Molecular determinants involved in the allosteric control of agonist affinity in the GABAB receptor by the GABAB2 subunit

The gamma-aminobutyric acid type B (GABAB) receptor is an allosteric complex made of two subunits, GABAB1 (GB1) and GABAB2 (GB2). Both subunits are composed of an extracellular Venus flytrap domain (VFT) and a heptahelical domain (HD). GB1 binds GABA, and GB2 plays a major role in G-protein activation as well as in the high agonist affinity state of GB1. How agonist affinity in GB1 is regulated in the receptor remains unknown. Here, we demonstrate that GB2 VFT is a major molecular determinant involved in this control. We show that isolated versions of GB1 and GB2 VFTs in the absence of the HD and C-terminal tail can form hetero-oligomers as shown by time-resolved fluorescence resonance energy transfer (based on HTRF technology). GB2 VFT and its association with GB1 VFT controlled agonist affinity in GB1 in two ways. First, GB2 VFT exerted a direct action on GB1 VFT, as it slightly increased agonist affinity in isolated GB1 VFT. Second and most importantly, GB2 VFT prevented inhibitory interaction between the two main domains (VFT and HD) of GB1. According to this model, we propose that GB1 HD prevents the possible natural closure of GB1 VFT. In contrast, GB2 VFT facilitates this closure. Finally, such inhibitory contacts between HD and VFT in GB1 could be similar to those important to maintain the inactive state of the receptor.


INTRODUCTION
γ-aminobutyric acid (GABA), a major inhibitory neurotransmitter in the central nervous system, activates GABA A ligand-gated Clchannels, as well as the G-protein-coupled receptor (GPCR) GABA B (1,2). This receptor is found in either pre-or post-synaptic elements in various types of neurons. As such, GABA B receptors play important role in brain function as illustrated by the antispastic activity of the GABA B selective agonist baclofen (Lioresal®), and its involvement in various types of epilepsy, as well as in nociception and drug addiction (2). GABA B receptors belong to the class-III GPCRs, together with metabotropic glutamate (mGlu), extracellular Ca 2+ -sensing, and some pheromone and taste receptors (3).
Each of these receptors is composed of an extracellular domain called Venus flytrap (VFT) where agonists bind, and an heptahelical domain (HD) responsible for the recognition and contains all the determinants required for G-protein coupling and plays a pivotal role in Gprotein activation by the heteromer (11)(12)(13)(14)(15). Third, GB2 increases agonist, but not antagonist affinity on GB1 (4,11), even though it does not appear to bind any natural ligand (16).
The ligand binding site of GB1 has been extensively studied (17)(18)(19)(20)(21)(22). Modeling and site-directed mutagenesis studies indicate ligands bind in the cleft that separates both lobes of the GB1 VFT, as observed for ligand binding in many similar protein modules (23) including the mGlu1 VFT (24,25). Antagonists are expected to prevent the closure of the GB1 VFT (19), as observed in mGlu receptors (25,26). Conversely, agonists interact with residues from both lobes of GB1 VFT and they stabilize a closed form of this domain (16,19). Such a domain closure of the GB1 VFT has recently been shown to be sufficient to activate this heterodimeric receptor (27).
Mechanism of the allosteric control of agonist affinity in GB1 by GB2 is unknown. It is likely that the GB2 subunit controls agonist affinity by further stabilizing the closed state of the GB1 VFT (28). Understanding this mechanism should have implications in the current model of GABA B receptor activation, as it may help to explain how GABA binding in the GB1 VFT can activate the GB2 HD. In addition, it may open new routes for the development of positive allosteric compounds known to stabilize the active conformation of the dimeric HDs and VFTs and to increase agonist affinity (29)(30)(31).
Here, we demonstrate that direct interaction between GB1 and GB2 VFTs is responsible for the increase in agonist affinity in two ways. First, it prevents the HD of GB1 to decrease agonist affinity, and second, the interaction between the two VFTs by itself further increases agonist affinity in GB1 VFT.

Western blotting
Twenty hours after transfection, HEK-293 cells were washed with PBS (Ca 2+ -and Mg 2+ -free) and harvested. The membranes were prepared as previously described (26). For each sample, 50 µg of total protein was subjected to SDS-PAGE by using 10 % polyacrylamide gels, transferred to nitrocellulose membrane (Hybond-C; Amersham Pharmacia), and probed with anti-HA mouse monoclonal antibody (clone 12CA5; Roche, Basel, Switzerland) at 0.1 µg/mL. Proteins were visualized by chemiluminescence (West Pico; Pierce, Rockford, IL).

Cell surface quantification by ELISA
Twenty hours after transfection with HA-tagged versions of the constructs, HEK-293 cells were washed twice with phosphate-buffered saline solution (PBS), fixed with 4% paraformaldehyde in PBS and then blocked with PBS plus 5% FBS. After 30 minutes reaction with primary antibody (monoclonal anti-HA clone 3F10; Roche, Basel, Switzerland) at 0.5 µg/mL) in the same buffer, the goat anti-rat antibody coupled to horseradish peroxidase (Jackson Immunoresearch, West Grove, PA) was applied for 30 minutes at 1 µg/mL. After intense washes with PBS, secondary antibody was detected and quantified instantaneously by chemiluminescence (Supersignal West Femto; Pierce, Rockford, IL) using a Wallac Victor 2 luminescence counter (Molecular Devices, St Grégoire, France).

Ligand binding assay
Ligand binding assay on intact HEK-293 cells was performed as previously described

Time-resolved FRET measurements
Time-resolved FRET experiments were conducted as described by Maurel et al. 1

GB2 increases agonist affinity, but not antagonist affinity on GB1
Although GB1 bind any known GABA B ligands, agonist affinities are about 100 times lower than those measured on native receptors (36). This is partly due to GB2, since its coexpression with GB1 results in a 10 fold increase in agonist affinity (4). This effect of GB2 does not result from the targeting of GB1 to the cell surface, and so to a mature glycosylation state, since a GB1 mutant able to reach the cell surface alone (GB1 ASA in which the ER retention signal RSRR is mutated into ASAR) still displays a low agonist affinity at the cell surface ( Fig. 1b and Table 1). A direct association between GB1 and GB2 appears therefore necessary to control agonist affinity in this receptor. Co-expression of GB1 with GB2/1, a chimeric subunit composed of the VFT of GB2 and the HD of GB1, or the replacement of the HD of GB1 by that of GB2 in the chimeric GB1/2 subunit, also resulted in an increased GABA affinity ( Fig. 1b and Table 1). These observations suggest that both the GB2 VFT and the HD of GB1 control agonist affinity on GB1.

VFT module constructs for GB1 and GB2
In order to elucidate the mechanism leading both GB2 VFT and GB1 HD to allosterically control agonist affinity on GB1, isolated GB1 and GB2 VFTs were prepared by removing the HD and C-terminal domains of each subunit. These constructs were generated to examine the possible direct interaction between GB1 and GB2 VFTs and the consequence on GABA affinity, regardless the other regions of the subunits. To express either GB1 or GB2 VFTs at the cell surface, two series of constructs were prepared. The first constructs called ∆GB1 and ∆GB2 were generated by introducing a stop codon at the end of the first intracellular loop of GB1 and GB2, respectively. As such, the VFTs were anchored in the plasma membrane by the first transmembrane helix (TM1) of these subunits (Fig. 2a). For the second constructs called GB1 GPI and GB2 GPI , the entire HD and C-terminal tail were replaced by the glycosylphosphatidylinositol (GPI) anchor sequence of the PrPc prion protein (see Material and Methods and Fig. 2a). In all cases, these constructs contain an N-terminal epitope, either c-myc or HA inserted after a signal peptide allowing their easy detection at the cell surface. Previous studies demonstrated that such epitope affected neither the pharmacology nor the function of these subunits (10,11).
All constructs are correctly expressed in HEK-293 cells, and display the expected molecular weight as shown in Western blot experiments (Fig. 2b). All constructs except GB2 GPI were found at the cell surface but at a density two to ten times lower than that of GB1 ASA as revealed by an anti-HA ELISA performed on intact cells (Fig. 2c). Finally, ∆GB1 and GB1 GPI bind a competitive and membrane non-permeant radio-labeled antagonist [ 125 I]CGP64213, and this binding can be displaced by GABA demonstrating these constructs retained their ability to bind GABA B ligands ( Fig. 2d and Table 1). All together, these results show that ∆GB1, ∆GB2 and GB1 GPI are correctly expressed at the cell surface, and that ∆GB1 and GB1 GPI are correctly folded.
To examine whether an interaction between GB1 and GB2 VFTs could be detected, the above described truncated GB1 and GB2 subunits as well as the wild-type subunits were co-expressed. GB2 and ∆GB2 increase expression of ∆GB1 and GB1 GPI at the cell surface.
Binding experiments show that total amount of radioligand bound to ∆GB1 at the cell surface is increased in the presence of ∆GB2 (Fig. 3a). A higher increase is observed when ∆GB1 is co-expressed with the full-length GB2 (Fig. 3a). Similarly, the amount of [ 125 I]CGP64213 bound to GB1 GPI at the cellular surface is also increased when this construct is co-expressed with either ∆GB2 and GB2 (Fig. 3a). Since neither GB2 nor ∆GB2 changes CGP64213 affinity on these GB1 constructs (Table 1), these data show that GB2 and ∆GB2 increase the number of [ 125 I]CGP64213 binding sites at the cell surface. We further confirm that GB2 and ∆GB2 increase the amount of truncated GB1 constructs at the cell surface using an ELISA on intact cells. The amount of HA-∆GB1 at the cell surface is increased after co-expression with c-myc-∆GB2 or c-myc-GB2 (Fig. 3b). Thus, GB2 VFT either stabilizes ∆GB1 and GB1 GPI at the cell surface, or facilitates their targeting to the plasma membrane.

GB1 and GB2 VFT modules assemble into heterodimeric complexes in the absence of heptahelical domain
To demonstrate more directly that GB1 and GB2 VFTs interact with each other, coimmunoprecipitation experiments were performed. Unfortunately, no interaction between GB1 and GB2 VFTs was detected (data not shown), possibly because such complexes are not stable enough and did not resist to the sample preparation. Indeed, although the deletion of the C-terminal coiled-coil domain of the GB1 and GB2 subunits does not prevent heterodimer formation, as shown by the normal functioning of the receptor, it largely decreases the amount of GB2 co-precipitated with GB1 (9). This shows that the coil-coiled domains in the Cterminal tails of these subunits strongly stabilize the heterodimer.
We therefore conducted time-resolved fluorescence resonance energy transfer (TR-FRET) experiments as previously described (37) ( and Maurel et al., submitted). In this assay, a FRET signal is measured at the surface of intact COS-7 cells between a donor molecule (europium cryptate-PBP) linked to an anti-HA monoclonal antibody, and an acceptor molecule (Alexa Fluor® 647) linked to an anti-c-myc-monoclonal antibody (Fig. 4a, Insert).
In this assay, the HA-tagged version of ∆GB1 and GB1 GPI and the c-myc tagged version of ∆GB2 and GB2 (or other control constructs) were used. As shown in Fig. 4a and 5a, a FRET signal is measured at 665 nm (the emission wavelength of Alexa Fluor® 647) after excitation of the europium cryptate-PBP at 337 nm in cells expressing either HA-∆GB1 or HA-GB1 GPI together with c-myc-∆GB2. To better compare the association efficiency of the different partners, FRET signal was normalized with regard to c-myc tagged construct expression (4c et 5c). This signal is significantly higher than that measured in mock transfected cells.
Moreover, such a FRET signal was not detected when ∆GB1 and GB1 GPI were co-expressed with the c-myc-tagged V2 vasopressin receptor, a class-I GPCR (Fig. 4a and 5a). As shown in Fig. 4b and 5b, this did not result from the low expression at the cell surface of the c-myc tagged partners. A significant FRET signal was however measured between ∆GB1 and ∆mG5 (the VFT domain of metabotropic mGlu5 receptor anchored to in the plasma membrane via the first TM) and to a lower extent between ∆GB1 and V2-R (Fig. 4a).
However, such signals were not observed with GB1 GPI (Fig. 5a) suggesting that TM1 of ∆GB1 is likely involved in a non-specific interaction with other TM containing proteins. This may be because the TM used to anchor ∆GB1 at the cell surface (TM1 of GB1) is usually associated with other TMs in the GB1 HD.
During these experiments, we also noticed GB1 and GB2 VFT modules form homomeric complexes, as previously reported for both full-length GB1 and GB2 subunits 1 .
Indeed, a FRET signal could be measured in cells expressing HA-GB1 GPI and c-myc-∆GB1.
This signal is similar to that obtained with HA-∆GB1 and c-myc-∆GB2 or HA-GB1 GPI and cmyc-∆GB2 ( Fig. 4a and 5a), consistent with GB1 VFT being able to oligomerize. Similar FRET signal between ∆GB1 expressed alone and GB1 GPI co-expressed with ∆GB1 suggest TM1 of ∆GB1 is not responsible for its homomerisation. Moreover, we found that the GB2 VFT can also form homo-oligomers in similar FRET experiments (data not shown).

Association between VFT modules of GB1 and GB2 increases agonist affinity on GB1
We then examined the possible influence of the GB1-GB2 VFTs interaction on agonist affinity. To that aim GABA affinity was measured by displacement of radioligand [ 125 I]CGP64213 on intact cells expressing the above described constructs alone or in combination (Table 1). In contrast to agonist, we verified that antagonist CGP64213 displaced with a similar potency the radioligand for every combination examined, indicating that none of the constructs co-expressed with GB1 affect antagonist affinity ( Table 1).
As shown in Fig. 6a, ∆GB2 increases by a factor 16 GABA affinity on GB1 ASA (Ki values were 16.0 ± 1.4 and 1.0 ± 0.2 µM for GB1 ASA expressed alone or with ∆GB2 respectively). This effect is specific since no change in GABA affinity was observed when GB1 ASA was co-expressed with ∆mG5 or ∆GB1(S246A), a GB1 construct unable to bind GABA B ligands (19). Association between GB1 A S A and ∆ GB2 or ∆GB1(S246A) was confirmed by measuring FRET signal between the co-expressed constructs (Fig. 6a, Insert).
These results further confirm ∆GB2 associates with GB1 ASA and show that this interaction increases GABA affinity on GB1.
Of interest, and as previously reported (22), GABA affinity on the isolated VFT of GB1 (either ∆GB1 or GB1 GPI ) is close to that measured when GB1 is co-expressed with GB2 (Ki values 1.0 ± 0.2, 0.7 ± 0.1 and 3.2 ± 0.2 µM, respectively), and 10-20 times higher than that on GB1 ASA (Table 1 and Fig. 6). This indicates that the HD of GB1 exerts an inhibitory action on agonist affinity. GABA affinity on these truncated GB1 constructs can still be further increased, although to a lower extent, after co-expression with either GB2 or ∆GB2 (3 and 2.5 fold, respectively) (Fig. 6b). Indeed, Ki values for GABA on ∆GB1 decreases from 1.0 ± 0.2 µM to 0.3 ± 0.1 µM and 0.4 ± 0.2 in the presence of GB2 and ∆GB2, respectively.
This shows that most of the increased agonist affinity resulting from GB1-GB2 association is due to GB2 VFT preventing the HD of GB1 to decrease agonist affinity, rather than a direct effect of GB2 VFT on the GB1 VFT closure.

DISCUSSION
In this study, we show how the association between the VFTs of GABA B receptor subunits GB1 and GB2 can allostericaly influence agonist affinity. GB2 VFT controls GB1 affinity for GABA in two ways. First, it exerts a direct action on GB1 VFT as GB2 VFT increases slightly agonist affinity on the isolated GB1 VFT. Second and most importantly, GB2 VFT prevents inhibitory interaction between the two main domains (VFT and HD) of the GB1 subunit.

Negative allostery within the GB1 subunit
Inhibitory contacts that maintain GB1 subunit alone in a low affinity state may have several origins. One possibility is that the direct interaction between the HD and the VFT domain within the GB1 subunit contrains the VFT in a low agonist affinity state.
Alternatively, GB1 can exist as homodimer and interactions between the GB1 VFTs in the homodimer could stabilize a low agonist affinity state. We exclude this latter possibility since the isolated VFT of GB1 (both ∆GB1 and GB1 GPI ) adopts a high agonist affinity state even though it is able to homodimerize. Accordingly, a direct interaction between the HD and the VFT domains of GB1 is likely responsible for the low agonist affinity state. Such an interaction has already been proposed by others but has never been demonstrated (22,38).
Such a negative effect of the HD on agonist affinity in the VFT has also been reported for other class-III GPCRs, the mGlu4 and mGlu8 receptors, using a similar approach (39,40).
Negative allostery of GB1 is controlled by GB2 VFT, and such a process is probably not a specificity of the expression of this subunit in heterologous cells. Indeed, only a high agonist affinity GABAB receptor is found in the brain. Moreover, GB1 needs to be associated with GB2 to reach the cell surface in the central nervous system, and neither GB1-GB1 nor GB2-GB2 homodimers couple efficiently to G-proteins (4,6).

How is the agonist affinity controlled?
Both GB1 and GB2 VFTs are structurally related to the VFT of mGlu1 metabotropic receptor and to bacterial periplasmic binding proteins, as suggested by modeling studies (16,17,19,21). Such domains are well-known to adopt either an open conformation (VFTo) stabilized by antagonists (25,26), or a closed conformation (VFTc) stabilized by agonists (24,26,27). VFT can oscillate between VFTo and VFTc states with an equilibrium constant K1= [VFTc]/[VFTo]. A bound ligand will affect K1 by a factor α>1 in the case of an agonist that stabilizes VFTc, and α<1 in the case of an antagonist that stabilizes VFTo or prevents the VFT to reach the VFTc state (28).
Accordingly, ligand affinity in a VFT (Kd) depends on both the affinity of the ligand in VFTo (K L ), K1 and α: According to this model, an increase in K1 (the receptor has a better tendency to reach the closed state in the absence of ligand) results in a large increase in agonist affinity (28).
Moreover, and as observed in the case of the GABA B receptor, changing K1 results in minor changes in antagonist affinity (28). We therefore propose that in the absence of GB2, HD of GB1 decreases K1 (favors the VFTo state), an effect that is prevented when GB2 VFT associates with GB1 VFT. In addition, GB1-GB2 VFTs association may further increase K1 (favors the VFTc state) and thus agonist affinity.

Implications for receptor activation
GB2 VFT and its association with GB1 VFT appear to play a crucial role in GABA B receptor activation. In the absence of GABA, GB2 VFT constrains the receptor in an inactive state whereas in the presence of agonist, it facilitates receptor activation. Indeed, the presence of both GB1 and GB2 VFTs in the dimeric receptor is necessary for agonist stimulation. A dimeric receptor constituted by GB1-GB2 HDs but possessing two identical VFTs (either GB1 or GB2) display a large constitutive activity that cannot be further stimulated by agonists (11).
Inhibitory contacts between HD and VFT in GB1 that controls agonist affinity could be similar to those important for GABA B receptor activation. This model is supported by several pieces of data. First, the fact that GB2 VFT increases agonist but not antagonist affinity suggests that GB2 VFT stabilizes the active state of GB1. Second, we showed GB2 VFT acts by releasing inhibitory contacts between HD and VFT of GB1. Third, positive allosteric compounds of GABA B that likely bind in the HD increase both affinity and efficacy of agonists (29). Finally, similar inhibitory interactions between HD and VFT in GB2 subunit could exist due to the allosteric nature of the GABA B receptor, and these contacts could also play a role in receptor activation.

Conclusion
In conclusion, our experiments emphasize the functional importance of VFT interaction within the GABA B heterodimeric receptor and potentially to other class-III GPCRs. Our data show that the direct interaction between the VFTs is not only important for agonist activation of the receptor, as already reported for both GABA B (11) and mGlu receptors (41), but also for the control of agonist affinity. Indeed, a recent study show that mutations in the mGlu1 VFT that possibly prevents direct VFTs interaction within the receptor dimer largely decrease agonist affinity (41). Perspectives of this work is to identify regions of VFT and HD responsible for inhibitory contacts both at the GB1-GB2 VFTs and HD-VFT interfaces.   (19)). ° values taken from (11). Values are means ± s.e.m. of at least three independent determinations.     In upper panel insert, association between HA-tagged GB1 ASA and c-myc-tagged ∆GB2,