␣4 and ␤4, respectively. A, immunoprecipitation with either anti-␣5 or the appropriate anti-␤ subunit was followed by Western blot analysis to determine the amount of associated ␣4. Quantitation of the blots in this panel shows that as the amount of ␣5 cDNA co-transfected with ␣4ϩ␤2 subunit is increased, associations between ␣4 and both ␤2 and ␣5, respectively, also increase. In contrast, when ␤4 is substituted for ␤2, the amount of ␣4 associated with ␤4 diminishes, whereas ␣5 is enhanced. All of the values were normalized to the no ␣5 input values, and the error bars were calculated and are shown as Ϯ S.E. for the results from three to seven independent experiments. B, [ 3 H]EB ligand binding to membranes from cells transfected as in A and membranes from cells treated with 1 M nicotine (gray bars). Increased ␣5 corresponds with decreased up-regulation induced by nicotine (fold increase is noted above each treatment set). Each experiment was conducted three to eight times, and the error bars reflect Ϯ S.E. for all experiments.
idates that these complexes do undergo nicotine-mediated upregulation. However, the relative degree of up-regulation diminished from almost 6-fold in non-␣5-transfected cells to only 2-fold in cells co-expressing ␣5 (0.5 g ␣5 cDNA transfected). Thus, ␣5 increases the number of ligand-binding sites in receptors composed of ␣4ϩ␤2 subunits, but the relative amount of nicotine-mediated up-regulation is diminished. This result is consistent with similar reports of ligand binding but poor up-regulation of ␣4␣5␤2 receptors in the rodent brain (14).
To assess the importance of the ␤ subunit in this process, mixtures where ␤2 was replaced by ␤4 were examined. Again, as the ␣5 concentration increases so do ␣5 associations with ␣4; however, associations between ␤4 and ␣4 are notably decreased (Fig. 2A). The impact by ␣5 occurs as the input cDNA concentration is equivalent to or exceeds that of ␤4 cDNA. The decreased ␣4ϩ␤4 association is accompanied by a decrease in [ 3 H]EB binding, suggesting disruption of ligand-binding complexes by ␣5 (Fig. 2B).
The ␣4␤4 receptors also exhibit up-regulation to chronic nicotine (1.5-fold), but this is much less than for ␣4␤2 receptors (9, 10, 32). However, upon increasing the ␣5 contribution, upregulation is completely abolished. Therefore, although ␣5 readily forms subunit associations in cells transfected with ␣4ϩ␤4, the net result is reduced [ 3 H]EB binding, opposite that seen for ␣4ϩ␤2 transfected cells. In both cases, inclusion of ␣5 corresponds with the overall reduced up-regulation in response to nicotine.
Cells Co-expressing ␣4ϩ␣5ϩ␤2ϩ␤4-As reported previously (13) when equal amounts of ␣4, ␤2, and ␤4 cDNA are co-transfected into 293 cells, a characteristic distribution of mature receptors of three different subunit composition results (␣4ϩ␤2, 15%; ␣4ϩ␤4, 55%; and ␣4ϩ␤2ϩ␤4, 30%). We examined how ␣5 impacts upon this distribution. For these experiments 293 cells were co-transfected with a constant amount of cDNA encoding ␣4ϩ␤2ϩ␤4 (0.25 g each) and varied amounts of ␣5 (0, 0.1, 0.25, and 0.5 g, respectively). The transfected cells were harvested and subjected to successive immunoprecipitation (4,13,14) where ␤2-containing complexes were removed first followed by a second immunoprecipitation with anti-␤4 (Fig. 3). The products of each immunoprecipitation were resolved by SDS-PAGE and the amount of ␣4 and ␣5 (when present) in association with the subunit being precipitated was measured by Western blot (Fig. 3A). Because both ␣4 and ␣5 are identified with the same epitope tag, both subunits can be resolved simultaneously on the blot by the difference in molecular weights (␣4, ϳ72 kDa; ␣5, ϳ48 kDa). In some experiments, the anti-␣5 antibody was used first to remove complexes containing ␣5 (not shown). The relative intensity of each band from the respective immunoprecipitation was then quantitated as shown in Fig. 3A. A summary of the results derived in percentages of complexes is shown (Fig.  3B). The results were similar to those obtained before (13) where receptors of ␣4ϩ␤2, ␣4ϩ␤4, and ␣4ϩ␤2ϩ␤4 were generated in different but distinct ratios. Associations between ␣4 and ␤4 are favored over ␣4 and ␤2 in the absence of ␣5 (not shown). As ␣5 input cDNA increased so did the proportion of ␣4 in association with ␣5. This included a dra-matic increase in ␣4ϩ␣5ϩ␤2ϩ␤4 complexes that reached on average 57% of the total associations in mixtures containing 0.5 g of ␣5 cDNA. However, there was a progressive decrease in [ 3 H]EB binding (Fig. 3C). The reduction to 65% is significant when equal cDNA input is used (0.25 g of each subunit; p Ͻ 0.03), and ligand binding decreases further to 44% when ␣5 amounts exceed those of other subunits. This decrease was not due to cell death because trypan blue exclusion was not increased relative to the other transfections (not shown). Notable is the convergence of values between the 57% value found for the production of ␣4ϩ␣5ϩ␤2ϩ␤4 complexes and the 56% decrease in ligand binding for cells receiving 0.5 g of ␣5 cDNA. Finally, complexes such as ␣4ϩ␣5ϩ␤4 were strongly disfavored (see the gel Fig. 3A where anti-␤4 immunoprecipitation follows anti-␤2). Collectively these results indicate that ␣5 is a favored assembly partner with ␣4 and ␤2. However, other subunits such as ␤4 alter this preferred association pattern, presumably leading to nonproductive subunit interactions.
To clarify further the receptor subunit distribution and ligand binding from emerging from mixtures of ␣4ϩ␣5ϩ␤2ϩ␤4, a combination of sucrose density centrifugation, ligand binding, and two-dimensional blue native gel electrophoresis (BNG) was used. Detergent-solubilized lysates from cells co-transfected with equal amounts of each cDNA (␣4ϩ␤2ϩ␤4 with and without ␣5; 0.25 g/subunit) were subjected to sedimentation velocity centrifugation through sucrose gradients. Individual fractions were subsequently analyzed by SDS-PAGE. Because each subunit migrates on SDS-PAGE at different molecular weights, they could each be detected simultaneously on Western blots, and specific [ 3 H]EB binding to corresponding fractions of gradients prepared in parallel could be measured.
The majority of ␣4-containing receptor complexes in the absence of ␣5 migrate as a peak fraction of ϳ10 S (Fig. 4A). To examine this region of the gradient in greater detail, smaller fractions (100 l) were collected between the 9 and 11 S marker fractions to increase the resolution of receptor distribution as reflected by the distribution of ␣4, which localized mostly to a peak consistent with the 10 S fraction (Fig. 4A). When ␣5 was included in the transfection mixture, the sucrose gradient profile was altered. Although there was an apparent increase in the amount of ␣4 and ␣5 in the 10 S peak, the presence of a prominent 10.5 S shoulder was also detected, indicating that ␣5 shifts, at least in part, the receptor shape.
The binding of [ 3 H]EB to protein within the gradient fractions collected in parallel are also consistent with a 10 and 10.5 S distribution that is largely dependent upon the presence of ␣5 (Fig. 4B). As shown there, without co-expression of ␣5, the majority of ligand binding is associated with the 10 S fraction. However, a persistent 10.5 S shoulder is distinguished by ligand binding that is not resolved well using Western blot analyses. Possible explanations for a shift to a 10.5 S shoulder include altered binding conformation, changes in glycosylation, binding of other intracellular proteins or altered receptor subunit stoichiometry. What is clear is that when ␣5 is present, ligand binding in the 10.5 S peak is increased substantially, although binding in the 10 S peak remains the same. Thus, ligand binding and Western blot profiles of ␣4 containing fractions between the 9 and 11 S markers correspond well with at least two major binding components of 10 and 10.5 S, but the 10.5 S fraction is enhanced by the presence of ␣5, suggesting an altered receptor shape dependent upon inclusion of ␣5.
To investigate further the receptor subunit differences on receptor shape suggested by sucrose gradient analysis of ␣4ϩ␣5ϩ␤2ϩ␤4 mixtures, we used two-dimensional BNG (see "Experimental Procedures"). This method is particularly useful for resolving the subunit composition of intact protein complexes including ligand gated ion channels, and we introduced this method to examine the relative subunit composition of nAChRs assembled from ␣4ϩ␤2ϩ␤4 subunit mixtures (13). A gel typical of the two-dimensional BNG method produced from cells co-transfected with ␣4ϩ␣5ϩ␤2ϩ␤4 is shown in Fig. 4C.
In this system ␣5 is present in large complexes consistent with mature nAChRs as well as in a single spot at ϳ55 kDa that is likely to be monomeric subunits. Similarly, ␤4 is present in a ϳ100-kDa spot that is of dimer size. Although the majority of nAChR immunoreactivity co-localizes in the complex mixtures of different ␣4ϩ␣5ϩ␤2ϩ␤4 combinations, three prominent components are resolved (Fig. 4C). First, the largest complex resolved by two-dimensional BNG is composed of ␣4ϩ␣5ϩ␤2.
Because this peak appears only in ␣5 transfected cells, it is likely to be related to the 10.5 S peak seen in sucrose gradients reported above. The second complex, the most prominent one in terms of protein content, is composed of a mixture of ␣4ϩ␣5ϩ␤2ϩ␤4 subunits, which are likely to reflect the relatively complex mixture of ␣4-containing assemblages suggested by the results of Fig. 3. Finally there is a smaller sized fraction that is also consistent with ␣4ϩ␣5ϩ␤2 composition. The presence of this additional peak suggests that ␣4ϩ␣5ϩ␤2 receptors are present in two conformations or possibly bound to other intracellular proteins whose identity will require further investigation. . Crude membranes were solubilized in detergent and subjected to progressive immunoprecipitation with either anti-␤2 followed by anti-␤4 or reciprocally anti-␤4 followed by anti-␤2 as described (see "Experimental Procedures"). A, representative results from cells transfected as described above. A Western blot reveals the relative amount of R3b epitope-tagged ␣4 or ␣5 as indicated in association with ␤2 or ␤4 after removal of all ␤2 complexes. Quantitation of these blots is shown where all of the values are normalized to samples with no ␣5 co-transfected. The error bars reflect plus or minus the standard error of the mean as calculated from three independent experiments done in the same manner. B, average percentages of the relative amounts of mixed subunit complexes that were either directly measured or derived from measuring the relative amount of ␣4 remaining after successive anti-␤ subunit immunoprecipitations as in A. The averages are from three independent experiments. C, ligand binding ([ 3 H]EB) to total membranes decreases as ␣5 input is increased. The loss of ϳ56% of the total radioligand binding relative to non-␣5 transfection mixtures corresponds well with the increase of ␣4␣5␤2␤4 complexes (B).
If indeed a receptor of ␣4ϩ␣5ϩ␤2 is generated from a subunit mixture of ␣4ϩ␣5ϩ␤2ϩ␤4, then this complex should be evident in samples where all ␤4-containing complexes are removed before gradient analysis. To test this, 293 cells were transfected with ␣4ϩ␣5ϩ␤2ϩ␤4 (0.25 g of cDNA/subunit), the solubilized membranes were cleared by immunoprecipitation with anti-␤4, and the remaining complexes were analyzed using sucrose gradient density centrifugation. The results in Fig. 4D show the band density profiles of a Western blot from these gradients. As predicted, there was essentially no detectable ␤4 (not shown), although ␣4ϩ␣5ϩ␤2 proteins remained. Of these, greater than 80% of all receptor-related protein signals were in the 10 -10.5 S fractions, although the 10.5 S peak dominated. A smaller fraction closer to 9.5 S in these gradients harbored almost equivalent amounts of ␣4ϩ␣5ϩ␤2 complexes (Fig. 4D), as did a large amount of ␣5 immunoreactivity near the FIGURE 4. Sucrose density gradients and two-dimensional BNG reveals ligand-binding sites of the ␣4؉␣5؉␤2 subunit composition. A, cells were transfected with equal amounts of cDNA (0.25 g of each) encoding ␣4R3b and ␤2-HA with or without ␣5R3b. Solubilized membranes from these cells were subjected to sucrose gradient sedimentation, and gradient fractions were collected for analysis of the epitope marked ␣4 and ␣5 subunits (when present) using SDS-PAGE gel electrophoresis. As shown, the majority of ␣4 is localized to fractions consistent with a 10 S complex. The profile revealed by ␣4 immunoreactivity results when gradient resolution between the 9 -11 S markers is increased. When ␣5 is included in the transfection mixtures, there is a broadening of the sedimentation profile curve to include a new distribution covering the 10 -10.5 S range. B, the corresponding [ 3 H]EB binding to similar gradient fractions is shown. Inclusion of ␣5 increases the amount of ligand bound to the 10.5 S peak over the 10 S peak similar to the distribution of the ␣4 or ␣4ϩ␣5, respectively, measured by Western blots. C, mixtures of ␣4ϩ␣5ϩ␤2ϩ␤4 were co-transfected, and the cell membranes were prepared for twodimensional BNG analysis. All of the subunits were detected using Western blot analysis for epitope-tagged ␤2 and ␤4 followed by stripping and reprobing to reveal both epitope-tagged ␣ subunits. The gel was then reassembled electronically to show all of the subunits as identified. Most immunoreactivity is of various (mixed) subunit composition, although two major complexes (designated as ␣4␣5␤2) do not appear to include ␤4. Possible ␤4 dimers and ␣5 monomers are marked. In the lower left gel, ␣5 was omitted from the subunit co-transfection mixture (only ␣4 is shown for clarity); the majority of ␣4 resides in a smear of complexes, although a prominent peak of immunoreactivity as measured in the gel densitometry tracing below the blot is observed. However, when ␣5 is included in the transfection mixture, the molecular weight of the overall complex is shifted to add a "lower" molecular weight component that includes both ␣4 and ␣5 species. These are the same complexes that are dominated by the presence of ␣4ϩ␣5ϩ␤2 in the reference gel above. D, to determine whether ␣4ϩ␣5ϩ␤2 subunits are present, ␣4ϩ␣5ϩ␤2ϩ␤4 were co-transfected (equivalent amounts of cDNA as above) only before separating the complexes on sucrose gradients, complexes harboring ␤4 were removed using anti-␤4 immunoprecipitation. The remaining complexes were then separated on a sucrose gradients, the fractions were collected, the Western blots were prepared, and the density traces of the two-dimensional analysis were done as in C. As shown in the line profiles, two prominent peaks of the remaining ␣4, ␣5, and ␤2 were present that include one at the 10.5 S peak and another near 10 S. A third and smaller peak found in immunoprecipitated samples was composed of mixed subunits of near equivalent amounts at ϳ9.5 S and presumed monomers of ␣5 near the top of the gradient. The results reflect data collected from between two and six independently performed experiments. gradient top, which is consistent with the monomers suggested by two-dimensional BNG analysis (Fig. 4C).
These results show receptor complexes harboring ␣4 in combination with ␤2 and/or ␤4 that bind [ 3 H]EB as reported previously (13,30,33). The addition of ␣5 produces both ␣4ϩ␣5ϩ␤2 complexes consistent with a 10.5 S peak revealed by Western blot and ligand binding. Other complexes of mixed ␣4ϩ␣5ϩ␤2ϩ␤4 subunits are also present as predicted from the results in Fig. 3.
IL-␤ and TNF␣ Do Not Impact upon Assembly of ␣4-containing nAChRs when ␣5 Is Co-expressed-Pro-inflammatory cytokines also impact upon the expression of receptors assembled from ␣4 and various mixtures of ␤2 and/or ␤4 subunits (13). The ␣4-based subunit associations measured above were repeated in the presence of ␣5. For these experiments, the conditions described above were repeated, but cells were also treated with either IL-1␤ or TNF␣. Crude membranes were then prepared from cells of each treatment group and analyzed for the relative amount of ␣4 co-immunoprecipitated with either of the respective ␤ subunits or ␣5. As shown in Fig. 5A, the results from these experiments reveal no impact by cytokines on subunit associations in the presence of ␣5 (compare results with those in Fig. 2A and Ref. 13). When complex mixtures of subunits were co-transfected, and subunit profiles of the receptors were examined using two-dimensional BNG (Fig.  5B), the receptor profiles emerging, regardless of IL-1␤ or TNF␣ treatment, exhibited equivalent assembly to the control. Thus, the assembly of ␣4 when ␣5 is present is unaffected by the presence of these pro-inflammatory cytokines.

DISCUSSION
The results of this study suggest that ␣5 promotes ␣4ϩ␤2 associations but interferes with ␣4ϩ␤4 as detailed in Fig. 6. A possible explanation for this result is that in the starting subunit pools of ␣4ϩ␤2, ␣5 provides an "extra" subunit to promote direct assembly into mature receptors as defined by ligand binding. However, if the starting pool is ␣4ϩ␤4, then ␣5 binding to ␣4 is favored but occurs at the expense of ␣4ϩ␤4 interactions, resulting in less efficient production of mature receptors. This basic priority in subunit association is retained in more complex subunit mixtures such as ␣4ϩ␣5ϩ␤2ϩ␤4, where again ␣5 competes with ␤4 for association with ␣4, leading to complexes that fail to bind ligand as well as promote the final outcome of favoring ␣4ϩ␣5ϩ␤2 complexes.
In terms of up-regulation, one of several mechanisms suggested to regulate this process (for review and discussion see Ref. 8) is improved assembly caused by stabilization of associations between ␣ and ␤ subunits upon binding of ligands such as nicotine (45). However, because ␣5 strongly favors ␣4ϩ␤2 assembly, binding of ligand to ␣4ϩ␣5ϩ␤2 receptors would have only a minimal effect on subsequent events related to upregulation. In turn, this also reveals the importance of the ␤ subunit to this process. Because the presence of ␤4 favors ␣4ϩ␣5 associations, which do not form a ligand-binding site, both up-regulation and overall assembly of receptors from these subunit combinations would be discouraged. This is consistent with other reports where the role of ␣5 in receptor formation in the presence of different ␤ subunits has been examined. For example, Lindstrom and colleagues (34,35) found that FIGURE 5. The presence of ␣5 moderates the influence of the pro-inflammatory cytokines IL-1␤ and TNF␣ on nAChR receptor assembly. A, similar to experiments in Fig. 2 only IL-␤ or TNF␣ was added for the duration of the transfection (control gels are the same as in Fig. 2). Western blot analysis of cells co-transfected with fixed amounts of ␣4 and the indicated ␤ subunit and increasing amounts of ␣5 cDNA. The relative amount of association of ␣4 with the indicated ␤ subunit is detected using Western blot, and the relative band density is plotted. B, in similar experiments cell membranes were examined by two-dimensional BNG analysis, and the results for just ␣4 and ␣5 are shown for clarity. Plots of ␣4 immunoreactivity in cells treated with either IL-␤ or TNF␣ and co-transfected with ␣5 or not co-transfected (as labeled) are placed below the blots. Control experiments were conducted between three and eight times as in Fig. 2. Cytokine experiments are based upon an n value of 3.
␣5 promoted receptor assembly of ␣3ϩ␤2 receptors, but assembly was diminished if ␤4 was substituted for ␤2. Also similar to the present study, ligand binding was present only in regions of sucrose gradients harboring receptors of pentameric structure as measured by centrifugation and inclusion of the ␣5 subunit.
These studies also appear to be relevant to nAChRs identified in the animal. Chick ciliary ganglia that express predominantly ␣3 subtype receptors are most prevalent in ganglia, where 80% of these expressed were co-assembled with ␤4 and ␣5, whereas others harbor ␣3ϩ␣5ϩ␤2ϩ␤4 (36). Similar detailed examinations of the coincident expression of nAChR subunit expression by a subset of inhibitory interneurons in the CA1 hippocampal region suggest that although ϳ85% of these express ␣4 also co-express ␣5, there are those that do not (8,21,24). Thus, in a model of local cellular heterogeneity of nAChR expression, our findings would predict that the variability in the co-expression of ␣5 with varying ␤ subunits would render some cells highly resistant to change promoted by exogenous conditions, whereas others could modify their receptor expression. This could also address the issue of the relatively limited evidence for the presence of functional ␣4ϩ␤4 receptors in the brain despite multiple lines of evidence that support the expression of this subunit (46 -48). In this case the production of ␣4ϩ␣5ϩ␤2 receptors or possibly ␣3ϩ␤4(Ϯ␣5) receptors (37,38) would be favored, whereas those of ␣4ϩ␤4 composition would fail to be detected.
Finally, nAChRs can modulate and be modulated by endogenous agents controlling the pro-inflammatory environment (32, 39 -43). As noted earlier ␣7 nAChRs suppress inflammatory responses in neuronal and non-neuronal systems including the digestive system, lung, and skin (39 -44). However, in turn IL-1␤ or TNF␣ significantly modifies certain nAChR receptor assembly where IL-1␤ favors ␣4ϩ␤2 assembly, and TNF␣ favors the formation of effectively any transfected ␣4ϩ␤2ϩ␤4 receptor subunit combination (13). TNF␣ dramatically enhances nicotine-mediated up-regulation of ␣4ϩ␤2 receptors through a mechanism that requires participation of the p38 MAPK pathway (32). However, in the current study the presence of ␣5 stabilizes assembly from these agents. Thus, in addition to ligand, the inflammatory environment may also impact the expression of certain nAChRs. However, these results also suggest that the occurrence of reciprocal regulatory interactions between the pro-inflammatory and nicotinic cholinergic receptors systems would be facilitated by the presence or absence of ␣5 and stabilization of key receptor expression.