Desensitization of NMDA receptors depends of their association with plasma membrane sodium-calcium exchangers in lipid nanoclasters

The plasma membrane Na+/Ca2+-exchanger (NCX) has recently been shown to regulate Ca2+-dependent N-methyl-d-aspartate receptor (NMDAR) desensitization, suggesting tight interaction of NCXs and NMDARs in lipid nanoclaster or “rafts”. To evaluate possible role of this interaction we studied effects of Li+ on NMDA-elicited whole-cell currents and Ca2+ responses of rat cortical neurons in vitro before and after cholesterol extraction by methyl-β-cyclodextrin (MβCD). Substitution Li+ for Na+ in the external solution caused a concentration-dependent decrease of steady-state NMDAR currents from 440 ± 71 pA to 111 ± 29 pA in 140 mM Na+ and 140 mM Li+, respectively. Li+ inhibition of NMDAR currents disappeared in the absence of Ca2+ in the external solution (Ca2+-free), suggesting that Li+ enhanced Ca2+-dependent NMDAR desensitization. Whereas the cholesterol extraction with MβCD induced NMDAR current decrease to 136 ± 32 pA in 140 mM Na+ and 46 ± 15 pA in 140 mM Li+, the IC50 values for the Li+ inhibition were similar (about 44 mM Li+) before and after this procedure. In Ca2+-free Na+ solution steady-state NMDAR currents after the cholesterol extraction were 47 ± 6 % of control currents. Apparently this amplitude decrease was not Ca2+-dependent. In 1 mM Ca2+ Na+ solution the Ca2+-dependent NMDAR desensitization was greater when cholesterol was extracted. Obviously, this procedure promoted its development. In agreement, Li+ and KB-R7943, an inhibitor of NCX, both considerably reduced NMDAR-mediated Ca2+ responses. The cholesterol extraction itself caused a decrease of NMDAR-mediated Ca2+ responses and, in addition, abolished the effects of Li+ and KB-R7943. Taken together our data suggest that NCXs downregulate the Ca2+-dependent NMDAR desensitization. Most likely, this is determined by co-localization and tight functional interaction of NCX and NMDAR molecules in membrane lipid rafts. Their destruction is accompanied by an enhancement of NMDAR desensitization and a loss of NCX-selective agent effects on NMDARs.

Introduction 49 50 N-methyl-D-aspartate activated glutamate receptors (NMDARs) are ligand gated ion 51 channels which naturally transfer currents determined by Na + , K + and Ca 2+ permeation. High 52 permeability of NMDARs to Ca 2+ makes them involved in synaptic plasticity [1,2], while their 53 hyperactivation during ischemia or stroke causes neuronal Ca 2+ overload and apoptosis [3]. Ca 2+ -54 dependent desensitization of NMDARs represents a feedback regulation of the NMDAR open 55 probability by the Ca 2+ entry into neurons [4][5][6][7][8]. The Ca 2+ entry via NMDAR pores produces a concentration-dependent manner because of Ca 2+ -dependent NMDAR desensitization [9]. 60 Recently it has been demonstrated that the inhibition of the plasma membrane Na + /Ca 2+  The Li + therapy is widely used to stabilize mood disorders, including bipolar disorders 72 and depression as well as suicidal behaviors [12]. There are some experimental indications that 73 KB-R7943 reduces 4-aminopyridine-induced epileptiform activity in adult rats [15]. It is still not 74 clear whether NCXs could represent a target of pharmacological action to compensate NMDAR-75 related neuronal pathologies and whether an acceleration of Ca 2+ -dependent NMDAR 76 desensitization by Li + is at least partially contributed to the Li + therapeutic effects. To provide 77 more clues for understanding of these aspects of the NMDAR pharmacology here we study the 78 concentration dependence of Li + effects on NMDAR currents and the role of functional 79 interaction between NCXs and NMDARs that presumably requires their close spatial localization 80 in lipid rafts. The procedure of culture preparation from rat embryos was previously described [16,17].       Fig 1A and B). The stepwise substitution of Li + for Na + in 183 the external solution after the MβCD treatment further decreased the NMDA-evoked currents to 184 the mean steady-state amplitude of 46.8 ± 15.3 pA (n = 10, p < 0.008, Student's two-tailed t-185 test). The IC 50 value for the Li + inhibition of NMDA-activated currents after the MβCD 186 treatment was 42 ± 20 mM (Fig 1B and C) which did not differ significantly from the value 187 obtained under the MβCD untreated control conditions. It should be noted, however, that the 188 degree of the NMDAR current inhibition in the Li + -containing bathing solution was less 189 pronounced after the MβCD treatment than before this procedure and were 59 ± 4 % (n = 10) 190 and 77 ± 3 % (n = 10) inhibition (p < 0.03, Student's two-tailed t-test), respectively (Fig 1D).  (Fig 2A). 205 In the absence of Ca 2+ in the external solution the ratio of amplitudes of NMDA-activated 206 steady-state currents, recorded after and before 5 min MβCD treatment was 47 ± 6 % (n = 6).

207
The decrease of the NMDAR current steady-state amplitude after the treatment is caused by the 208 direct effect of the cholesterol extraction on NMDARs, because under these particular conditions 209 the Ca 2+ -dependent desensitization was not observed (Fig 2A). In the presence of 1 mM Ca 2+ in 210 the bathing solution, however, the Ca 2+ -dependent desensitization of NMDARs, measured as the 211 ratio of the steady-state current amplitudes in the absence and presence of Ca 2+ before and after 212 the MβCD treatment was significantly greater when cholesterol was extracted (Fig 2A and B), 213 suggesting that this procedure enhanced the Ca 2+ -dependent NMDAR desensitization. In  (Fig 2C). The direct effect of MβCD treatment on NMDARs was pronounced and the 218 ratio values obtained in the presence and absence of external Ca 2+ were similar (Fig 2C and D).

219
In 1 mM intrapipette BAPTA the steady-state NMDAR currents decreased after the extraction to 220 about 10 % of their amplitudes (Fig 2D), whereas in experiments when the intracellular media 221 was natural in terms of Ca 2+ buffering the NMDAR currents decreased in a much smaller extent 222 (about 47 %, Fig 2A).

223
Based on these experiments we may assume that in lipid rafts NCX weakens Ca 2+ -  in the Na + -containing solution were 35 ± 9 % (n = 3, overall 98 neurons) and in the Li + -245 containing solution were 36 ± 9 % (n = 3, overall 98 neurons) of the Ca 2+ responses, obtained 246 before the treatment in the Na + -containing solution (Fig 3A and B). Because these values were 247 significantly smaller, than those obtained before the treatment in the Na + solution (p < 0.0001,   NMDARs and NCXs should co-localize and interact that could be achieved in membrane 291 cholesterol rich nanocluster or lipid rafts (Fig. 4A). Actually the co-localization of NMDARs and 292 NCXs in lipid rafts at the distance less than 80 nm was recently demonstrated using FRET further support our conclusion that the destruction of lipid rafts abolishes the influence of NCXs 318 on NMDARs (Fig. 4B).

319
These experiments allow us to suggest, that the NCX inhibition prevents the 320 maintainance of low Ca 2+ level in the proximity of the intracellular domains of NMDARs by the 321 Ca 2+ extrusion to the outside, which elevates pre-membrane local Ca 2+ concentration, but limits 322 total Ca 2+ entry into neurons. Modeling of interaction between CaM and C-terminal of GluN1 323 subunits of NMDARs reveals that it could occur if these molecules are co-localized within the 324 distance of tens of nanometers [24]. Spatial uncoupling of NCXs and NMDARs weakens the 325 NCX influence on the Ca 2+ -dependent NMDAR desensitization. Thus, the inhibition of NCXs 326 with Li + or KB-R7943 after the disaggregation of molecules within former lipid rafts does not 327 significantly influence the cytoplasmic Ca 2+ accumulation in response to NMDAR activation.