Do N-arachidonyl-glycine (NA-glycine) and 2-arachidonoyl glycerol (2-AG) share mode of action and the binding site on the β₂ subunit of GABA_A receptors?

Introduction

GABA is the major inhibitory neurotransmitter in mammalian brain. Its fast effects are mediated by synaptic and extrasynaptic GABA_A receptors. These receptors are composed of five subunits that surround a central chloride ion channel (Macdonald & Olsen, 1994; Sieghart, 1995; Sieghart & Sperk, 2002; Sigel & Steinmann, 2012). The major receptor isoform consists of α₁, β₂, and γ₂ subunits (Olsen & Sieghart, 2008). Different approaches led to the widely accepted 2α:2β:1γ subunit stoichiometry (Chang et al., 1996; Tretter et al., 1997; Farrar et al., 1999; Baumann, Baur & Sigel, 2001; Baumann, Baur & Sigel, 2002; Baur, Minier & Sigel, 2006) with a subunit arrangement γβαβα anti-clockwise as seen from the synaptic cleft (Baumann, Baur & Sigel, 2001; Baumann, Baur & Sigel, 2002; Baur, Minier & Sigel, 2006). The pharmacological properties are dependent on subunit composition (Sigel et al., 1990) and arrangement (Minier & Sigel, 2004).

Neurosteroids (Belelli & Lambert, 2005) and the endocannabinoid 2-AG (Sigel et al., 2011) have been documented as endogenous ligands of GABA_A receptors. A binding site for 2-AG has been shown to be present specifically on each of the two β₂ subunits present in a pentameric receptor. Moreover it has been proposed that 2-AG dips into the membrane and binds to the fourth trans-membrane sequence (M4) of β₂ subunits (Baur et al., 2013).

NA-glycine has been shown to be present in amounts of about 50 pmol/g dry weight in rat brain tissue and about 140 pmol/g dry weight in spinal cord (Huang et al., 2001). This can be compared to the levels of 2-AG that have been reported to be 4 and 50 nmol/g wet weight in brain (Sigel et al., 2011) and spinal cord (Guasti et al., 2009), respectively. Thus, NA-glycine is about 1000-fold less abundant than 2-AG. NA-glycine has no functional affinity for CB₁ receptors (Sheskin et al., 1997), but may activate the G protein coupled receptors GPR18 (Kohno et al., 2006) and GPR92 (Oh et al., 2008) and target Na⁺/Ca²⁺-exchanger NCX (Bondarenko et al., 2013) and T-type Ca²⁺-channels (Barbara et al., 2009). Interestingly, NA-glycine exerts analgesic properties in different rodent models of pain (Huang et al., 2001; Succar, Mitchell & Vaughan, 2007). Recently, it has been reported that NA-glycine can also interact with glycine receptors, potentiating α₁ and interestingly inhibiting α₂ and α₃-containing glycine receptors (Yevenes & Zeilhofer, 2011). It should be noted that NA-glycine functionally interacts with glycine receptors with an apparent affinity >10 µM. This should be compared to the functional affinities to GABA_A receptors, NCX and T-type Ca²⁺-channels that are estimated <10 µM.

We have previously demonstrated that NA-glycine allosterically potentiates GABA_A receptors (Baur et al., 2013), but it remained unclear whether this occurred via the same binding site as 2-AG. The aim of the present study was therefore to establish whether NA-glycine acts at the identical site on GABA_A receptors as 2-AG. In most studied aspects Na-glycine acts similar to 2-AG, indicating a common binding site. However, some point mutations that abrogate modulation by 2-AG leave the initial modulation by NA-glycine nearly unaffected and only over time the modulation is gradually decreased to zero. Differential solubilization effects of 2-AG and NA-glycine may account for this phenomena. Thus, at least during initial phases of interaction with GABA_A receptors the mode of binding is different for the two compounds.

Methods

Results

Both NA-glycine and 2-AG allosterically potentiate recombinant α₁β₂γ₂ GABA_A receptors expressed in Xenopus oocytes. Both compounds share the arachidonoyl tail structure but differ in their hydrophilic head groups (Fig. 1). Please note that at physiological pH, NA-glycine is negatively charged. Here we wanted to compare the GABA_A receptor binding site for NA-glycine with the well-characterized binding site for 2-AG.

Figure 2A shows current traces of a cumulative concentration–response curve of the allosteric potentiation of α₁β₂γ₂ GABA_A receptors at a GABA concentration of 1 µM. At the highest concentration used the current trace displayed the typical signs of an open channel block, rapid apparent desensitization and an off current. This phenomenon made a precise curve fit impossible as maximal potentiation could not be determined precisely. The averaged concentration–response curve (Fig. 2B) was fitted with the assumption of different maximal potentiation. From these fits it was estimated that the EC₅₀ was between 1 and 10 µM (not shown). Direct activation by 3 µM NA-glycine elicited no significant current (<2 nA) in oocytes where 100 µM GABA elicited a current >7 µA.

Allosteric potentiation by 3 µM NA-glycine was determined at different concentrations of the endogenous agonist GABA. Figure 3 shows that the degree of potentiation was rapidly decreasing with increasing concentrations of GABA. The comparable properties of 2-AG are also shown in Fig. 3. We tried to rationalize these findings using a model that has previously been proposed on the basis of other observations (Baumann, Baur & Sigel, 2003; Fig. 4A). This model assumes binding of GABA to two sites differing in their binding affinity and transition to the open state with low propensity of singly ligated states and high propensity of the doubly ligated state. Figure 4B shows computed current amplitudes in dependence of the GABA concentration. In addition, a predicted curve is shown where it is assumed that NA-glycine promotes transition of the singly ligated receptor from the closed to the open state. Figure 4C shows a computed GABA concentration-dependence of the current potentiation expected in this case. The model predicts that sizeable potentiation is limited to very low concentrations of GABA.

We have previously shown that the CB₁ receptor ligand DEA antagonizes potentiation by 2-AG (Baur et al., 2013). Therefore, we compared the ability of DEA to antagonize potentiation by NA-glycine and 2-AG. Based on the structural similarity of the three compounds we assumed a competitive behaviour. Figure 5 compares the cumulative concentration inhibition curves for 3 µM NA-glycine and for 3 µM 2-AG. Potentiation by NA-glycine was inhibited half-maximally at 72 ± 36 µM (n = 4) DEA and potentiation by 2-AG at 1.4 ± 0.6 µM (n = 6). If the two ligands displayed a similar apparent affinity for potentiation at the same site where DEA acts, providing equal water solubility and lipid solubilization, a similar inhibitory potency of DEA would have been expected. In order to investigate if DEA and NA-glycine act competitively, we repeated a concentration inhibition curves at 6-times lower concentration (0.5 µM) of NA-glycine. Half-maximal inhibition was observed at 96 ± 41 µM (n = 4) DEA (Fig. 5). This could be interpreted as non-competitive interaction of DEA with NA-glycine. Similarly, we performed concentration inhibition curves at 1 and 15 µM 2-AG. The higher concentration of 2-AG elicits direct current that amount to less than 0.1% of the maximal current amplitude elicited by GABA in the same oocytes. As expected for a competitive interaction between 2-AG and DEA, half-maximal inhibition was shifted to the left upon decrease of the 2-AG concentration from 3 µM to 1 µM, but the inhibition curve became much flatter, indicating the DEA becomes partially inactive at higher concentrations. Half-maximal inhibition for 15 µM was not reached at concentrations up to 100 µM (n = 4) DEA (Fig. 5). In case of a non-competitive interaction of DEA with 2-AG an IC₅₀ of about 1.4 µM and in case of a competitive interaction an IC₅₀ of about 7 µM would be expected in the latter case. The observed results cannot be explained by classical receptor theory and we therefore speculate that 2-AG and NA-glycine exhibit a different water solubility and lipid solubilisation in the experimental setup. In the discussion we mention possible explanations.

NA-glycine shows higher efficacy than 2-AG for potentiation of currents elicited by GABA. 2-AG is metabolically stable in Xenopus oocytes as no degradation by serine hydrolases was found (not shown). In case NA-glycine competes for the same binding site as 2-AG and both molecules have a similar apparent affinity to this binding site, it would be anticipated that the degree of potentiation by both agents at the same concentration would result in an intermediate potentiation as compared to the individual agents. The apparent affinity of 2-AG has been determined as 2 µM, while the apparent affinity of NA-glycine is estimated 1–10 µM here. As shown in Fig. 6A, combined application results surprisingly in nearly the same extent of potentiation as application of NA-glycine alone.

Again, this may be caused by a differential water solubility and membrane solubilisation behaviour of NA-glycine and 2-AG. We therefore measured the critical micelle concentrations of both molecules. The apparent CMC was >100 µM for NA-glycine and 4.2 ± 0.5 µM for 2-AG, pointing to significant self-assembly and detergent behaviour of 2-AG.

As 2-AG fails to potentiate in GABA_A receptors where the β₂ subunit is replaced by β₁, we tested potentiation by NA-glycine in α₁β₁γ₂ receptors (Fig. 6B). Similarly to 2-AG, potentiation by NA-glycine depends on the presence of β₂ subunits. We studied potentiation by NA-glycine in concatenated receptors containing either two β₂ subunits, two β₁ subunits or one each β₁ and β₂ in different positions in the receptor pentamer (Fig. 7). Receptors containing two β₂ subunits exhibited strong potentiation while receptors containing two β₁ subunits showed very weak potentiation. Intermediate potentiation was observed in receptors containing one each, β₁ and β₂. This strongly indicates that the NA-glycine binding site is located on the β₂ subunit as previously shown with 2-AG (Sigel et al., 2011).

A number of point mutations have been described to interfere with the potentiation by 2-AG. We tested the effect of the point mutations β₂W428C, β₂S429C, β₂F432C, β₂F439L and β₂V443C. Original current traces are shown for the mutant receptor α₁β₂S429Cγ₂. These traces are compared with traces from wild type receptors (Fig. 8A). While wild type receptors show a time-independent potentiation by NA-glycine, mutant receptors showed initially a potentiation that rapidly decayed over time. As these mutant receptors show a similar dependence on GABA as wild type receptors, and the experiment were carried out at very low GABA concentrations this current transient is not due to desensitization. In the case of 2-AG the effect of these mutations is a reduction of the potentiation independent of the time of exposure to 2-AG. This behaviour is observed with NA-glycine for the potentiation of α₁β₂F432Cγ₂ receptors, but not the other mutant receptors studied.

The mutation studies indicate a site of action in the inner leaflet of M4 of the β₂ subunit. In this case NA-glycine has to traverse the lipid bilayer either by diffusion or mediated by a transport system and this may require some time. In order to test the time-dependence of action of NA-glycine we exposed an oocyte to GABA followed by GABA and NA-glycine (Fig. 9). Indeed, onset of modulation was slow and did not reach a steady level within 1 min. Upon switch of the medium to GABA only, a slow decay of the potentiation was observed.

Discussion

NA-glycine allosterically potentiates GABA_A receptors like the major endocannabinoid 2-AG. We aimed at localizing the site of interaction of Na-glycine with recombinant α₁β₂γ₂ GABA_A receptors relative to the site for 2-AG. An interpretation of our results is hampered by the fact that the apparent affinity for the potentiation by NA-glycine could not be determined accurately. However, we can estimate this value to be in the range of 1–10 µM, which compares well with the value of 2 µM for 2-AG (Sigel et al., 2011). The fact that we find significant potentiation of GABA_A receptors by >0.1 µM NA-glycine may reflect the better water solubility of NA-glycine over 2-AG at low concentrations and even indicate biological relevance as the average in vivo concentration in the central nervous system may be estimated from the dry tissue content as about 15–50 nM and NA-glycine is unlikely to be randomly distributed.

The following observations argue for a similar mode of action of NA-glycine and 2-AG. First, both substances only act exclusively at low GABA concentration (Fig. 3), putatively by enhancing the opening of singly ligated receptor channels (Fig. 4). A leftward shift of the concentration response curve for GABA as observed in the case of benzodiazepines does not abrogate potentiation below EC₅₀ (Sigel & Steinmann, 2012). To our knowledge, this is a new mode of action of a ligand. Second, investigation of receptors with different β subunits (Fig. 5) and experiments with concatenated receptors containing either no, one, or two β₂ subunits (Fig. 6) strongly indicate that both ligand binding sites are located on the β₂ subunit. A common binding site in the inner leaflet of the fourth trans-membrane region (M4) of this subunit is suggested by the fact that modulation by both agents is either reduced or abolished in five identical mutant receptors, at least in the late phase of action of NA-glycine (Fig. 8B). The onset of action for both substances was found to be slow (Fig. 9; Baur et al., 2013). On the basis of these observations, it is tempting to assume a common binding site for the two ligands.

The following observation cannot be explained by classical receptor theory in case NA-glycine and 2-AG use an identical binding site, i.e., display a similar apparent affinity and interact with each other in a competitive way. Combined application of two compounds with similar affinities at identical concentrations is then expected to result in an intermediate potentiation as compared to that by individual compounds. Instead, the observed potentiation is similar to the one by NA-glycine alone. A second observation is difficult to reconcile with a common binding site for NA-glycine, 2-AG and the inhibitor of the potentiation of 2-AG, DEA. Namely, DEA prevents potentiation by NA-glycine only at 50-fold higher concentrations as that caused by 2-AG. As mentioned in the result section, the interaction between DEA and 2-AG cannot be explained by classical receptor theory. Since NA-glycine exerts a significant higher CMC than 2-AG, differential solubilisation of NA-glycine and 2-AG with Xenopus oocytes may account for some of the effects observed in this study. The way these lipids are organized in an aqueous environment will affect entry of the molecules into the bilayer, binding equilibrium, and the way the receptor is occupied. If this holds true the observations with co-application of NA-glycine and 2-AG as well as the inhibition of NA-glycine and DEA have to be seen in a new light. In spite of our observations the three agents could still all bind to largely overlapping sites within an extended surface able to bind flexible hydrophobic structures.

The mutant receptors α₁β₂W428Cγ₂, α₁β₂S429Cγ₂, α₁β₂F439Lγ₂ and α₁β₂V443Cγ₂ all largely abrogate modulation by NA-glycine after 1 min of combined application of GABA with NA-glycine. This abrogation is not present at the beginning of the combined application, but sets in rather slowly. We have no explanation for this observation. Solubility considerations do not help to explain this phenomenon.

Overall, most arguments point to a similar action and possibly overlapping binding site for NA-glycine and 2-AG. No matter what the exact mode of interaction of NA-glycine with the GABA_A receptor is, this agent represents by far the stronger positive allosteric modulator than 2-AG, although the latter is more abundant in brain. The implications of our findings for the analgesic effect of NA-glycine remain to be studied.

Abbreviations

GABA

γ-aminobutyric acid

GABA_A receptor

γ-aminobutyric acid type A receptor

NA-glycine

N-arachidonyl-glycine

2-AG

2-arachidonoyl glycerol

DEA

docosatetraenylethanolamide