1. Activation and desensitization of N-methyl-D-asparate (NMDA) receptors were studied in large outside-out patches excised from cultured embryonic neurones dissociated from mouse forebrain. The patches were exposed to rapid changes of NMDA or L-glutamate concentrations in the presence of glycine at concentrations (10-20-mu-M) saturating the modulatory site of the NMDA receptor. 2. Immediately after formation of the patch the responses to NMDA and L-glutamate showed a slow and small desensitization, even with high concentrations of agonist. During the following hour, the peak response either decreased or remained relatively stable, but in all cases the desensitization increased and accelerated until it stabilized. In this 'stabilized' state, the desensitization produced by high concentrations of NMDA (1 mM) or L-glutamate (300-mu-M) had an exponential time course, with a time constant of about 30 ms. The ratio of the peak over the steady-state current was in the order of 40 for NMDA and about 30 for L-glutamate. 3. Concentration-response curves were built for the peak and the plateau responses, for NMDA and for L-glutamate. The comparison of these curves indicated that (i) the EC50 of the peak (K(app)) was always higher than the EC50 of the plateau (K(ss)); (ii) the two EC50 values for NMDA (K(app) and K(ss)) were higher than those for L-glutamate; (iii) the Hill coefficient was close to 1.4 for each of the four curves. 4. The application of NMDA or L-glutamate at a low concentration for 3 s periods reduced the response to a subsequent application of the same agonist at a saturating concentration. The IC50 of this 'predesensitization', termed K(pre), was lower than the EC50 of the steady-state response, K(ss). 5. The onset rates of desensitization increased with the concentration of agonist. The EC50 of this relation was close to the value of K(app). 6. The decay of the currents at the end of a 3 s application of agonist was usually well described by the sum of two exponentials both of which were faster for NMDA than for L-glutamate. 7. The recovery from desensitization after a long (3 s) pulse of agonist was approximately exponential, with a time constant of about 0.5 s for NMDA and about 3.5 s for L-glutamate. 8. The results can be interpreted by using a model similar to those used for the nicotinic receptor assuming that (i) the NMDA receptor exists, in the absence of agonist, in both an activatable (R) and two desensitized (D) states; (ii) the R and D states have two equivalent binding sites for NMDA agonists; (iii) the affinity of the R state for the agonist is lower than that of the D states. However, in contrast with the hypotheses made for the nicotinic receptor, it is not possible to assume that the reactions leading to the opening of the channel through activation of the R states are much faster than the reactions connecting the R and D states. 9. The differences between L-glutamate and NMDA responses suggest that L-glutamate differs from NMDA by a higher affinity for the various receptor states as well as by a higher efficacy. 10. The evolution of the NMDA receptor behaviour between the moment at which the patch is formed and the time at which it reaches a stable state suggests that the progressive deepening of desensitization is due to an acceleration of the transition between the doubly liganded R state and the doubly liganded D states. In contrast, both the equilibrium between the non-liganded R and D states and the kinetics of the activation pathway appear relatively stable over time.
