NMDA receptor expression in Xenopus oocytes
To investigate the influence of NMDA receptor subunit composition on the modulatory effects of neuroactive steroids, mRNA coding for the NR1100 subunit was coinjected into Xenopus laevis oocytes along with mRNA coding for either the NR2A, NR2B, NR2C, or NR2D subunit. All four diheteromeric subunit combinations resulted in expression of functional NMDA receptors 1 – 5 days after injection, as indicated by an inward current in response to application of 80 μM NMDA plus 10 μM glycine.
Dependence of PS modulation upon subunit composition
As shown in Figure 1, the choice of NR2 subunit dictates the direction of modulation by PS. Because the NMDA EC50 differs for the various subunit combinations, we compared the modulatory effects of PS using a concentration of NMDA close to its EC50 for each subunit combination (80 μM for NR1/NR2A, 25 μM for NR1/NR2B and NR1/NR2C, and 10 μM for NR1/NR2D). Glycine was present at a saturating concentration (10 μM). In oocytes expressing NR1/NR2A receptors, the NMDA induced current is increased by 62±8% (n=8) in the presence of 100 μM PS. Similarly, with oocytes expressing NR1/NR2B receptors, the NMDA-induced current is enhanced by 78±9% (n=4) in the presence of 100 μM PS. In contrast, NMDA responses of oocytes expressing NR1/NR2C or NR1/NR2D receptors are inhibited by 35±3% (n=4) and 26±1% (n=9), respectively, in the presence of 100 μM PS.
Figure 1. Inverse modulation of NMDA receptor subtypes by PS. (a – d), examples of traces obtained from oocytes previously injected with (a) NR1/NR2A, (b) NR1/NR2B, (c) NR1/NR2C, or (d) NR1/NR2D mRNAs. The bar indicates the period of drug application. Interval between consecutive current traces was 45 s. Receptors were activated by co-application of 10 μM glycine plus 80 μM NMDA (NR1/NR2A), 25 μM NMDA (NR1/NR2B and NR1/NR2C), or 10 μM NMDA (NR1/NR2D). Co-application of 100 μM PS to NR1/NR2A or NR1/NR2B receptors resulted in an increase in the agonist response, whereas co-application of 100 μM PS to NR1/NR2C or NR1/NR2D resulted in a decrease in the agonist response. (e) Concentration-response curves for PS effect on NR1/NR2 receptors. Data points are averaged values of normalized peak current responses from oocytes injected with NR1/NR2A (n=8), NR1/NR2B (n=8), NR1/NR2C (n=4) or NR1/NR2D (n=4) RNAs. Responses were normalized to the control response obtained by application of 10 μM glycine plus 80 μM NMDA (NR2A), 25 μM NMDA (NR2B, NR2C) or 10 μM NMDA (NR2D). Error bars indicate s.e.mean. Smooth curves are calculated from the two-state model (equation 1) using the parameters in Table 3. (f) Effect of holding potential on modulation of the NMDA/glycine response by PS. Points are averaged relative currents obtained in the presence of 100 μM PS, standardized relative to the response induced from the same oocyte by 10 μM glycine plus 80 μM (NR1/NR2A, n=4), 25 μM (NR1/NR2B, n=7; NR1/NR2C, n=3), or 10 μM NMDA (NR1/NR2D, n=3). Symbols are defined as in e. Error bars indicate s.e.mean.
Download figure to PowerPoint
As shown in Figure 1e, PS is about equally potent in potentiating NR1/NR2A and NR1/NR2B receptors, and 3.4 to 5.6 fold less potent as an inhibitor of NR1/NR2C and NR1/NR2D receptors (Table 1). Enhancement of NR1/NR2A and NR1/NR2B receptors and inhibition of NR1/NR2C and NR1/NR2D receptors exhibits little if any voltage dependence (Figure 1f).
Table 1. Concentration dependence of PS modulation of the NMDA response
To determine how PS enhances or inhibits the response of the NMDA receptor, the glutamate, NMDA, and glycine concentration-response curves were determined in the presence and absence of PS. As shown in Figures 2 and 3, the nature of the modulatory effect of PS depends not only upon subunit composition, but also upon the specific agonist used. With NR1/NR2A receptors, PS enhances the efficacy of NMDA, glutamate (Figure 2a) and glycine (Figure 3a). At NR1/NR2B receptors, however, PS primarily enhances the efficacy of NMDA, but primarily enhances the potency of glutamate (Figure 2b) and glycine (Figure 3b).
Figure 2. The choice of NR2 subunit determines the direction of PS modulation of the glutamate and NMDA concentration-response curves. Data points are averaged normalized peak NMDA-induced current responses obtained from oocytes injected with (a) NR1/NR2A, (b) NR1/NR2B, (c) NR1/NR2C, or (d) NR1/NR2D mRNAs. Concentration-response data for NMDA and for L-glutamate were obtained in the presence of 10 μM glycine. The data were normalized relative to the current response from the same oocyte induced by co-application of 200 μM NMDA and 10 μM glycine. Error bars represent s.e.mean. Smooth curves are calculated from equation (1) using the parameters in Table 3.
Download figure to PowerPoint
Figure 3. The choice of NR2 subunit determines the direction of PS modulation of the glycine concentration-response curve. Data points are averaged normalized peak current responses obtained from oocytes injected with (a) NR1/NR2A, (b) NR1/NR2B, (c) NR1/NR2C, or (d) NR1/NR2D mRNAs. Concentration-response data for glycine were obtained in the presence of 10 μML-glutamate and in the absence and presence of 100 μM PS. The data for each oocyte were normalized relative to the current response induced by co-application of 200 μM NMDA plus 10 μM glycine. Error bars represent s.e.mean. Smooth curves are calculated from equation (1) using the parameters in Table 3.
Download figure to PowerPoint
Negative modulation by PS of NR1/NR2C and NR1/NR2D receptor activation is a consequence of a decrease in the efficacies of glutamate, NMDA (Figure 2c,d), and glycine (Figure 3c,d). Agonist potencies are not decreased, indicating that PS does not compete for either the glutamate or glycine recognition sites.
Inhibitory potency of 3α5βS depends upon the NR2 subunit
As shown in Figure 4a – d, 100 μM 3α5βS reversibly inhibits NMDA-induced currents of Xenopus oocytes expressing NR1/NR2A (Figure 4a), NR1/NR2B (Figure 4b), NR1/NR2C (Figure 4c), or NR1/NR2D (Figure 4d) receptors. However, the extent of inhibition is significantly greater with NR1/NR2C and NR1/NR2D receptors than with NR1/NR2A and NR1/NR2B receptors (P<0.01, ANOVA with Scheffés post-hoc test). Concentration-response analysis (Figure 4e) indicates that this difference is primarily due to an approximately 4 fold lower potency of 3α5βS at NR1/NR2A and NR1/NR2B receptors than at NR1/NR2C and NR1/NR2D receptors (see Table 2 for EC50s). Inhibition of the NMDA induced current by 3α5βS exhibits little if any voltage dependence from −100 to +20 mV (Figure 4f).
Figure 4. The choice of NR2 subunit influences 3α5βS inhibition of the NMDA response. a – d, examples of traces obtained from oocytes previously injected with NR1/NR2A, NR1/NR2B, NR1/NR2C, or NR1/NR2D mRNAs, respectively. The bar indicates the period of drug application. Interval between consecutive current traces was 45 s. The receptors were activated by co-application of 10 μM glycine plus 80 μM NMDA (NR1/NR2A, a), 25 μM NMDA (NR1/NR2B, b and NR1/NR2C, c), or 10 μM NMDA (NR1/NR2D, d). Typical results are shown; mean inhibition was 53±5% (n=3) for NR1/NR2A, 58±3% (n=3) for NR1/NR2B, 97±2% (n=6) for NR1/NR2C, and 83±3% (n=3) for NR1/NR2D. e, concentration-response curves for 3α5βS effect on NR1/NR2 receptors. Data points are averaged values of normalized steady-state current responses from oocytes injected with NR1/NR2A (n=4), NR1/NR2B (n=3), NR1/NR2C (n=6) or NR1/NR2D (n=4) RNAs. Current responses are expressed relative to the current response in the absence of 3α5βS. Error bars represent s.e.mean. Smooth curves are derived from fits to the logistic equation. f, dependence of 3α5βS effect on membrane potential. Points are averaged relative current obtained in the presence of 100 μM 3α5βS. (NR1/NR2A, n=5; NR1/NR2B, n=10) or 10 μM 3α5βS (NR1/NR2C, n=4; NR1/NR2D, n=10).
Download figure to PowerPoint
Table 2. Concentration dependence of 3α5βS modulation of the NMDA plus glycine response
To determine how 3α5βS inhibits the glutamate response, concentration-response curves were constructed for glutamate (in the presence of 10 μM glycine) and glycine (in the presence 10 μM glutamate) in the presence and absence of 100 μM 3α5βS. As shown in Figure 5, 3α5βS decreases the efficacy with which glutamate and glycine activate NR1/NR2A (Figure 5a,b), NR1/NR2B (Figure 5c,d), NR1/NR2C (Figure 5e,f), and NR1/NR2D (Figure 5g,h) receptors.
Figure 5. Effect of 3α5βS on glutamate (a, c, e, g) and glycine (b, d, f, h) concentration-response curve of oocytes expressing NR1/NR2A (a, b), NR1/NR2B (c, d), NR1/NR2C (e, f) or NR1/NR2D (g, h) subunits. Glutamate concentration-response data was obtained in the presence of 10 μM glycine and in the absence or presence of 100 μM 3α5βS. Glycine concentration-response data was obtained in the presence of 10 μM glutamate and in the absence or presence of 100 μM 3α5βS. Data points are averaged normalized peak current responses of three to seven oocytes. Smooth curves are fits to the logistic equation. The data for each oocyte were normalized to standard current responses induced by co-application of 200 μM NMDA and 10 μM glycine. Concentration-response data for glutamate and glycine alone is the same as in Figure 2, and is repeated for comparison.
Download figure to PowerPoint
Modelling the interaction of PS with the NMDA receptor
It seems unlikely that the fundamental mechanism of action of PS would be different for receptors of different subunit composition, or that PS would have different mechanisms of enhancing NMDA and glutamate responses of NR1/NR2B receptors. We therefore considered whether a single allosteric model could accommodate the different effects of PS on agonist concentration-response curves for the four types of NMDA receptors tested. The simplest possible allosteric model is the two-state model, in which a receptor is assumed to exist in either an inactive (closed) or an active (open) conformation, with each conformation having its own characteristic affinity for ligands (Karlin, 1967; Monod et al., 1965). In a two-state model, the efficacy of an agonist is dependent upon the ratio (K′agonistKagonist−1) of its affinities for the active and inactive states, and upon the gating equilibrium constant, M=[R′] [R]−1, which is the ratio of active to inactive receptors in the absence of agonist (Colquhoun, 1998).
Evidence suggests that the NMDA receptor is tetrameric, with two sites for glutamate/NMDA and two sites for glycine (Clements & Westbrook, 1991). We modelled the receptor on this basis, adding two additional sites for PS (two sites seeming more likely than one on the basis of symmetry) (Figure 6). Activation is treated as concerted, with all subunits activating or deactivating simultaneously. Because little if any desensitization is observed in our experiments (see Figure 1), a desensitized state is not included. This model entails a total of 10 parameters: the dissociation constants for binding of each of the four ligands to the active and inactive states of the receptor, the resting ratio of active to inactive receptors, and a scaling factor related to the number of receptors (the current if all receptors were simultaneously active, expressed relative to the 200 μM NMDA response).
Figure 6. Allosteric model of NMDA receptor modulation by PS. Activation of the receptor (gating) is assumed to be concerted and described by a two-state model. The model includes six binding sites, two each for glutamate/NMDA, glycine, and PS. High affinity is indicated by a deep ‘slot’ for the corresponding ligand, while low affinity is indicated by a shallow slot. Affinity of the active state for PS is greater than that of the resting state for NR1/NR2A and NR1/NR2B, but less than that of the resting state for NR1/NR2C and NR1/NR2D.
Download figure to PowerPoint
For each subunit combination, the concentration-response data for all agonists in the presence and absence of PS were simultaneously fitted to the two-state model (equation 1). The two state allosteric model readily accommodates the co-agonist interaction between glutamate/NMDA and glycine, which arises because each of the co-agonists individually has very low efficacy. Simultaneous binding of the co-agonists to the glutamate and glycine sites results in a synergistic interaction, such that their combined efficacy is much greater than the sum of their individual efficacies.
As shown in Figures 1, 2 and 3, in which the smooth curves are calculated from the two-state model using the fitted parameters (Table 3), this model also produced a good fit to the data for the effects of PS on all four subunit combinations. The model provides an explanation for the different effects of PS on the agonist concentration-response curves for NR1/NR2A and NR1/NR2B receptors. In this type of model, an allosteric modulator is expected to influence both agonist potency and agonist efficacy, but one effect on the other may predominate. If an agonist has high efficacy to begin with, such that it is capable of activating nearly all receptors, then a positive allosteric modulator is predicted to primarily enhance the potency of that agonist, shifting its concentration-response curve to the left. In contrast, a positive allosteric modulator will primarily enhance the efficacy of a partial agonist.
Table 3. Parameters from fits of the two-state model to concentration response date in the presence and absence of PS
PS increases the efficacy of glutamate and glycine at NR1/NR2A receptors, but primarily increases the potency of glutamate and glycine at NR1/NR2B receptors, suggesting that glutamate and glycine have lower efficacy at NR1/NR2A than at NR1/NR2B receptors. As calculated from the fitted parameters (Table 3), saturating glutamate and glycine activate only 16% of NR1/NR2A receptors, but 88% of NR1/NR2B receptors. In contrast, NMDA has low efficacy at both NR1/NR2A and NR1/NR2B receptors, producing (at saturating glycine) 13 and 62% maximal activation, respectively, so PS enhances the efficacy of NMDA in both cases.
Conversely, the two-state model predicts that a negative allosteric modulator will primarily decrease the potency of a high-efficacy agonist, while decreasing the efficacy of a partial agonist. In the case of NR1/NR2C and NR1/NR2D receptors, the major effect of PS is a reduction in the efficacies of glutamate, NMDA, and glycine, suggesting that all of these agonists are relatively inefficient in activating NR1/NR2C and NR1/NR2D receptors. However, while the model produced a good fit to the data, it was not possible to obtain a unique set of parameter estimates (i.e. more than one set of values produced a good fit) for the NR1/NR2C or NR1/NR2D combinations, indicating that the information in the concentration-response data is not adequate to fully define all 10 parameters for the inhibitory effects of PS on these subunit combinations. Because the fits for NR1/NR2A and NR1/NR2B both yielded estimates of about 7×10−5 for M (the resting ratio of active to inactive receptors), the concentration response data for NR1/NRC and NR1/NR2D were fit with M fixed to this value, thereby reducing the number of free parameters (Table 3).
In contrast to the results obtained with PS, the two-state model was not able to adequately fit the 3α5βS concentration-response data for any subunit combination, suggesting that the mechanism of action of 3α5βS is different from that of PS.