The GABA_C receptor was first characterized as a bicuculline-insensitive, baclofen-insensitive GABA-activated conductance (Johnston, 1996). Molecular studies identified three subunits, ρ1–ρ3, which combine to form pentameric receptors that have properties similar to GABA_C receptors in native cells (Enz and Cutting, 1998). Although the GABA_C and GABA_A receptors show molecular similarities, they are quite distinct in terms of their cellular localizations, probably physiological roles, and pharmacological properties (Johnston, 1996; Bormann, 2000; Johnston et al., 2003; Lukasiewicz et al., 2004). Neuroactive steroids are known to be the most potent and efficacious endogenous modulators of the GABA_A receptor (Belelli et al., 2006), but their activities on GABA_C receptors have been less well characterized. Initial studies of receptors expressed in Xenopus oocytes after injections of retinal mRNA found that the bicuculline-insensitive responses to GABA were insensitive to both potentiating and inhibiting neurosteroids (Woodward et al., 1992). However, a subsequent study found that the apparent insensitivity resulted from the fact that relatively high concentrations of GABA were used to elicit responses; when low concentrations of GABA were used, both potentiation and inhibition could be shown (Morris et al., 1999). However, the effects required higher steroid concentrations than those necessary for modulation of the GABA_A receptor. We have undertaken a further study of the actions of steroids on the ρ1 GABA_C receptor, to examine a broader range of endogenous steroids and to better define the structure-activity requirements for steroid actions.

There were two primary motivations for these studies. The first motivation was to determine whether some endogenous steroids have higher potency or efficacy than those examined to date. Because steroids are so effective in actions on the GABA_A receptor, it seemed possible that a particular steroid might demonstrate a physiologically relevant selectivity for the GABA_C receptor. The GABA_C receptor plays a less well defined role in the function of the nervous system (Zhang et al., 2001). It has a clear role in the retina (Lukasiewicz et al., 2004), and studies of genetically modified mice suggest that it may play a role in processing of nociception in the spinal cord (Zheng et al., 2003). Activation of GABA_C receptors by synaptically released GABA has been shown to occur in the rat hippocampus (Alakuijala et al., 2006), and a role for GABA_C receptors in short-term memory has been demonstrated in young chickens (Gibbs and Johnston, 2005). Furthermore, there is accumulating evidence that ρ subunits may coassemble with GABA_A receptor subunits both in vitro (Pan et al., 2000; Pan and Qian, 2005) and in vivo (Milligan et al., 2004). There is also increasing interest in developing drugs targeting the GABA_C receptor (Johnston et al., 2003). Hence, identifying possible endogenous modulators of receptors containing the ρ1 subunit is of general interest.

The second motivation was to obtain additional information for comparisons of steroid actions on the GABA_A and GABA_C receptor. The GABA_A receptors are the major type of receptors that underlie the physiologically and clinically relevant effects of neurosteroids, and they are the most extensively studied type (Belelli et al., 2006). The GABA_C receptor has been less studied, but it shows both similarities to and differences from the GABA_A receptor. In particular, for pairs of steroids such as pregnanolone and allopregnanolone (which differ only in terms of the stereochemistry at the C5-position), both members potentiate GABA_A receptors, whereas the 5β steroid (pregnanolone) inhibits and the 5α steroid (allopregnanolone) potentiates GABA_C receptors (Morris et al., 1999). Because the GABA_A and GABA_C receptors show sequence similarities and hence may have some similarity in steroid binding sites, it is of interest to explore the structure-activity relationship for potentiation and inhibition of the GABA_C receptor by steroids.

Materials and Methods

All experiments were conducted using human ρ1 GABA_C receptors expressed in Xenopus laevis oocytes, as described in Li et al. (2006b). In brief, a full-length construct of the human ρ1 subunit was provided by D. Weiss (University of Texas Health Center, San Antonio, TX) and transferred to pcDNA3 (Invitrogen, Carlsbad, CA). The construct was sequenced, and it was found to correspond to the published sequence (GenBank accession no. M62400). cRNA was synthesized using mMessage mMachine T7 kit (Ambion, Austin, TX). The cRNA was dissolved in RNAase-free water, and 8 ng in a volume of 23 nl was injected into each oocyte. Oocytes were incubated for 40 to 60 h at 18°C before being studied electrophysiologically.

A large number of steroids and analogs were tested (Table 1). Commercially available steroids were purchased from Steraloids (Newport, RI). Other steroids were synthesized by published methods (Table 1). Stock solutions of steroids were made at a concentration of 10 mM in DMSO, and then they were diluted to the final concentration in bath solution containing 200 nM GABA. At the highest concentration of DMSO used (0.2%), DMSO had no effect on responses (also see Morris et al., 1999; Li et al., 2006b).

Responses were recorded 2 to 5 days after injecting the oocytes, at a holding potential of –50 mV (unless otherwise noted) using a two-electrode oocyte clamp (Warner Instruments, Hamden, CT). Both voltage and current electrodes were patch-clamp electrodes filled with 3 M KCl, and they had resistances of 0.5 to 1 MOhm. The bath solution contained 96 mM NaCl, 2 mM KCl, 1 mM MgCl₂, 1.8 mM CaCl₂, and 10 mM HEPES, pH adjusted to 7.5 with NaOH. The bath had a volume of approximately 0.1 ml, and it was perfused at approximately 7 ml/min. Bath solution was perfused between all test applications. Solutions were switched by hand, using Teflon rotary valves (Rheodyne, Rohnert Park, CA). Solutions were applied using glass reservoirs, fluorocarbon valves, and fluorocarbon or metal tubing, to reduce adsorption. Most studies were performed using 200 nM GABA, a concentration producing a response approximately 5% of the maximal response (Li et al., 2006b). In each case, GABA alone was applied, then GABA + steroid, then GABA alone, and the effect was calculated by taking the ratio of the response to GABA + steroid to the mean response for the bracketing control applications. Applications were separated by 3 to 5 min. In studies of mutated receptors, an initial concentration-response curve was obtained for GABA, and the concentration that produced approximately 5% of maximal response was used (see Results).

Data were acquired using a Labmaster analog-to-digital converter (Molecular Devices, Sunnyvale, CA). Digitized traces were analyzed using pClamp software (Molecular Devices), and reduced data were analyzed with Excel (Microsoft, Redmond, WA) and SigmaPlot (Systat Software, Inc., San Jose, CA). Data values are presented in the text and shown in the figures as mean ± S.D. (N cells). The significance of effect for a given compound was assessed using a two-tailed Student's t test for paired observations for a difference to a ratio of 1. To compare effects of multiple drugs a one-way analysis of variance with a Dunnett's post hoc correction (SYSTAT; Systat Software, Inc.) was used to compare effect ratios.

GABA concentration-response relationships were obtained by interspersing responses to a control concentration (1 μM) and normalizing responses for that egg to the control. The relationships were analyzed using the following equation: relative , where maximum is the maximal relative response, EC₅₀ is the concentration producing half-maximal response, and n is the Hill coefficient.

Inhibition curves were analyzed using the following equation: relative , where Y0 is the relative response at a saturating drug concentration, IC₅₀ is the concentration producing half-maximal inhibition, and n is the Hill coefficient. In both cases, the equations were fit to all the individual data points pooled from all eggs with that drug.

Results

Structure-Activity Relationship for Steroids and Analogs

The structure-activity relationships were evaluated using a drug concentration of 10 μM, for effects on the response to 200 nM GABA. This concentration of GABA is about the EC₅ concentration (Li et al., 2006b), and it was chosen because it has been shown that both potentiation and inhibition by steroids are maximal at a low concentration of agonist (see below; Morris et al., 1999). We also note that it has been proposed that GABA_C receptors produce a relatively tonic response to low concentrations of GABA; thus, a low concentration may be most appropriate for exploring physiologically relevant actions. A single concentration of steroid was adopted for comparisons among multiple drugs. However, it means that a difference in effect could be produced by a change in potency, efficacy, or both.

We use a standard set of abbreviations for the steroids used (full names and abbreviations are given in Table 1). For example, the pair allopregnanolone and pregnanolone are abbreviated as 3α5α17βP and 3α5β17βP (e.g., the 3-hydroxyl group is in the α orientation; the 5-hydrogen is α or β, respectively; and the 17-methylketone is β).

The overall results for 41 compounds are summarized in Fig. 1 (with representative structures), and selected comparisons are presented below. Structures for all steroids and analogs are provided in the Supplemental Material.

Comparison of 5α- and 5β-Reduced Compounds. We compared pairs of drugs with identical structures except for the reduction at the C5-position (Fig. 2). In general, if drugs in a pair had activity, the 5α-reduced compound potentiated and the 5β-reduced compound inhibited. These results are consistent with previous studies (Morris et al., 1999; Goutman and Calvo, 2004). The major exception is the pair of sulfated steroids 3αSO₄5α17βP and 3αSO₄5β17βP, for which these C5 diastereomers inhibited equally. This observation was surprising enough that we confirmed it with new supplies of both steroids, and we also performed a full concentration-effect study (see below). It suggests that inhibition by the sulfated steroids may differ from inhibition by unsulfated steroids.

Comparison of 3α- and 3β-Reduced Compounds. We also compared pairs of drugs that differed only in the orientation of the substituent at the 3-position (Fig. 3). In most cases, including sulfated steroids, a compound with a 3β substituent showed greatly reduced ability to potentiate or block. This probably explains why pregnenolone sulfate (3βSO₄5Δ17βP) is a weak inhibitor of responses of the ρ1 GABA_C receptor (Woodward et al., 1992; Fig. 1). Indeed, 3β5Δ17βP itself is inactive, whereas 3α5Δ17βP is a potentiator (Fig. 3). 3βSO₄5Δ17K also is only weakly active (Fig. 1). One possible exception is 3β5β17βP (Fig. 3), which showed significant (although relatively weak) potentiation.

Comparison of 17α- and 17β-Reduced Compounds. Steroid actions on GABA_A receptors show a strong dependence on the orientation of the substituent at the 17-position, with 17α compounds showing very weak activity (Phillipps, 1975). This requirement for activity on ρ1 receptors was tested using four compounds based on 3α5α17βCN. 3α5α17βCN potentiated [2.00 ± 0.20 (6)], whereas 3α5α17αCN did not [1.01 ± 0.13 (5); P < 0.001 for the significance of difference between the diastereomers]. Like-wise, 3α5β17βCN inhibited [0.71 ± 0.09 (7)], whereas 3α5β17αCN did not [0.97 ± 0.03 (5); P < 0.001]. We also examined four androstanediols, which were all essentially inactive on the ρ1 receptor under the conditions tested (Fig. 1). However, 17α-Est was equally effective at producing inhibition as 17β-Est [17β-Est: 0.44 ± 0.12 (11); and 17α-Est: 0.40 ± 0.07 (7)]. This observation suggests that estradiols may have a different site from that of the 5β-reduced steroids. In general, these results support the idea that the sites mediating either potentiating (3α5α) or inhibiting (3α5β) effects require the same orientation for the substituent at C17.

Structure of the Substituent at the 17β-Position. We examined a number of drugs that were identical at the 3- and 5-positions but that differed in terms of the group attached at the 17β position, to determine whether the sites mediating potentiation and inhibition could be distinguished based on this structural feature (Fig. 4). Overall, we found that the two types of effector sites showed similar structure-activity relationships, at the resolution afforded by our methods. That is, drugs with carbonitrile, methylketone, and hydroxymethylketone groups in the 17β-position all were able to produce potentiation or inhibition. Ketone and hydroxyl groups resulted in compounds that were essentially inactive (Fig. 1). These observations suggest similarities between the sites mediating potentiation by 5α-reduced steroids and inhibition by 5β-reduced steroids.

Inhibition by Estradiols. We have already noted that both 17β-Est and 17α-Est inhibit responses of the ρ1 GABA_C receptor. We also examined the requirement for a free 3-OH group, and we found that methylation (3MeO17β-Est) significantly reduces inhibition (Fig. 1). The ability of estrogens to potentiate responses of the human neuronal nicotinic α4β2 receptor shows similar structural requirements, including the lack of selectivity between 17α- and 17β-Est (Paradiso et al., 2001), so we tested an additional compound (17α-ethynyl17β-Est), which potentiates the nicotinic receptor. This drug showed reduced inhibitory effects on the ρ1 GABA_C receptor compared with 17β-Est (Fig. 1).

Sulfated and Carboxylated Steroids. Both 3αSO₄5α17βP and 3αSO₄5β17βP inhibit responses (Fig. 2), whereas 3βSO₄5α17βP and 3βSO₄5β17βP do not (Fig. 2). In addition, 3βSO₄5Δ17βP and 3βSO₄5Δ17K are only weak blockers (Fig. 1). We also tested the effects of adding a sulfate to either the 3- or 17-position of 17β-Est; both sulfated estradiols were significantly less active than 17β-Est (Fig. 1). This result suggests that the sites mediating inhibition by sulfated steroids and by estradiols are distinct.

To examine the role of a different anionic group at the 3-position, we tested a carboxylic acid derivative (3αCOOH5β17βP). This proved to be a very effective blocker (Fig. 2). Because the sulfated steroids inhibited whether the structure was 5α- or 5β-reduced, we also examined 3αCOOH5α17βP. This compound was inactive (Fig. 2). Accordingly, the carboxylated and the sulfated compounds have a different structure-activity relationship.

Comparisons of Enantiomer Pairs. Previous studies of the ρ1 GABA_C receptor have shown that the enantiomer of 3α5α17βP(ent-3α5α17βP) is inactive (rather than potentiating) and that ent-3α5β17βP actually potentiates rather than blocking currents (Li et al., 2006b; see Fig. 1). We also compared 17β-estradiol and ent-17β-estradiol; again, the unnatural enantiomer is inactive [17β-Est: 0.44 ± 0.12 (11); ent-17β-Est: 0.92 ± 0.10 (6)]. This suggests that there is a specific site involved, although it can, apparently, recognize both 17α- and 17β-estradiol. The enantiomer pairs 3βSO₄5Δ17βP and ent-3βSO₄5Δ17βP and 3βSO₄5Δ17K and ent-3βSO₄5Δ17K are only weakly active (Fig. 1), but they are consistent in the sense that if the natural enantiomer has a discernible effect, the unnatural enantiomer has a reduced, absent or reversed effect. Accordingly, it seems most likely that all of the effects are mediated by interactions between the steroid and the ρ1 GABA_C receptor (Li et al., 2006b).

Studies of the Mechanism of Steroid Action

Concentration Effect Studies. We first determined the consequences of testing a drug at 10 μM and eliciting responses with different concentrations of GABA (Fig. 5). The goal was to determine whether there is an indication that a drug acts preferentially on the resting or activated states of the receptor. The potentiation produced by 3α5α17βCN and the inhibition produced by 17β-Est or 3αSO₄5β17βP are maximal at a low concentration of GABA, and they are reduced at a high (saturating) concentration of GABA. We have also confirmed that the inhibition by 3α5β17βP is reduced at higher concentrations of GABA (Goutman and Calvo, 2004); 1 μM3α5β17βP reduced the response to 200 nM GABA to 0.65 ± 0.14 (N = 27), whereas it only reduced the response to 10 μM GABA to 0.96 ± 0.06 (N = 14). Similar observations have been made for potentiation by 3α5α17βPOH and inhibition by 3α5β17βPOH (Morris et al., 1999). In general, it is expected that there will be reduced potentiation when a high concentration of agonist is used, since many mechanisms for potentiation will be ineffective when a maximal response is elicited. In contrast, there is no reason to think that any of these inhibiting drugs act as competitive inhibitors of GABA binding. The fact that inhibition is reduced at higher [GABA] suggests that the inhibitors preferentially bind to resting states of the receptor.

We also examined the inhibition produced by various concentrations of inhibitors on responses elicited by 200 nM GABA. The goal was to determine whether inhibition was partial or complete, over the accessible range of steroid concentrations. As shown in Fig. 6, 3αCOOH5β17βP is the most potent inhibitory steroid. The diastereomers of estradiol are about equipotent with picrotoxinin and with each other. The sulfated steroids also are essentially equipotent with each other, although less potent than picrotoxinin. These compounds produced full inhibition. However, 3α5β17βP produced only a partial block (Fig. 6). Morris et al. (1999) have tested 3α5β17βP and other 5β-reduced steroids (including 3α5β17βPOH), and they also found that they did not produce full block but showed only partial block at maximal inhibition (also see Goutman and Calvo, 2004). These observations suggest that 3α5βP and other 5β-reduced steroids are allosteric inhibitors. These results also indicate that the screening protocol (a drug concentration of 10 μM and an EC₅ concentration of GABA) is appropriate for enhancing the magnitude of any effects.

Voltage Dependence of Inhibition. Charged inhibitors may show voltage dependence in the potency for block, if the charge on the drug interacts with the membrane potential field (for example, in open channel block; see Neher and Steinbach, 1978). For both 3αCOOH5β17βP and 3αSO₄5β17βP, there was no change in inhibition over a 100-mV range of potential. For 3αCOOH5β17βP, the relative response at –100 mV was 0.57 ± 0.03, at –50 mV was 0.63 ± 0.04, and at 0 mV was 0.54 ± 0.21 (no difference significant; three oocytes tested using 1 μM3αCOOH5β17βP). For 3αSO₄5β17βP, the values were 0.56 ± 0.03, 0.58 ± 0.08, and 0.48 ± 0.12, respectively (no difference significant; five oocytes tested using 5 μM3αSO₄5β17βP). Accordingly, it seems unlikely that the charge interacts with the membrane field in the mechanism producing inhibition.

Consequences of Mutations in the Second Membrane-Spanning Helix. We then explored the consequences of mutations in the ρ1 subunit. We selected two residues in the predicted second membrane-spanning region (TM2) for mutation, based on the likelihood that they would produce changes in the actions of inhibitory drugs. We evaluated consequences for the effects of 10 drugs, including three potentiating steroids, the corresponding 5β inhibitory steroids, and representative sulfated inhibitory steroids, carboxylated inhibitory steroids, and 17β-estradiol. The first mutation is to the residue in the second position of TM2, TM2 P2′S (ρ1 P294S numbered according to the first residue in the predicted mature subunit). Residues homologous to this have been identified as important for inhibition in several studies. Initially, it was found to be a critical residue for conferring resistance to dieldrin in an insect GABA receptor (rdl TM2 P2′S; Ffrench-Constant et al., 1993). Subsequently, it was identified as a determinant for block of glycine receptors by cyanotriphenylborate (Rundström et al., 1994) and lactones (glycine α3 TM2 G2′A, Steinbach et al., 2000). Finally, mutation of this residue in the α1 subunit of the GABA_A receptor reduces block by pregnenolone sulfate (α1 TM2 V2′S; Akk et al., 2001), by a benz[e]indene (Li et al., 2006a) and by 3β-hydroxysteroids and sulfated steroids (Wang et al., 2002). This mutation does not affect block by picrotoxinin of responses of ρ1 GABA_C receptors to low concentrations of GABA (Wang et al., 1995). However, we note that mutations to ρ1 TM2 P2′ do affect activation by agonists, including changes in the efficacy of partial agonists (Carland et al., 2004).

This mutation shifted the concentration response curve for GABA to higher concentrations (data not shown; EC₅₀ 3.4 ± 0.2 μM; Carland et al., 2004). Accordingly, we used 1 μM GABA to elicit responses to test the actions of steroid.

The consequences of this mutation on the actions of drugs are shown in Table 2. The ability of 5β-reduced steroids to inhibit the response is diminished to such an extent that it did not differ from no effect (Table 2). However, the actions of a sulfated steroid, a carboxylated steroid, 17β-estradiol, and picrotoxinin are less affected. Potentiation is not significantly affected. The results suggest that the 5β-reduced steroids form a group that is distinct from the other inhibitors. The fact that potentiation is not strongly affected also suggests that the incomplete inhibition by 3α5β17βP is not the result of combined inhibition plus potentiation, with differing EC₅₀ values.

Consequences of a Mutation to TM2 T6′. The second mutation was TM2 T6′F(ρ1 T298F). This residue was first implicated in block by picrotoxin in studies of the β subunit of the glycine receptor (Pribilla et al., 1992), and it was later shown to be critical for picrotoxin block of the GABA_A receptor (Gurley et al., 1995) and the ρ1 GABA_C receptor (Zhang et al., 1995). This mutation did not shift the activation curve for GABA (data not shown; EC₅₀ = 0.6 ± 0.2 μM); thus, 200 nM GABA was used to elicit responses.

As expected, this mutation essentially removes block by picrotoxinin (Table 3). However, the consequences of this mutation on the steady-state effects of steroids were startling (Table 3). First, it removes block by 17β-Est, suggesting similarities between the actions of estradiol and picrotoxinin. Second, it increases block by several drugs, including converting 3α5α17βCN into a relatively strong blocker.

To understand these effects, we examined the time course of the responses. As shown in Fig. 7, a large “tail” forms when drug and GABA are removed, most markedly for 3αCOOH5β17βP, 3α5α17βCN, and 3α5β17βCN. Tails are not apparent with 17β-Est, picrotoxin, 3α5α17βPOH, or 3α5β17βP, and they are less marked for other compounds. This tail is ascribed to a rapid relief of inhibition. If recovery from inhibition occurs more rapidly than deactivation of the receptor, then as receptors unblock the response actually increases when drug is removed. Often, the type of block producing these tails is an “open channel” block, but this specific mechanism is not required. It simply has to be the case that recovery from block is more rapid than deactivation. Recordings from Xenopus oocytes are not suitable for detailed kinetic analysis, so we did not attempt to correct for the time courses of the multiple processes occurring. Instead, we simply reanalyzed the traces, but we measured the amplitude of the tail current (Fig. 8; Table 3). When this is done, it seems that this mutation does not affect potentiation. It does remove block by 17β-Est and picrotoxinin. It does not strongly affect block by 5β-reduced steroids, except in the case of 3α5β17βPOH. Finally, it partially removes block by sulfated and carboxylated steroids. Again, 17β-Est and picrotoxinin fall in a group that is distinct from the other types of compounds examined.

What process produces the tails (and the greatly altered profile for steady-state block)? It seems most likely that the TM2 T6′F mutation has either produced or made apparent a novel inhibitory mechanism. The major reason for this conclusion is that the structure-activity relationship for the process differs from any of the other interactions we have examined. The additional process does not require a particular stereochemistry at the 5-position, because 3α5α17βCN and 3α5β17βCN both produce tails. However, the dependence on the nature of the substituent at the 17-position is very strong; carbonitrile is much preferred over hydroxymethylketone, whereas methylketone essentially removes the ability to produce inhibition by this mechanism. The relief from block by this mechanism also is significantly faster than for the other inhibitory mechanisms we have examined. In this context, we should note that in some applications of picrotoxinin to wild-type receptors we have noted a small tail response, and others have reliably seen such tails (Qian et al., 2005). We might have missed them for technical reasons (e.g., slow perfusion). However, we think that it is unlikely that the same mechanism is involved in picrotoxin block of wild-type receptors and steroid block of T6′F receptors. A major reason is simply that the mutation removes block by picrotoxinin and 17β-Est.

Discussion

These results have extended the analysis of steroid actions at the ρ1 GABA_C receptor in several ways. As a result of our studies of enantiomer pairs (also see Li et al., 2006b), it seems that these steroids interact with specific chiral sites, most probably on the ρ1 GABA_C receptor itself. Because the comparisons are relatively extensive, we start by considering potentiation by 5α-reduced steroids and inhibition by 5β-reduced steroids (Morris et al., 1999).

Potentiation by 5α Compounds and Inhibition by 5β Compounds. Previous work has shown that potentiation and inhibition seem to be independent processes. When 3α5β17βP (10 μM) was applied in the presence of 10 μM 3α5α17βP, the amount of block was indistinguishable from the block in the absence of 3α5α17βP (Li et al., 2006b). This observation indicates that the two types of drug do not compete for the same site. However, the present data show similarities in the structural requirements for steroid action. For both inhibition and potentiation, drugs containing a 3β-hydroxy group are inactive, as are drugs containing a 17α-carbonitrile. Furthermore, the types of the substituent at the 17-position that support potentiation or inhibition are broadly similar, with hydroxy or keto groups resulting in relatively weak compounds for either action. Thus, the sites mediating the two effects may be similar in structure, although distinct. Neither of the mutations in the TM2 region that we examined had a significant effect on the ability of 5α-reduced steroids to potentiate, suggesting that the mechanisms for potentiation and inhibition are distinct. We discuss other studies of mutated receptors below.

5β-Reduced steroids interact preferentially with inactive receptors, because the inhibition is reduced at higher concentrations of GABA, and it is likely to be allosteric, because the maximal inhibition can be less than complete (also see Morris et al., 1999; Goutman and Calvo, 2004). Somewhat surprisingly, a mutation (TM2 P2′S) homologous to a mutation that has been shown to reduce the ability of sulfated steroids to block the GABA_A receptor has the largest effect on inhibition by 5β-reduced steroids. It has previously been argued that this residue is unlikely to form part of a binding site for 3βSO₄5Δ17βP (pregnenolone sulfate) on the GABA_A receptor (Akk et al., 2001). We note that an earlier study found that a different mutation at this site (TM2 P2′A) did not affect inhibition by 3α5β17βP (Morris and Amin, 2004).

A comparison to the steroid structure-activity relationship for actions on the GABA_A receptor is more complex, for several reasons. First, 5β steroids potentiate the GABA_A receptor. The consequences of a 3β substituent orientation are also somewhat different. For the ρ1 GABA_C receptor, this orientation removes both potentiation or block (at least for the concentrations tested), whereas for the GABA_A receptor 3β steroids (e.g., 3β5α17βP) block weakly by a mechanism similar to that of 3βSO₄5Δ17βP (Wang et al., 2002). However, the lack of potentiation by steroids with 3β substituents is in agreement with our observations on the ρ1 GABA_C receptor. The requirement for the 17β configuration for potentiation is found for the GABA_A receptor (Phillipps, 1975), although it does not seem to have been examined for inhibitory steroids. Therefore, overall, there are some strong similarities between the structural requirements for steroid actions at the GABA_A and ρ1 GABA_C receptors. The major difference is in the effects of 5β-reduced steroids, which potentiate GABA_A and block ρ1 GABA_C receptors. It should be noted that there are some indications that 5α- and 5β-reduced steroids may have nonidentical binding sites on the GABA_A receptor (Gee et al., 1988; Mennerick et al., 2004), so the possibility exists that the sites may be similar on both types of receptor but transduction mechanisms differ. It is also known that potentiating steroids interact with more than one site on the GABA_A receptor (Akk et al., 2004); the studies of ρ1 GABA_C receptors were not performed with techniques that could address this issue.

We note that many steroids and analogs inhibit the rodent nicotinic α4β2 receptor (Sabey et al., 1999; Paradiso et al., 2000). The structure-activity relationship for this inhibition suggests that the site involved is distinct from any of the sites we have characterized on the ρ1 GABA_C receptor.

Inhibition by Sulfated or Carboxylated Compounds. Turning to inhibition of ρ1 GABA_C receptors, several observations were noted under Results that suggest that there is more than one mechanism for producing inhibition of the ρ1 GABA_C receptor. Inhibition could be produced by 5β-reduced steroids, by sulfated or carboxylated steroids, and by estrogens. In comparing sulfated to 5β-reduced steroids, both the 5α- and 5β-reduced sulfates inhibited. This is a clear difference to the inhibition by unsulfated steroids, for which the 5β configuration is required. Inhibition by sulfated steroids is reduced at higher GABA concentrations, suggesting that they interact preferentially with the inactive receptor. It does not seem to involve binding in which the charge on the steroid interacts with the membrane field, because the inhibition is voltage-independent. Somewhat surprisingly, a mutation homologous to a mutation that has been shown to reduce the ability of sulfated steroids to block the GABA_A receptor has no effect on block by 3αSO₄5β17βP, whereas it reduced inhibition by the uncharged 5β steroids tested. Overall, it seems likely that the receptor-steroid interactions that mediate inhibition by sulfated steroids (and possibly the mechanism as well) differ from those mediating inhibition by nonsulfated 5β steroids.

Many aspects of the effects of sulfated steroids differ between the ρ1 GABA_C and GABA_A receptors. One aspect is that the 3β-sulfated steroids do not block the ρ1 GABA_C receptor, whereas they do block the GABA_A receptor (Mennerick et al., 2001). Block of GABA_A receptors by 3αSO₄5β17βP is voltage-dependent, and it is enhanced by higher concentrations of GABA (Mennerick et al., 2001), possibly reflecting a form of open-channel block. Finally, block is reduced by a mutation, α1V2′S (Wang et al., 2002), that did not strongly affect block of the ρ1 GABA_C receptor. 3βSO₄5Δ17βP differs from other sulfated steroids in that the block of the GABA_A receptor is not voltage-dependent (Akk et al., 2001; Eisenman et al., 2003). The mechanism by which pregnenolone sulfate acts on GABA_A receptors is not known, although it has been proposed that it enhances desensitization (Shen et al., 2000; Akk et al., 2001).

A 3-carboxysteroid (3αCOOH5β17βP) produces the largest inhibition of any of the drugs we tested at 10 μM. In common with the sulfated steroids, inhibition does not show voltage dependence. In clear distinction to the sulfated steroids, 3αCOOH5α17βP had little effect at either block or potentiation. This may mean that this compound shares some aspects of inhibition with other 5β-reduced compounds. However, the effects of two mutations showed a rather different pattern than for uncharged 5β-reduced steroids. This analog potentiates GABA_A receptors when applied at 10 μM concentrations, but it acts as a voltage-dependent blocker at higher concentrations (Mennerick et al., 2001), and its mechanism of action is not clear.

Inhibition by Estradiols. The most surprising observation with natural steroids is the block by 17β-estradiol, which is an effective inhibitor. 17β-Estradiol is relatively inactive on the GABA_A receptor (Akk et al., 2007). It potentiates the human nicotinic α4β2 receptor (Paradiso et al., 2001), whereas it inhibits the rat isoform (Paradiso et al., 2000). At ρ1 GABA_C receptors, estradiols are also unusual in that the 17 hydroxy group can be oriented in either the 17α or 17β configuration, with equal effect. The saturated androstanediols are largely inactive, irrespective of stereochemistry, indicating the importance of the unsaturated A ring for this action. Adding a sulfate group to either the 3- or 17-position of 17β-Est reduces activity, suggesting a difference to inhibition by sulfated steroids. To our surprise, inhibition by 17β-Est is removed completely by a mutation, which also removes inhibition by picrotoxinin. Inhibition by estradiols does not fit well with the structural requirements for inhibition by either 5β-reduced steroids or sulfated steroids, and it is likely to reflect an additional site.

Other Studies of the Effects of Mutations to the ρ1 GABA_C receptor on Steroid Actions. An extensive study has been made of the consequences of mutations to a residue in the TM2 region of the ρ1 GABA_C receptor, I307, which is near the extracellular end of the helix (Morris and Amin, 2004). This residue is homologous to a residue that has been implicated in several drug effects on the GABA_A receptor. Initially, it was reported that in the β subunits the nature of this residue is critical for the action of etomidate (Belelli et al., 1997). Subsequent work has also indicated the importance of this residue, in the α subunit, in the actions of volatile anesthetics and alcohols (Mascia et al., 2000; Jenkins et al., 2001). Finally, the mutation ρ1 I307S makes the receptors sensitive to potentiation by pentobarbital (Belelli et al., 1999). It has been proposed that this residue forms part of a pocket behind the TM2 membrane-spanning helix, into which a drug may insert itself to produce effects (Mascia et al., 2000; Jenkins et al., 2001). A large series of substitutions have been made of ρ1 I307, and the consequences for actions of two inhibitory steroids (3α5β17βP and 5β-dihydroprogesterone or 3K5β17βP) and one potentiating steroid (3α5α17βP) were determined (Morris and Amin, 2004). The pattern of results is complicated and difficult to interpret. In brief, inhibitory steroids, in general, could be converted to show either mixed effects (inhibition at lower concentrations and potentiation at higher concentrations) or only potentiating effects. A more limited set of mutations was tested on the potentiating steroid, but, in general, potentiation was enhanced. The initial interpretation was that neuroactive steroids had their actions by effects in the lipid surrounding the ρ1 GABA_C receptor, which were sensed by this residue and transduced into functional changes (Morris and Amin, 2004). This initial interpretation is made much less likely by the strong enantioselectivity of many steroid actions (see above; Li et al., 2006b), but the pattern of observations is difficult to explain by a unifying theory. However, we note that in our more limited examination of the consequences of mutations we have obtained evidence that a point mutation may have complicated effects. In the TM2 T6′F mutation, we were fortunate that the appearance of a novel form of inhibition was signaled by the tail currents.

Summary. In terms of the first objective, examination of a greater number of endogenous steroids, two major observations can be made. First, estradiols are much more effective inhibitors of the ρ1 GABA_C receptor than of GABA_A receptors. Second, the 3α configuration is required for all other steroid actions examined. In terms of the second objective, to clarify the structure-activity relationship for the actions of steroids on the GABA_C receptor and compare it with that for the GABA_A receptor, the results of this initial survey of the structure-activity relationship for steroid actions on the ρ1 receptor suggest multiple sites for interaction between steroids and the ρ1 receptor. The sites mediating potentiation by 5α- and inhibition by 5β-reduced steroids have similar qualitative requirements, and some similarities to those for potentiation of GABA_A receptors. In contrast, the sites mediating inhibition by sulfated steroids and by estradiols seem to differ from those sites, from each other, and from sites characterized on other members of the nicotinic receptor family.