Site-specific effects of neurosteroids on GABAA receptor activation and desensitization

This study examines how site-specific binding to three identified neurosteroid-binding sites in the α1β3 GABAA receptor (GABAAR) contributes to neurosteroid allosteric modulation. We found that the potentiating neurosteroid, allopregnanolone, but not its inhibitory 3β-epimer epi-allopregnanolone, binds to the canonical β3(+)–α1(-) intersubunit site that mediates receptor activation by neurosteroids. In contrast, both allopregnanolone and epi-allopregnanolone bind to intrasubunit sites in the β3 subunit, promoting receptor desensitization and the α1 subunit promoting effects that vary between neurosteroids. Two neurosteroid analogues with diazirine moieties replacing the 3-hydroxyl (KK148 and KK150) bind to all three sites, but do not potentiate GABAAR currents. KK148 is a desensitizing agent, whereas KK150 is devoid of allosteric activity. These compounds provide potential chemical scaffolds for neurosteroid antagonists. Collectively, these data show that differential occupancy and efficacy at three discrete neurosteroid-binding sites determine whether a neurosteroid has potentiating, inhibitory, or competitive antagonist activity on GABAARs.


INTRODUCTION
cleft between the α subunits, with the C3-hydroxyl substituent of the steroids interacting directly with Q242 23 in the α subunit (αQ242). PAM-NS activate these chimeric receptors, and their action is blocked by αQ242L 24 and αQ242W mutations. These studies posit a single canonical intersubunit binding site for NS action that receptors, since both open and desensitized GABAAR exhibit enhanced orthosteric ligand binding affinity 11 (37). 12 13 The results show that 3α5αP binds to the canonical β(+)-α(-) intersubunit site, stabilizing the open state 14 of the receptor, whereas the 3-diazirinyl NS (KK148 and KK150) bind to this site but do not promote 15 channel opening, and 3β5αP does not occupy this site. These data indicate that NS binding to the 16 intersubunit sites is largely responsible for PAM activity and that the 3α-OH is critical for NS activation. In contrast, 3α5αP, 3β5αP and the 3-diazirinyl NS all bind to both the α1 and β3 intrasubunit sites. Occupancy desensitization (13,14) eliminated [ 3 H]muscimol binding enhancement by 3β5αP but not 3α5αP ( Figure 3).

23
We infer that 3α5αP increases [ 3 H]muscimol binding by stabilizing an active state of the receptor, whereas 24 3β5αP increases [ 3 H]muscimol binding by stabilizing a desensitized state of the receptor which is absent in 25 the α1(V256S)β3 receptor. Collectively, these data indicate that 3β5αP and KK148 stabilize a desensitized 8 state of the GABAAR, thus enhancing orthosteric ligand affinity.

Quantitative comparison of the effects of 3β5αP on [ 3 H]muscimol binding and receptor desensitization.
3 While there is qualitative agreement between the relative effects of the various NS analogues on 4 orthosteric ligand binding and receptor desensitization, there is an apparent quantitative discrepancy in the 5 magnitude of the effects. For example, 3β5αP enhances [ 3 H]muscimol binding by two-fold ( Figure 1E), 6 whereas it reduces steady-state current by only ~25% ( Figure 2C). To address this difference, we first state. To address the quantitative differences in results from the two assays, we analyzed the 13 electrophysiological data in the framework of the three-state Resting-Open-Desensitized model (44,45). 14 We assumed that both the open and desensitized states had higher affinity for muscimol than the resting 15 state, and that the affinities were similar and could be treated as equal. We then calculated the predicted 16 occupancy of the high affinity states (Popen + Pdesensitized) using parameters derived from the functional 17 responses, to compare to the observed changes in binding. The raw current amplitudes of peak and steady-Application of 1 mM GABA elicited a current response that had a peak Popen of 0.71 ± 0.25 (n = 16).

23
The Popen of the steady-state response was 0.121 ± 0.033 (n = 7), that was reduced to 0.077 ± 0.013 (n = 5) 24 with 3 µM 3β5αP. Analysis of steady-state currents using the Resting-Open-Desensitized model indicates 9 steroid. The relatively small increase in the sum of (Popen + Pdesensitized) (from 0.95 to 0.97) is due to the use 1 of saturating GABA in these experiments. In the presence of lower concentrations of GABA, i.e., lower 2 levels of activation, the predicted increase in the sum of (Popen + Pdesensitized) is greater. For example, for the 3 condition where GABA elicits a peak response with Popen of 0.01, the predicted steady-state Popen is 0.0094 4 and the predicted Pdesensitized 0.0643 (sum of the two equals 0.0737). In the presence of GABA + 3β5αP, the 5 predicted steady-state Popen is 0.0090 and the predicted Pdesensitized 0.1045 (sum of 0.1135), thus showing a 6 54% increase in the sum of (Popen + Pdesensitized). Incidentally, this example demonstrates the need to use high 7 concentrations of GABA to observe a meaningful reduction in steady-state Popen in the presence of 3β5αP.

9
To more directly compare the data from the radioligand binding and electrophysiology experiments, we exposed oocytes containing α1β3 GABAARs to 20 nM muscimol and recorded currents before and after co-11 application of 3 μM 3β5αP. The percent reduction in steady-state current following 3β5αP exposure was   indicate that relatively small changes in steady-state current can be associated with relatively large changes 20 in the occupancy of high-affinity states.

22
Binding site selectivity for NS analogues.

23
To determine whether KK148 and 3β5αP stabilize the desensitized conformation of the GABAAR by 24 selectively binding to one or more of the identified NS binding sites on the GABAAR (11), we first 25 determined which of the identified NS sites they bind. We have previously shown that the 3α5αP-analogue photolabeling reagent, KK200 labels the β3(+)-α1(-) intersubunit (β3G308) and α1 intrasubunit (α1N408) 1 sites on α1β3 GABAARs ( Figure 4A), and that photolabeling can be prevented by a ten-fold excess of 3α5αP 2 (11). As a first step to determine the binding sites for 3β5αP, KK148 or KK150, we examined whether a 3 ten-fold excess of these compounds (30 μM) prevented KK200 (3 μM) photolabeling of either binding site.

4
Photolabeling was performed on membranes from HEK293 cells transfected with epitope-tagged α1His-5 FLAGβ3 receptors, mimicking the conditions used in the [ 3 H]muscimol binding assays and photolabeled 6 residues were identified and labeling efficiency was determined using middle-down mass spectrometry (11).

7
KK148, KK150, 3α5αP and 3β5αP all prevented KK200 photolabeling of α1N408 in the α1 intrasubunit site 8 ( Figure 4B), consistent with their binding to this site. In contrast, KK148, KK150 and 3α5αP but not 3β5αP 9 prevented labeling of β3G308 in the intersubunit site ( Figure 4C), indicating that 3β5αP does not bind to 10 the intersubunit site.

12
The KK148-and KK150-photolabeled samples were also analyzed to directly identify the sites of supplement 2); Y415 is the same residue labeled by KK123 at the α1 intrasubunit site (11). The KK148 and 20 KK150 labeled peptides in TM4 of the β3 subunit and corresponding unlabeled peptide were identified by 21 fragmentation spectra as β3TM4 I426-N445. These data support labeling of the β3 intrasubunit site by KK148 and KK150. Fragmentation spectra of the peptide-sterol adducts were not adequate to determine 23 the precise labeled residue because of low photolabeling efficiency (0.13% for KK148; 0.19% for KK150, labels nucleophilic residues (28,34,47); such residues are not present in the intersubunit site.

Orthosteric ligand binding enhancement by NS analogues is mediated by distinct sites.
To determine which of the previously identified binding sites contributes to NS enhancement of 11 [ 3 H]muscimol binding, we performed site-directed mutagenesis of the NS binding sites previously 12 determined by photolabeling ( Figure 6A) (11). Specifically, α1(Q242L)β3 targets the β3(+)-α1(-)  Table 1). The residual enhancement of [ 3 H]muscimol binding observed in receptors 17 with mutations in the intersubunit or α1 intrasubunit site occurs at ten-fold higher concentrations of 3α5αP 18 than WT and receptors with mutations in the β3 intrasubunit site (

19
It is important to note that the [ 3 H]muscimol binding curves in Figure 6 are normalized to control. The     3α5αP produced a small reduction in steady-state current in α1(Q242L)β3 receptors with mutations in neither of the intrasubunit sites ( Figure 9A). This inhibitory effect was eliminated by α1(V256S)β3,

23
indicating that it was due to receptor desensitization ( Figure 9D). In receptors with combined mutations in 24 the intersubunit and α1 intrasubunit sites [i.e. α1(Q242L/N408A/Y411F)β3], 3α5αP significantly inhibited the steady-state current ( Figure 9B), an effect that was markedly reduced by mutations in the β3 intrasubunit desensitization ( Figure 9D). Notably, 3α5αP exerted only a modest inhibitory effect in α1(Q242L)β3 3 receptors in which occupancy of the β3 intrasubunit site should promote inhibition. This may be due to a 4 counterbalancing action at the α1 intrasubunit site, where 3α5αP binding contributes more to receptor 5 activation as demonstrated by our previous observation that mutations in the α1 intrasubunit site 6 significantly reduce 3α5αP potentiation of GABA-elicited currents (11). These results suggest that in 7 addition to activation, 3α5αP enhances receptor desensitization. Enhanced desensitization by the PAM-NS 8 3α5αP (54) and 3α5α-THDOC (55,56) has been observed in prior studies supporting the current finding 9 with 3α5αP.

1
In this study we examined how site-specific binding to the three identified NS sites on α1β3 GABAAR 2 (11) contributes to the PAM vs. NAM activity of epimeric 3-OH NS. We found that the PAM-NS 3α5αP, 3 but not the NAM-NS 3β5αP, binds to the canonical β3(+)-α1(-) intersubunit site that mediates receptor 4 potentiation, explaining the absence of 3β5αP PAM activity. In contrast, 3β5αP binds to intrasubunit sites 5 in the α1 and β3 subunits, promoting receptor desensitization. Binding to the intrasubunit sites provides a 6 mechanistic explanation for the NAM effects of 3β5αP (14). 3α5αP also binds to the β3 intrasubunit site 7 explaining the previously described desensitizing effect of the PAM-NS 3α5αP (54) and 3α5α-THDOC 8 (55,56). Two synthetic NS with diazirine moieties at C3 (KK148 and KK150) were used to identify NS 9 binding sites and shown to bind to the intersubunit as well as both intrasubunit sites. Neither of these ligands 10 potentiated agonist-activated GABAAR currents, reinforcing the importance of the 3α-OH group and its 11 interaction with α1Q242 in PAM actions. KK148 is an efficacious desensitizing agent, acting through the 12 α1 and β3 intrasubunit NS binding sites. KK150, the 17α-epimer of KK148, binds to all three NS binding 13 sites, but neither activates nor desensitizes GABAARs, suggesting a potential chemical scaffold for a general 14 NS antagonist. Collectively, these data show that differential occupancy of and efficacy at three discrete NS 15 binding sites determines whether a NS ligand has PAM, NAM, or potentially NS antagonist activity on 16 GABAARs.

18
The observation that 3β5αP and KK148 enhance orthosteric ligand binding but do not potentiate GABA-

19
The selective binding of 3β5αP to a subset of identified NS binding sites provides an explanation for its 20 NAM activity. 3β5αP stabilizes desensitized receptors by binding to the α1 and β3 intrasubunit sites, but 21 does not activate the receptor because it does not bind to the intersubunit site. This site-selective binding is 22 unexpected for several reasons. First, docking and free energy perturbation calculations in a prior study 23 predicted that epi-pregnanolone (3β5βP) binds to the intersubunit site in a similar orientation and with free 24 energies of binding that are equivalent to pregnanolone (3α5βP) (23). The modeling suggested that 3β5βP studies also show similar binding energies and orientations of 3β5αP and 3α5αP binding in the β3(+)-α1(-) 1 intersubunit site (Supplementary file 2). We have also shown that binding affinity or docking scores of NS 2 binding to the intersubunit site is not significantly affected by mutations (α1Q242L, α1Q242W, α1W246L) 3 that eliminate NS activation, although binding orientation is altered (28). These data indicate that NS 4 binding in the intersubunit site is tolerant to significant changes in critical residues and NS ligand structure, 5 and are consistent with our findings that NS analogues, such as KK148 and KK150, can bind to the 6 intersubunit site but have no effect on activation (Figures 1 and 4). Thus, the peculiar lack of 3β5αP binding 7 to the intersubunit site suggests that either: (1) details in the structure of the intersubunit site in the open 8 conformation that explain the absence of 3β5αP binding are not apparent in current high-resolution 9 structures or; (2) 3β5αP does not bind for other reasons. One plausible explanation is that 3β5αP, like cholesterol, has low chemical activity in the membrane and does not achieve sufficiently available 11 concentration to bind in this site (58). This explanation would require that the chemical activity of 3β5αP 12 differs between the inner and outer leaflets of a plasma membrane since 3β5αP binds to the intrasubunit 13 sites.

15
The functional analysis of mutations in each of the three NS binding sites demonstrates that the     Figure 10A). In contrast, 3β5αP preferentially stabilizes the desensitized 7 state through binding to both intrasubunit sites ( Figure 10B). KK148, like 3β5αP, stabilizes the desensitized 8 state by binding to the intrasubunit sites ( Figure 10C). KK148 also binds to the intersubunit site, presumably 9 with no state-dependence, since it is neither an agonist nor an inverse-agonist ( Figures 1C and 10C). KK150, 10 which neither activates nor desensitizes GABAARs and is not an inverse agonist, binds to all three sites, 11 again presumably with no-state dependence (Figures 1D and 10D). This model predicts that KK148 should act as a competitive antagonist to PAM-NS at the intersubunit site. Indeed, KK148 reduces 3α5αP

20
The site-specific model of NS action ( Figure 10) has significant implications for the synaptic 21 mechanisms of PAM-NS action. At a synapse, GABAARs are transiently exposed to high (mM) concentrations of GABA leading to a channel Popen approaching one (61,62). GABA is quickly cleared from 23 the synapse leading to rapid deactivation with minimal desensitization (57,63). In the presence of a PAM-increased inhibitory current (18,55,64,65). This effect is largely attributable to stabilization of the open 1 state, presumably by binding to the intersubunit and α1 intrasubunit binding sites. A second effect has been 2 observed in which the PAM-NS 3α5αP (54) and 3α5α-THDOC (55) prolong the slow component of 3 GABAAR desensitization and slow recovery from desensitization. This results in increased late channel 4 openings (55,57) and IPSC prolongation (64,65). When the frequency of synaptic firing is rapid, the 5 desensitizing effect of NS may also contribute to frequency-dependent reduction in IPSC amplitude (55,66).

6
The desensitizing effect of 3α5αP is predominantly mediated by binding at the β3 intrasubunit site. The      The human α1 and β3 subunits were subcloned into pcDNA3 for molecular manipulations and cRNA 3 synthesis. Using QuikChange mutagenesis (Agilent Technologies, Santa Clara, CA), a FLAG tag was first 4 added to the α1 subunit then an octa-histidine tag was added to generate the following His-FLAG tag tandem 5 (QPSLHHHHHHHHDYKDDDDKDEL), inserted between the 4th and 5th residues of the mature peptide.

6
The α1 and β3 subunits were then transferred into the pcDNA4/TO and pcDNA5/TO vectors (Thermo Fisher 7 Scientific), respectively, for tetracycline inducible expression. Point mutations were generated using the 8 QuikChange site-directed mutagenesis kit and the coding region fully sequenced prior to use. The cDNAs 9 were linearized with Xba I (NEB Labs, Ipswich, MA), and the cRNAs were generated using T7 mMessage 10 mMachine (Ambion, Austin, TX).

13
Cell culture was performed as described in previous reports (11). The tetracycline inducible cell line Tin a humidified atmosphere containing 5% CO2. Cells were passaged twice each week, maintaining          The oocytes were injected with a total of 12 ng cRNA. The ratio of cRNAs was 5:1 ratio (α1:β3) to minimize 1 the expression of β3 homomeric receptors. Following injection, the oocytes were incubated in ND96 (96 2 mM NaCl, 2 mM KCl, 1.8 mM CaCl2, 1 mM MgCl2, 5 mM HEPES; pH 7.4) with supplements (2.5 mM 3 Na pyruvate, 100 U/ml penicillin, 100 μg/ml streptomycin and 50 μg/ml gentamycin) at 16 o C for 2 days 4 prior to conducting electrophysiological recordings. The electrophysiological recordings were conducted 5 at room temperature using standard two-electrode voltage clamp. The oocytes were clamped at -60 mV.

6
The chamber (RC-1Z, Warner Instruments, Hamden, CT) was perfused with ND96 at 5-8 ml/min. Solutions    [X] is the concentration of agonist present, and other terms are as described above. In practice, the value 15 of LΓ was calculated using the experimentally-determined Popen of the peak response, and then used as a 16 fixed value in estimating Q from Popen,steady-state.

17
The Pdesensitized was calculated using:   these were then clustered geometrically using the program DIVCF (76). The resulting clusters were ranked 1 by Vina score and cluster size, and then visually analyzed.         The data for WT in panels 6B and 6D is a replot of the same data shown in Figure 1E.      ). 3β5αP stabilizes a desensitized state through the β3 and α1 intrasubunit sites. (C) Same as (A) for KK148 ( ). KK148 allosterically stabilizes a desensitized state through the β3 and α1 intrasubunit sites, and equally stabilizes all three states of the receptor through the β3-α1 intersubunit site. The width of orange arrows indicates relative affinities of KK148 for each state of the receptor. (D) Same as (A) for KK150 ( ). KK150 equally stabilizes all three states of the receptor through the β3 and α1 intrasubunit sites, and the β3-α1 intersubunit site.

WTα1β3
EC50 (nM) Hill slope  Potentiation response ratio Reduction of potentiation in the presence of competitor (%) # Cells 0.1 μM 3α5αP alone + 1 μM competitor 3α5αP vs. KK148 5.1 ± 2.0 4.0 ± 1.2 24 ± 16 6 3α5αP vs. KK150 6.4 ± 2.1 5.8 ± 1.9 12 ± 3 5 SUPPLEMENTARY FILE 3: KK148 and KK150 decrease allopregnanolone-induced potentiation. Neurosteroid potentiation of α1β3 GABAARs in Xenopus laevis oocytes is expressed as potentiation response ratio, calculated as the ratio of the peak responses in the presence of GABA and 0.1 μM allopregnanolone (3α5αP) to the peak response in the presence of GABA alone. The GABA concentrations were selected to generate a response of 3-9% of the response to saturating GABA (30 μM). Co-application of KK148 or KK150 reduces the potentiating effect of 3α5αP on GABAelicited currents. Data are shown as mean ± SD.