Abstract
Endogenous steroids can modulate the activity of transmitter-gated channels by directly interacting with the receptor. 17β-Estradiol potentiates activation of neuronal nicotinic α4β2 receptors by interacting with a 4 aa sequence at the extreme C terminus of the α4 subunit, but it is not known whether potentiation requires that the sequence be placed on a specific subunit (e.g., an α4 subunit that is involved in forming an acetylcholine-binding site). By using concatemers of subunits and chimeric subunits, we have found that the C-terminal domain can be moved from the α4 to the β2 subunit and still result in potentiation. In addition, the sequence can be placed on a subunit that contributes to an acetylcholine-binding site or on the structural subunit. The data indicate that this estradiol-binding element is a discrete sequence and suggest that the effect of 17β-estradiol is mediated by actions on single subunits and that the overall consequences for gating occur because of the summation of independent energetic contributions to overall gating of this receptor.
Introduction
Steroids are endogenous modulators of membrane channel function. Although many actions of steroids are mediated by alterations in gene expression initiated by binding to nuclear receptors, steroids can have rapid and reversible actions on both transmitter-gated and voltage-gated ion channels (Belelli and Lambert, 2005; Schlichter et al., 2006). A particularly well studied example is the GABA type A (GABAA) receptor, for which neurosteroids are among the most potent and efficacious potentiators (Belelli and Lambert, 2005; Akk et al., 2007). The fact that steroids can modulate the function of synaptic receptors provides a rapid link between endocrine and nervous system functions.
The ligand-gated ion channel gene family includes the subunits for the vertebrate nicotinic, GABAA, serotonin type 3, and glycine receptors, and a number of related proteins in invertebrates (Brejc et al., 2001; Akabas, 2004; Sine and Engel, 2006). These receptors form as pentamers of homologous subunits (see Fig. 1), arranged in a rosette around a central ion channel formed from membrane-spanning α-helical regions contributed from all subunits. We are studying the neuronal nicotinic receptor containing α4 and β2 subunits, to define the sites and mechanisms by which potentiating agents act on the receptor. The α4 subunit is expressed in many brain regions, and the α4β2* receptor is one of the most common receptor subtypes (Gotti et al., 2007). The major physiological role of these receptors is to modulate the release of other neurotransmitters (Dani and Bertrand, 2007). The endogenous steroid 17β-estradiol potentiates the response of the α4β2 nicotinic receptor and requires a specific amino acid sequence at the extreme C terminus of the α4 subunit (Paradiso et al., 2001; Curtis et al., 2002). We sought to better define the structural requirements for estradiol potentiation of this receptor.
The sites at which potentiators interact with receptors in the ligand-gated ion channel family have been defined for some drugs. In several cases, the recognition region for the drug is formed by residues from two subunits at an interface between subunits (Hsiao et al., 2006; Moroni et al., 2008; Seo et al., 2009). In others, binding occurs between residues in a single subunit (Jenkins et al., 2001; Hosie et al., 2006; Li et al., 2006). By using concatemeric constructs of subunits and mutated subunits, we examined the effects of placing the α4 C-terminal sequence on either the α4 or β2 subunit, or on subunits that contribute an interface to an ACh-binding site or that serve as the structural subunit (see Fig. 1). Our results show that the C-terminal sequence may be placed on either subunit and the subunit may be in either position in the assembled receptor. These observations indicate that the C-terminal domain is a discrete and transferable element underlying 17β-estradiol potentiation. Potentiation increases geometrically with the number of C-terminal domains in the receptor, which suggests that binding of 17β-estradiol has an independent effect on an individual subunit, which adds a constant amount of energy to stabilize the open-channel form of the receptor.
Materials and Methods
cDNAs and molecular biology.
cDNA constructs for human nicotinic receptor α4 and β2 subunits were kindly provided by J. Lindstrom (University of Pennsylvania, Philadelphia, PA) (α4, accession number NM_000744; β2, accession number NM_000748). The constructs were transferred to pcDNA3 (Invitrogen).
Two concatemers were constructed, which linked two subunits, as previously described (Zhou et al., 2003). From the N to C termini, they are β2-EF(AGS)6-α4 (abbreviated β/α) and α4-EF(AGS)8-β2 (α/β). The signal sequence of the second subunit was deleted. The β/α concatemer is very similar to that described previously (Zhou et al., 2003) [EG(AGS)6], whereas the α/β concatemer has a linker two residues longer than the α4-(AGS)6-β2 concatemer described previously [QEGT(AGS)6TG]. Chimeric subunits were constructed by overlap extension and smaller mutations were constructed using QuikChange (Stratagene). The locations of chimera joining points are shown in Figure 1.
All constructs were sequenced through the entire coding region.
Receptor expression and oocyte voltage clamp.
cRNA was synthesized using the mMessage mMachine T7 kit (Ambion). The concentration of mRNA was estimated from the OD260 value. When combinations of free subunits were injected, the ratio of construct with an α4 N terminus to that with a β2 N terminus was 8:1 (mass ratio), unless otherwise specified. When concatemers were injected with free subunits, the ratio was 2:1.
Xenopus oocytes were prepared in the laboratory of Dr. C. Zorumski (Washington University, St. Louis, MO) using an approved protocol. Oocytes were injected with 12–15 ng of cRNA in a volume of 18–23 nl. Oocytes were maintained at 18°C for 2–7 d before physiological study.
Standard methods were used for two-electrode voltage clamp of Xenopus oocytes (Steinbach et al., 2000; Paradiso et al., 2001; Jin et al., 2009), using an OC-725C voltage clamp (Warner Instruments). Currents were filtered at 20 Hz, and then digitized at 50 Hz (Digidata 1200 interface; Molecular Devices) and stored using pClamp 8.0 (Molecular Devices). Transients were analyzed with Clampfit (Molecular Devices). Oocyte recordings were performed in a small chamber that was continuously perfused with saline. Drug applications were made using a manually controlled perfusion system. The system was made with glass, stainless-steel, or Teflon components, to reduce steroid adsorption. The applications were relatively slow, with bath exchange times of ∼1 s. The external solution contained the following (in mm): 96 NaCl, 2 KCl, 1.8 BaCl2, 1 MgCl2, and 10 HEPES, pH 7.3. External Ca2+ was replaced with Ba2+, to avoid activation of Ca2+ activated channels. We did not use atropine to block muscarinic receptors, as it potentiates α4β2 receptors (Zwart and Vijverberg, 1997). Occasional oocytes showed delayed responses to ACh; these oocytes were discarded.
The concentration–response relationship for activation by ACh was characterized by fitting the Hill equation, Y([ACh]) = Ymax(1/(1 + (EC50/[ACh])∧nHill)), where Y is the response to a concentration of ACh, Ymax is the maximal response, EC50 is the concentration producing half-maximal activation, and nHill is the Hill coefficient. Concentration–response data were collected for an individual cell, and data were normalized to the response to 1 mm ACh. The fit was rejected if the estimated error in any fit parameter was >60% of the fit value, and all parameter estimates for that fit were discarded. The relationship was analyzed for each cell, and then overall mean values were calculated for oocytes injected with that set of constructs.
Potentiation by 17β-estradiol is strongest for low concentrations of ACh (Paradiso et al., 2001; Curtis et al., 2002). Since the EC50 for activation by ACh depends on the subunit combinations expressed (see Results), each oocyte was tested with 1 mm ACh, to estimate the maximal response. A low concentration of ACh, chosen to be able to evoke <20% of the maximal current, was then applied. After the response to ACh had reached a stable level, the application was switched to ACh plus 10 μm 17β-estradiol. The application was switched to bathing solution, followed by repeat of the control low concentration. The relative response in the presence of 17β-estradiol was then calculated. 17β-Estradiol was not preapplied, as the onset and offset of potentiation are rapid (Paradiso et al., 2001). ACh or ACh plus 17β-estradiol were applied for 10–20 s, and applications were separated by 3–4 min, to allow full washout.
The amplitudes of currents expressed by some combinations of constructs were much larger than for others (supplemental Table 1, available at www.jneurosci.org as supplemental material). We are exploring possible reasons for this, but do not have an explanation at present. Experimentally, to accept a value for potentiation by 17β-estradiol, the control response had to be at least 5 nA. For analysis of agonist concentration–response relationships, the maximal response had to be at least 50 nA. To avoid problems of clamp control, the maximal response also had to be <30 μA. For some constructs, the low response amplitude required us to use a higher than usual concentration of ACh, which might have reduced the estimated potentiation (supplemental Table 2, available at www.jneurosci.org as supplemental material).
Values are presented as arithmetic mean ± SE (number of observations). Statistical tests were made using Excel (Microsoft) or Systat (Systat Software). Unless otherwise indicated, statistical tests were two-tailed t tests with unequal variance.
Receptor extraction and Western blots.
Groups of 40–50 oocytes were injected with mRNA, as described previously. Membrane proteins were extracted basically as described previously (Carbone et al., 2009). In brief, oocytes were suspended in an ice-cold homogenization buffer containing the following (in mm): 150 NaCl, 2 CaCl2, 2% Triton X-100, 20 Tris-HCl, pH 7.4, supplemented with protease inhibitor mixture (Sigma-Aldrich; P8465), at a ratio of 5 μl of buffer per oocyte. Oocytes were homogenized by passing through a 20 gauge needle seven times and a 27 gauge needle three times, and then extracted for 30 min on ice. Homogenates were centrifuged twice at 1000 × g for 5 min at 4°C to remove the yolk, and the supernatants were then recentrifuged at 10,000 × g for 10 min at 4°C. The cleared supernatants were collected and diluted 50:50 in 2× Laemmli sample buffer (Bio-Rad) supplemented to 100 mm DTT. Samples were placed at room temperature for 30 min, and then aliquots were loaded on precast 7.5% gels (Bio-Rad). After electrophoresis, proteins were transferred to PVDF (polyvinylidene difluoride) membranes (Millipore). Membranes were blocked for 1 h at room temperature with 100% Odyssey block solution (LI-COR Biosciences), followed by overnight incubation at 4°C in a solution of 50% Odyssey block solution/50% PBS (137 mm NaCl, 2.7 mm KCl, 4.3 mm Na2HPO4, 1.4 mm KH2PO4, pH 7.3) containing 0.4% Tween 20 (Sigma-Aldrich) with primary antibody. Polyclonal rabbit antibody to α4 subunit (sc5591; Santa Cruz Biotechnology) and goat antibody to β2 (sc1449; Santa Cruz Biotechnology) were used at 1:300. Membranes were washed four times with PBS containing 0.2% Tween 20, and then incubated with secondary antibody in the dark for 45 min at room temperature. Goat anti-rabbit and donkey anti-goat labeled with IR dye 680 (LI-COR Biosciences) were used at dilutions between 1:5000 and 1:20,000. Membranes were washed five times with PBS plus 0.2% Tween 20 and scanned on an Odyssey Infrared Imaging System (LI-COR Biosciences).
Drugs.
17β-Estradiol (CAS 50-28-2), acetylcholine chloride (ACh) (CAS 60-31-1), and 19-norpregna-1,3,5(10)-trien-20-yne-3,17-diol, (17α)-(9CI) (17α-vinylestradiol) (CAS 57-63-6) were purchased from Sigma-Aldrich. 5-Iodo-3-(2(S)-azetidinylmethoxy)pyridine (5I A85380) (CAS 213764-92-2) was purchased from Tocris. Steroids were prepared as a 20 mm stock solution in DMSO and diluted into external solution on the day of an experiment. ACh was prepared as a 1 m stock solution in bath solution and stored frozen at −20°C. 5I A85380 was prepared as a 50 μm stock solution in bath solution and stored frozen at −20°C. Working solutions were prepared on the day of experiments.
Characterization of receptors containing concatemers.
We wanted to control the subunit stoichiometry and subunit position in α4β2 receptors. To accomplish this, we generated two concatemers, following the approach described by Zhou et al. (2003), one with the β2 subunit at the N terminus (abbreviated β/α) and the other with α4 (α/β). We expressed each concatemer in oocytes, both alone and with free α4 or β2 subunits (Table 1). We confirmed that concatemers expressed alone produce functional receptors. However, in the presence of a free subunit, the resulting receptors appear to reflect the properties of receptors composed of two concatemers plus a single free subunit (Zhou et al., 2003). Expression with a free β2 subunit results in a population of receptors with a small EC50 for activation by ACh, whereas free α4 subunit results in receptors with a large EC50 (Table 1). The values are similar to those for free subunits expressed at a low α4:β2 ratio or a high α4:β2 ratio, respectively (Table 1) (Zwart and Vijverberg, 1998; Moroni et al., 2006). We also examined gating by the subtype-selective agonist, 5I A85380. This agonist has different EC50 values for activation of the two forms: ∼10 nm for (α4)2(β3)3 versus ∼18,000 nm for (α4)3(β3)2 (Zwart et al., 2006). In addition, the maximal response is actually greater than the maximal response to ACh for the (α4)2(β3)3 receptor (Zwart et al., 2006). We determined the response to 1 μm 5I A85380 relative to the response to 1 mm ACh. 5I A85380 activates receptors incorporating free β2 (resulting in receptors containing three copies of β2) much more strongly than those with free α4, as expected from studies with free subunits (Table 1). Overall, these results indicate that the concatemers assemble with free subunits to generate pentameric receptors with properties appropriate for the stoichiometry predicted for incorporation of the free subunit, as reported previously (Zhou et al., 2003).
The fact that we produce surface receptors by combining subunit concatemers with mutated free subunits raises the possibility that some combinations assembled inappropriately (for example excluding the free subunit). We cannot rule this out, but for most combinations we determined the EC50 for activation by ACh and the relative gating by 1 μm 5I A85380 compared with 1 mm ACh (supplemental Table 1, available at www.jneurosci.org as supplemental material). As mentioned previously, these values provide an indication of the numbers of α4 and β2 subunits present in the functional receptors. Our unpublished data indicate that the N-terminal extracellular domain is the primary determinant for both of these parameters; that is, when three β2 extracellular domains are present in the receptor, the EC50 will be small (10 μm or less) and the relative response to 5I A85380 will be large (1.0 or greater), in contrast to the case when only two are present. Figure 2 shows a plot of relative response against log(EC50) and shows that the data fall into two distinct groups. The groups indicate that the functional receptors contain the number of β2 extracellular domains expected if the free subunits incorporated into the pentameric receptor. As shown in Results, we also found that a single free α4 subunit confers potentiation by 17β-estradiol, indicating that it also incorporates into the assembled pentameric receptors. Accordingly, the constructs we have used behave in the expected way, and the majority of pentameric receptors include both concatemers and free subunits. We cannot rule out the possibility that there is a small fraction of receptors that did not incorporate free subunits.
We note that our data indicate that both the α/β and β/α concatemers when expressed without free subunits behave as though receptors include three α4 subunits. This differs from the results obtained by Zhou et al. (2003), who observed that the β/α receptor had a small EC50 for ACh.
We also confirmed by Western blot that concatemers do not significantly degrade into individual subunits in the oocytes (Fig. 3) (Zhou et al., 2003; Carbone et al., 2009).
We sought to incorporate the free subunit selectively into a defined position in the pentamer, either a position in which it contributed to an agonist-binding site or one in which it did not. Accordingly, we confirmed the arrangements proposed by Zhou et al. (2003) for subunits in receptors formed by a concatemer plus a free subunit (Fig. 1). To do this, we used a mutation in the E-loop of the agonist-binding site contributed by the β2 subunit, which we had previously shown to affect activation by 5I A85380 (β2F119Q) (Hamouda et al., 2009). We expressed β2F119Q with each concatemer and determined activation by 5I A85380 and ACh. As shown in Figure 4, when β2F119Q is expressed with the β/α concatemer, activation is indistinguishable from when wild-type β2 is expressed. In contrast, when expressed with the α/β concatemer, activation by 5I A85380 is shifted toward higher concentrations and lower efficacy. The EC50 for activation by ACh is not affected [5 ± 1 μm (five cells) for α/β&β2F119Q and 9 ± 3 μm (seven cells) for β/α&β2F119Q], as expected from results with free subunits (Hamouda et al., 2009). These observations confirm the proposed arrangement (Zhou et al., 2003), that the free subunit contributes to an agonist-binding interface when expressed with the α/β concatemer but not the β/α concatemer (Fig. 1).
Results
Previous studies of the ability of 17β-estradiol to potentiate the nicotinic α4β2 receptor have found that potentiation requires a specific sequence at the C-terminal of the α4 subunit (Paradiso et al., 2001; Curtis et al., 2002). Mutagenesis of the sequence (Paradiso et al., 2001) demonstrated that not only a specific set of residues is required but that the position of the sequence with respect to the final (fourth) transmembrane helix is critical: insertion or deletion of a single residue to move the domain further from or closer to the external end of the helix abolishes potentiation. Finally, addition of even a single residue at the end of the domain also abolishes potentiation by 17β-estradiol.
In light of these observations, our initial hypothesis was that potentiation requires that the C-terminal domain must be placed on an α4 subunit and, additionally, that the α4 subunit must participate in ACh binding. To test this hypothesis, we need to control the number and position of mutated subunits in the assembled receptor. The use of concatemeric constructs of subunits (Zhou et al., 2003; Carbone et al., 2009) allows this control. We generated two concatemers, one with the α4 subunit at the N terminus (referred to as α/β) and the other with the β2 subunit at the N terminus (β/α) using the approach developed by Zhou et al. (2003) (see Materials and Methods). Our characterization of the concatemers demonstrates that the functional receptors generated when a concatemer is expressed with a free subunit show the properties of a receptor that includes two copies of the concatemer with one copy of the free subunit (see Materials and Methods). Furthermore, the use of a point mutant in the β2 subunit confirmed the subunit positions defined by Zhou et al. (2003) (see Materials and Methods) (Fig. 1). When a free subunit is expressed with the β/α concatemer, it will occupy the structural (non-agonist-binding) position in the receptor, whereas with the α/β subunit it will occupy an agonist-binding position (Fig. 1).
The concentration of ACh used to elicit responses and to test potentiation was relatively low (0.1 to 1 μm) and, for almost all constructs, elicited <20% of the response to 1 mm ACh (supplemental Table 2, available at www.jneurosci.org as supplemental material). We used 10 μm 17β-estradiol as the standard test concentration of potentiator.
Sample traces of potentiation for the most important combinations of constructs are shown in Figure 5. The data for the potentiation ratios for all combinations studied are shown in supplemental Table 2 (available at www.jneurosci.org as supplemental material).
17β-Estradiol potentiation is conferred whether the α4 subunit participates in ACh binding or serves as the structural subunit
We confirmed that the α/β concatemer, expressed alone, is not potentiated by 10 μm 17β-estradiol, whereas the β/α concatemer is (Zhou et al., 2003) (Table 1). This indicates that potentiation requires one or more intact, untethered C-terminal domains on the α4 subunit. When the α/β concatemer is expressed with free β2 subunit (abbreviated as α/β&β2), potentiation is also absent, whereas coexpression of the α/β concatemer with free α4 subunit results in potentiation (Fig. 5, Table 1). This observation indicates that a single intact α4 C terminus is sufficient to allow some potentiation. Overall, comparing β/α&α4 to α4&β2 (8:1) indicates that potentiation for the β/α&α4 receptor is greater than for receptor composed of free subunits (p = 0.04) as well as for β/α&β2 compared with α4&β2 (1:8) receptors (p = 0.01).
Coexpression of the α/β concatemer with free α4 subunit results in a receptor in which the α4 subunit that has an intact C terminus also participates in forming an ACh-binding interface (Fig. 1). To determine whether this is required, we tested additional constructs.
The first manipulation tested the effects of a mutation of the α4 subunit C terminus. Potentiation by 17β-estradiol is lost when the α4 C-terminal WLAGMI is mutated to WLAAC (abbreviated αWLAAC) (Paradiso et al., 2001). Expression of αWLAAC with α/β removes potentiation (Fig. 6), as expected. We then mutated the untethered C terminus of the α4 subunit in the β/α concatemer (β/αWLAAC). When the β/αWLAAC concatemer is expressed with free α4, potentiation is present, but not when it is expressed with free β2 (Fig. 6). These observations indicate that potentiation can occur whether the α4 subunit contributes to an agonist-binding interface (when expressed with the α/β concatemer) or acts as a structural subunit (when expressed with the β/αWLAAC concatemer). They also confirm that only a single subunit need contain the WLAGMI sequence to underlie potentiation.
Incorporation of a single α4 subunit with an untethered WLAGMI domain confers estradiol potentiation on the receptor. We compared two combinations of constructs. In the case of α/β&α4 to α/β&α4WLAAC, the difference in potentiation ratio is significant at p = 5 × 10−8 (t test). For β/αWLAAC&α4 to β/αWLAAC&α4WLAAC, the difference is significant at p = 8 × 10−7. These observations support the conclusion that free α4 subunits incorporate efficiently when expressed with either concatemer.
A 17β-estradiol binding element can be placed on either α4 or β2 subunits
We then examined the question of whether potentiation required the rest of the α4 subunit or was based on the C-terminal region alone. The initial constructs were chimeric subunits between α4 and β2 subunits, with a join just after the end of the third membrane spanning region (abbreviated α-M3-β and β-M3-α) (Fig. 1). In these chimeras, the N-terminal extracellular domain and the first three transmembrane domains are from one subunit, whereas the large cytoplasmic loop, the fourth transmembrane domain and the C-terminal tail are transferred. As expected, replacing the C-terminal domain of the α4 subunit with sequence from β2 removes potentiation when expressed with the α/β concatemer (Figs. 5, 6; supplemental Table 2, available at www.jneurosci.org as supplemental material). More surprisingly, transferring the α4 sequence to β2 confers potentiation (Fig. 6). This observation indicates that the N-terminal extracellular domain and the first three transmembrane domains are not sufficient for potentiation. Potentiation is transferred whether the free subunit occupies the structural position or contributes to an agonist-binding interface (Fig. 6).
We constructed two additional chimeric subunits. These involved transferring the fourth transmembrane segment plus the WLAGMI sequence (βM4→C) from the α4 subunit to the β2 subunit, or only the WLAGMI sequence (βWLAGMI). The 6 aa sequence was transferred, as previous work has indicated that the length of the C-terminal sequence is critical for potentiation (Paradiso et al., 2001). Both of these chimeras allow potentiation when expressed with the β/αWLAAC or the α/β concatemer (Fig. 6). These data indicate that the terminal residues are critical for potentiation. However, the amount of potentiation is significantly greater for the βM4→C construct than for α4 or βWLAGMI when expressed with either the α/β or β/αWLAAC concatemers, and for β-M3-α expressed with α/β (one-way ANOVA for each concatemer separately, with Bonferroni's correction). We note, however, that mutation of WLAGMI to WLAAC in the intact α4 subunit removes potentiation.
We noted that transferring the WLAGMI sequence to β2 resulted in the C-terminal sequence QPWLAGMI. Previous work (Paradiso et al., 2001) showed that mutation of both prolines in α4 (PPWLAGMI to AAWLAGMI) abolishes potentiation, so we mutated the glutamine to proline in β2WLAGMI to produce β2PPWLAGMI. Expression of the resulting construct did not increase potentiation to the same level as seen with the βM4→C mutation (Fig. 6).
Overall, these data indicate two points. First, transfer of the TM4 region plus the tail produces more potentiation than the simple transfer of the C-terminal tail. Second, the cytoplasmic loop of the α4 subunit (from the end of TM3 to the start of TM4) appears to diminish potentiation compared with the amount produced when only the TM4 to the C terminus is transferred.
The observation that the WLAGMI domain can be moved from one subunit to another is quite surprising. Our data suggest that potentiation does not differ greatly depending on whether the tail is placed on the α4 or β2 subunit. Pooling the data for receptors containing single copies of WLAGMI, the mean values for potentiation are, for β2 N-terminal constructs, 2.3 ± 0.3-fold (mean ± SE; N = 8 combinations of constructs), and, for α4 N-terminal constructs, 1.6 ± 0.1-fold (N = 2) (p = 0.03). Similarly, it does not appear to matter whether the domain is placed on the structural subunit (mean potentiation, 2.4 ± 0.5-fold; four combinations) or on a subunit contributing to an ACh-binding interface (2.0 ± 0.2-fold; six combinations) (p = 0.8). These conclusions are tentative, given the relatively small number of cases examined and the possibility for confounding factors.
Our observations indicate that potentiation can occur when only a single copy of a free WLAGMI domain is incorporated in the assembled receptor. Furthermore, the WLAGMI sequence can be placed on the α4 or β2 subunit, and the subunit can participate in forming an agonist-binding interface or serve as the fifth, structural, subunit. There are trends suggesting that potentiation may be larger when the receptor includes concatemers, and may be larger when the domain is present on a β2 subunit rather than an α4 subunit. There is also an indication that some portions of the α4 cytoplasmic loop (between the third and fourth transmembrane regions) might reduce potentiation, whereas some portions of the fourth transmembrane region may enhance potentiation. These observations suggest the existence of additional factors that influence the extent of potentiation, which will have to be examined in additional experiments.
Relationship between copy number and potentiation
There is a significant increase in potentiation as the total number of WLAGMI C-terminal domains increases in a pentameric receptor. The average value for potentiation by 10 μm 17β-estradiol increases from 0.97 ± 0.02 (0 domains, 10 combinations tested) to 2.2 ± 0.2 (1 domain, 10), 2.4 ± 0.4 (2 domains, 5), 3.3 ± 0.2 (3 domains, 4), to 3.7 ± 0.3 (5 domains, 3). Regression of potentiation on the number of untethered domains gives a slope of 0.6 (p < 10−7 that the slope is zero).
Our data were obtained using a constant concentration of 17β-estradiol, and it is possible that changes in both potency and efficacy occurred for some of the constructs. Accordingly, we determined the concentration–effect relationship for 17β-estradiol for combinations of subunits that have one to five untethered WLAGMI sequences. Two combinations of constructs were tested for each number of untethered WLAGMI. The combinations were chosen to keep the number of β2 N termini constant at 2 in the assembled pentamers, but some combinations were formed without the use of concatemers and the domains were placed on agonist-binding or structural subunits. The concentration–potentiation data are shown in Figure 7, and the fit parameters are given in Table 2.
The fit maximal potentiation is plotted against the number of untethered WLAGMI domains in Figure 8. There is a clear increase in maximal potentiation with increasing untethered WLAGMI. The increase is greater than linear, with each added untethered WLAGMI increasing the potentiation by ∼1.6-fold. The geometric fit shown in Figure 8 is a better description than the linear fit, although there is an indication that the increase from three to five domains is not as great as might be expected. This could result from several possible factors. Some are technical, for example as a result of particular constructs providing a somewhat greater or lesser amount of potentiation than others. However, it is also possible that there is a “ceiling” on the amount of potentiation. This could arise because there is a maximal potentiation possible by the mechanism used by estradiol (so that five domains occupied by 17β-estradiol would not be 1.62 as efficacious as three), or it could be that there is some interaction among bound estradiols so that there is a maximal possible number that can bind. Additional experiments will be necessary to distinguish among these possibilities.
Our test concentration of 17β-estradiol (10 μm) is not a saturating concentration for any of these constructs. Accordingly, the potentiation ratios we calculate for the screening data may reflect both efficacy and potency. However, the qualitative conclusions about placement of the untethered WLAGMI domain are clear.
Potentiation by 17α-vinylestradiol
The steroid analog 17α-vinylestradiol also potentiates the α4β2 receptor. Previous work (Paradiso et al., 2001) has found that potentiation by this analog does not require the final four residues of the α4 C terminus, although potentiation is larger when the AGMI sequence is present. To remove potentiation, it is necessary to also mutate a tryptophan residue just preceding the terminal sequence (i.e., to convert WLAGMI to LLAAC). Mutation only of the tryptophan (to LLAGMI) does not reduce potentiation by either 17α-vinylestradiol or 17β-estradiol, emphasizing the overall greater importance of the AGMI sequence for potentiation (Paradiso et al., 2001). Assuming that the transduction mechanism is the same for potentiation by 17α-vinylestradiol or 17β-estradiol, these observations indicate that neither the AGMI sequence nor the critical tryptophan is required for the transduction of molecular recognition into functional potentiation. Accordingly, this steroid analog provides a control to indicate that the transduction mechanism for potentiation is preserved when potentiation by 17β-estradiol is removed.
We tested a number of the combinations described above (a total of 17) (supplemental Table 3, available at www.jneurosci.org as supplemental material) and found that 17α-vinylestradiol potentiates all of them. For the five cases in which there are no untethered AGMI and two or three critical tryptophans, the mean potentiation by 17α-vinylestradiol is 2.3 ± 0.6, whereas 17β-estradiol on average has no effect (0.94 ± 0.04). Overall, potentiation increases with increasing numbers of critical tryptophans (associated with increasing numbers of untethered AGMIs) from 3.1 ± 1.5 (N = 2) for two tryptophans, to 6.1 ± 1.7 (N = 11) for three tryptophans, and 7.7 ± 2.7 (N = 4) for five tryptophans, although the regression coefficient (1.2 ± 1.3; fit value ± estimated SE of parameter) is not significantly different from 0 (p = 0.39).
We tested a subset of additional constructs in which the critical tryptophan was mutated. These were αLLAAC and β/αLLAAC. These constructs allowed us to test the effect of increasing numbers of critical tryptophans in the absence of an untethered AGMI sequence, again holding the number of subunits with β2 N termini constant at 2. The mean potentiation increased steadily with increasing numbers of tryptophans: β/αLLAAC&αLLAAC (zero tryptophans: 1.16 ± 0.03, five oocytes); β/αLLAAC&αWLAAC (one: 1.20 ± 0.09, four); β/αWLAAC&αLLAAC (two: 1.56 ± 0.15, four); β/αWLAAC&αWLAAC (three: 1.76 ± 0.05, four). Linear regression of potentiation on the number of critical tryptophans for this subset gives a regression coefficient of 0.22 ± 0.04, which differs from zero (p = 0.04).
Overall, the most significant observation is that removal of potentiation by 17β-estradiol does not remove potentiation by 17α-vinylestradiol, indicating that the transduction mechanism is retained.
Discussion
The goal of this study was to identify the critical portions of the nicotinic α4 and β2 subunits required for potentiation by 17β-estradiol. We extended previous studies that had determined that the C-terminal tail of the α4 subunit is necessary (Paradiso et al., 2001; Curtis et al., 2002). Our starting hypothesis was that potentiation would be subunit specific and likely would require that the specific sequence be present on a subunit in a particular position (e.g., on an α4 subunit involved in forming an agonist-binding site). To our surprise, this hypothesis is incorrect in both respects. The results indicate that the WLAGMI domain can be placed at the C terminus of either the α4 or β2 subunit to subserve potentiation. In addition, the subunit can participate in forming an agonist-binding site or serve as the fifth, structural, subunit in the receptor. As a corollary to these observations, the subunit with the domain can be placed between β2 and α4 subunits or α4 and β2 subunits in the assembled receptor, so potentiation does not appear to require a particular neighbor subunit.
Several observations support the idea that 17β-estradiol interacts with the receptor at the AGMI sequence. Because 17β-estradiol is a hydrophobic molecule, it can interact with the lipid membrane. However, the enantiomer of 17β-estradiol does not potentiate (Paradiso et al., 2001). This enantioselectivity indicates that potentiation requires interaction of steroid with an optically active site, perhaps on the receptor. Potentiation is also extremely sensitive to the structure of the WLAGMI domain—mutations, insertions, or deletions can greatly reduce potentiation. In addition, potentiation by 17α-vinylestradiol is reduced but not removed by mutation of AGMI (Paradiso et al., 2001; this study). To remove potentiation by 17α-vinylestradiol, it is necessary to mutate both the AGMI sequence and the neighboring tryptophan. This observation indicates that the AGMI sequence is not required for transduction, in that 17α-vinylestradiol still is capable of potentiation. Furthermore, the finding that complementary changes in steroid and receptor structures affect potentiation supports the idea that the D ring of the steroid associates with the C-terminal WLAGMI tail to underlie potentiation. Finally, the present observations indicate that moving this defined domain from one subunit to another can transfer potentiation. Overall, these data support the conclusion that the C-terminal domain is involved in molecular recognition (binding) rather than the conversion of binding at a different site into functional potentiation (transduction). However, previous work from our laboratory (Paradiso et al., 2001) has shown that the molecular structure of the A ring (the “other end” of the steroid molecule) also is important for potentiation. The binding domain for the A ring has not been localized, although the present data suggest that it must be either to a sequence that is found in both the α4 and β2 subunits, or possibly lies in the membrane. Accordingly, it seems most appropriate to call the C-terminal sequence a binding element or binding domain.
It is surprising that potentiation can be transferred between subunits simply by moving the WLAGMI sequence. Previous studies of potentiating drugs show more specificity in sites. Several potentiators interact with receptors in this gene family at subunit interfaces in the extracellular region. The classic example is benzodiazepine potentiation of GABAA receptor function (Sigel and Buhr, 1997), but potentiation of nicotinic α3β4* receptors by morantel (Seo et al., 2009) or α43β22 receptors by Zn2+ ions (Moroni et al., 2008) also requires specific residues in the two subunits forming an interface. For more hydrophobic compounds, etomidate appears to interact with transmembrane regions of two subunits in the GABAA receptor to potentiate (Li et al., 2006), whereas potentiating neurosteroids are proposed to interact with two transmembrane domains of a single GABAA subunit (Hosie et al., 2006). This estradiol-binding element appears to be remarkably discrete and effective at transferring potentiation.
There are additional parts of the subunit(s) that appear to influence the amount of potentiation (Curtis et al., 2002). The cytoplasmic loop of the α4 subunit may reduce the amount of potentiation transferred. This effect might result from an action of the loop on transduction of potentiation, as there have been reports in receptors in this family that the cytoplasmic loop can affect channel function (Bouzat et al., 1994; Wang et al., 2000; Kuo et al., 2005; Hales et al., 2006) or modulation (Swope et al., 1999; Yevenes et al., 2008). In contrast, the TM4 helix of the α4 subunit appears to increase potentiation. As will be discussed below, the action of 17β-estradiol to potentiate responses may result from an effect on the transmembrane helices to stabilize the open-channel state of the receptor. Accordingly, the structure of the TM4 region may influence the transduction mechanism. Alternatively, it might be that the A ring of the steroid interacts with some of the residues in the TM4 region. Additional experiments will be required to elucidate the bases for these effects.
The mechanism by which potentiation is produced is not known. Single-channel studies have shown that there is no increase in the single-channel conductance (Curtis et al., 2002). However, the probability of being open is increased, perhaps because of an increase in the duration of openings [Curtis et al. (2002), their Fig. 4]. The increased probability of being open is reminiscent of the effects of mutations of the conserved leucine at the ninth residue in the second membrane-spanning region. Numerous studies have reported that mutation of TM2 L9′ to more hydrophilic residues increases the open probability of nicotinic and related receptors (Revah et al., 1991; Labarca et al., 1995; Chang et al., 1996). The increase is produced by a mutation in any of the five subunits in the receptor, and the overall effect increases with number of mutated subunits (Labarca et al., 1995; Chang et al., 1996; Moroni et al., 2006). These observations indicate that all five subunits contribute to a conformational change before the channel becomes permeable for ions and that the L9′ mutation in any subunit therefore can shift the overall gating equilibrium. Perhaps the interaction with estradiol produces a conformational change in the transmembrane regions of that subunit that results in a similar stabilization of the open state. The observation that the extent of potentiation (efficacy) increases with increased numbers of untethered WLAGMI regions is consistent with this suggestion. The data suggest that each untethered WLAGMI increases the maximal potentiation by ∼1.6-fold. The idea of an independent action of 17β-estradiol on any subunit that carries an available WLAGMI sequence accounts well for the essential features of our observations.
Drugs that enhance the response of synaptic receptors, without directly interacting with the agonist-binding site, are of increasing interest as therapeutic agents that do not produce a response directly but enhance endogenous signaling. A number of drugs that enhance GABAA receptor activity are in clinical use as tranquilizers, sedatives, or hypnotics, first represented by the benzodiazepines (Rudolph and Möhler, 2004). More recently, potentiators of the nicotinic α7 receptor have received attention as possible agents to enhance cognition and memory (Lightfoot et al., 2008). The nicotinic α4 subunit is quite prevalent in the mammalian brain; the receptor comprising α4 and β2 subunits is the most common heteromeric receptor and the α4 subunit also participates in forming a variety of receptors of more complex stoichiometry (Gotti et al., 2007). These receptors have their major physiological effects by modulating the release of other neurotransmitters, rather than directly mediating postsynaptic responses in the brain (Dani and Bertrand, 2007). In particular, a role for α4-containing receptors in control of dopamine release has been proposed (Exley and Cragg, 2008), providing a possible link to the reward pathway. The actions of endogenous compounds, particularly steroids, are more difficult to define compared with the effects of exogenously added drugs. There are two principal difficulties in demonstrating a physiological role for estradiol potentiation specifically of α4-containing nicotinic receptors. The first is that the concentration of 17β-estradiol required for potentiation is high (>1 μm), much higher than the levels in the brain (∼1 nm) (Mukai et al., 2006), although local synthesis clearly occurs (Mukai et al., 2006; Cornil and Charlier, 2010) and could result in higher local levels. The second is that 17β-estradiol has a multiplicity of effects in the nervous system, both in sculpting development (cf. Gillies and McArthur, 2010) and in more rapid changes in function (cf. Mukai et al., 2006; Cornil and Charlier, 2010; Gillies and McArthur, 2010).
Overall, these observations indicate that the interaction between 17β-estradiol and the nicotinic α4β2 receptor is mostly determined by the discrete, C-terminal tail of a subunit. The ability to transfer potentiation between subunits and the relationship between numbers of WLAGMI domains and efficacy of potentiation suggest that the effect of 17β-estradiol is mediated by actions on single subunits and that the overall consequences for gating occur because of the summation of independent energetic contributions to overall gating of this receptor.
Footnotes
This work supported by National Institutes of Health Grants NS22356 and GM47969. J.H.S. is the Russell and Mary Shelden Professor of Anesthesiology. We thank Gustav Akk, John Bracamontes, and Chris Lingle for helpful comments and criticism. We thank Henry Lester for encouragement.
- Correspondence should be addressed to Joe Henry Steinbach, Department of Anesthesiology, Washington University School of Medicine, 660 South Euclid Avenue, Saint Louis, MO 63110. jhs{at}morpheus.wustl.edu