NMDA Channel Regulation by Channel-Associated Protein Tyrosine Kinase Src

See allHide authors and affiliations

Science  31 Jan 1997:
Vol. 275, Issue 5300, pp. 674-678
DOI: 10.1126/science.275.5300.674


The N-methyl-D-aspartate (NMDA) receptor mediates synaptic transmission and plasticity in the central nervous system (CNS) and is regulated by tyrosine phosphorylation. In membrane patches excised from mammalian central neurons, the endogenous tyrosine kinase Src was shown to regulate the activity of NMDA channels. The action of Src required a sequence [Src(40–58)] within the noncatalytic, unique domain of Src. In addition, Src coprecipitated with NMDA receptor proteins. Finally, endogenous Src regulated the function of NMDA receptors at synapses. Thus, NMDA receptor regulation by Src may be important in development, plasticity, and pathology in the CNS.

The NMDA receptor is a principal subtype of ionotropic excitatory amino acid receptor that plays a central role in development, neuroplasticity, and excitotoxicity in the CNS (1). The function of NMDA receptor is regulated by protein phosphorylation at serine or threonine (2) and at tyrosine (3, 4) residues. For serine or threonine kinases, protein kinase C is an endogenous kinase (5) that regulates NMDA channel function. In contrast, the endogenous tyrosine kinase that regulates NMDA channels has been elusive. Numerous receptor (6), as well as nonreceptor (7), tyrosine kinases are expressed in the CNS. We set out to identify the endogenous tyrosine kinase regulating NMDA channel function.

We recorded NMDA receptor single-channel currents using inside-out patches taken from cultured rat central neurons (8). To investigate whether the endogenous tyrosine kinase was a nonreceptor tyrosine kinase in the Src family, we made use of a high-affinity peptide, EPQ(pY)EEIPIA (9), which activates this family of kinases. The nonphosphorylated form of the peptide, EPQYEEIPIA, is inactive (9) and was used as a control. Application of EPQ(pY)- EEIPIA to the cytoplasmic side of the membrane produced an increase in the channel activity (10) while having no effect on single-channel conductance (Fig. 1A). On average, channel open probability (Po) increased to 260 ± 38% of the control value (mean ± SEM; n = 7 patches), and there was an increase in mean open time to 152 ± 18%. Peptide EPQ(pY)EEIPIA caused marked changes in the distribution of open and shut times (Fig. 1A), with alterations in the relative area of the components and no significant changes in the values of the time constants. In the distribution of open times, the area of the longest component was increased. For shut times, the area of the shortest component increased and that of the longest component decreased. Peptide EPQ(pY)EEIPIA also produced a significant increase in the duration of the bursts, clusters, and superclusters (Fig. 1B) and in the numbers of openings and mean total open time during the groupings (11). In contrast, control peptide EPQYEEIPIA had no significant effect on any of the parameters measured (Fig. 1B) (n = 5 patches). Omitting adenosine 5′-triphosphate (ATP) prevented the effects of EPQ(pY)EEIPIA, consistent with an action due to phosphorylation (Fig. 1B).

Fig. 1.

Modulation of NMDA channel activity by a peptide activator or an antibody inhibitor of Src family kinases. (Aa) A continuous record of NMDA channel open probability (Po) before and during cytoplasmic application of EPQ(pY)EEIPIA (1 mM; bar). NMDA channel Po was calculated in bins 10 s in duration. (Ab) Single-channel currents before (Control; upper trace) and during (lower trace) application of EPQ(pY)EEIPIA. (Ac) Current-voltage (I-V) relation for mean NMDA single-channel currents before (circles) and during (triangles) application of EPQ(pY)EEIPIA. (Ad) Dwell-time histograms of open and shut times before (upper panels) and during (lower panels) application of EPQ(pY)EEIPIA. In this and all other dwell-time histograms the dashed lines indicate individual exponential components, the solid line shows the sum of the components, and n is the number of events. The average time constants of the components (±SEM) were 0.12 ± 0.05, 0.98 ± 0.45, and 3.0 ± 0.9 ms (open times) and 0.15 ± 0.02, 1.7 ± 0.27, 26 ± 4.0, 380 ± 86, and 1700 ± 540 ms (shut times), before peptide administration (n = 7 patches). (B) Effect of EPQ(pY)EEIPIA (n = 7), EPQYEEIPIA (1 mM; n = 5 patches), and EPQ(pY)EEIPIA, in the absence of ATP (n = 5 patches), on NMDA channel activity. Values are the means ± SEM. The dashed line indicates control values before peptide application [Po and mean open time = (3.7 ± 0.9) × 10−3 and 2.0 ± 0.5 ms, respectively, and the duration of the bursts, clusters, and superclusters = 2.7 ± 0.8, 9.1 ± 1.7, and 203 ± 41 ms, respectively]. to, mean open time; B, burst length; C, cluster length; S, supercluster length. (C) A continuous record of the NMDA channel Po before and during application of anti-cst1 (10 μg/ml; bar). (D) Effect of anti-cst1 (n = 6 patches) or the nonselective IgG fraction (10 μg/ml; n = 4 patches) on activity of NMDA channels. *P < 0.05 [Mann-Whitney test for EPQ(pY)EEIPIA versus EPQYEEIPIA, or anti-cst1 versus the nonselective IgG fraction].

In a complementary series of experiments we used an antibody, anti-cst1 (12), which inhibits Src family kinases (13) (Fig. 1, C and D). Anti-cst1 (10 μg/ml; n = 6 patches) produced significant decreases in Po, mean open time, and burst and cluster duration (Fig. 1D). In addition, the numbers of openings and mean total open time in all of the groupings were decreased (11). In contrast, application of nonselective immunoglobulin G (IgG) fraction (10 μg/ml) had no significant effect (n = 4 patches). The effects of EPQ(pY)EEIPIA and anti-cst1 implied that a Src family kinase was present in the membrane patches and regulated NMDA channel gating.

To determine whether the endogenous kinase was Src itself, we studied the effects of anti-src1 (Fig. 2, A and B) (n = 6 patches), which blocks selectively the function of this kinase (13). Administration of anti-src1 decreased NMDA channel gating with no change in single-channel conductance: Po, mean open time, and burst and cluster length were significantly reduced, and the open-time distribution was shifted toward shorter openings and the shut times toward longer closings. The effects of anti-src1 were prevented when it was incubated with the corresponding immunogen peptide, Src(40–58), for 30 min just before use (n = 3 patches). Moreover, EPQ(pY)EEIPIA had no effect when applied to patches that had first been treated with anti-src1 (Fig. 2E) (n = 5 patches). In addition, application of recombinant pp60c-src (1 U/ml; n = 6 patches) increased Po to 257 ± 100% of control, mean open time to 125 ± 10%, burst length to 186 ± 23%, and cluster length to 220 ± 33%. These effects were prevented by boiling pp60c-src to heat-inactivate it just before use (n = 4 patches). Together, these results indicated that NMDA channels were regulated by endogenous Src.

Fig. 2.

Regulation of NMDA channel gating by endogenous Src. (Aa) A continuous record of NMDA channel Po in the absence or presence of anti-src1 (1:100 dilution; bar). (Ab) Dwell-time histograms of NMDA channel open and shut times before (upper panels) or during (lower panel) application of anti-src1. (B) Effect of anti-src1 (mean ± SEM; n = 7 patches) or anti-src1 after incubation with Src(40–58) (n = 3 patches). (C) A continuous record of NMDA channel Po in the absence or presence of Src(40–58) (0.03 mg/ml; bar). (D) Effect of Src(40–58) (n = 7 patches) or sSrc(40–58) (0.03 mg/ml; n = 5 patches). (E) Effect of EPQ(pY) EEIPIA alone, EPQ(pY)EEIPIA with anti-src1 (n = 5 patches), or EPQ(pY)EEIPIA with Src(40–58) (n = 4 patches) on NMDA channel activity. *P < 0.05 and **P < 0.01 (Mann-Whitney test).

It is possible that anti-src1 acted by preventing an interaction between the region in Src specific for the antibody and a complementary region in a target protein. To investigate this possibility, we studied the effects of Src(40–58) (Fig. 2, C and D). Po, mean open time, and duration, number of openings, and mean total open time during bursts and clusters were reduced by application of Src(40–58) (n = 7 patches). To control for length, net charge, and amino acid composition, we tested a peptide in which the amino acid sequence of Src(40–58) was scrambled (8). Scrambled Src(40–58) [sSrc(40–58)] had no effect on single-channel activity (Fig. 2D) (n = 5 patches). In other experiments, initial treatment with Src(40–58) prevented the effects of EPQ(pY)EEIPIA (Fig. 2E), but these effects persisted with sSrc(40–58) (n = 3 patches). Thus, an interaction between the region of Src(40–58), which is within the unique domain of Src, and another protein may be necessary for the effects of Src on NMDA channels.

To determine whether Src and NMDA channels are associated physically, we immunoprecipitated membrane proteins (14) with antibodies specific for the NR1 subunit of NMDA receptors (anti-NR1) or for Src (anti-Src). When we used nondenaturing conditions (15) to solubilize membrane proteins, immunoprecipation with anti-NR1 resulted in coprecipitation of Src (Fig. 3A). Conversely, immunoprecipitation with anti-Src resulted in coprecipitation of NR1 (Fig. 3B). In contrast, the K+ channel protein Kv3.1 (16) was not immunoprecipitated either by anti-Src or anti-NR1, indicating that precipitation of membrane proteins did not occur nonspecifically. The coprecipitation of Src by anti-NR1 and of NR1 by anti-Src was prevented when denaturing solubilization conditions were used, whereas the immunoprecipitation of NR1 or Src by the corresponding antibodies was not affected. Moreover, neither NR1 nor Src was precipitated by nonspecific IgG. Thus, Src may have been physically associated with NMDA channels in vitro.

Fig. 3.

Association of Src and NMDA channel proteins. Immunoprecipitation with anti-NR1 (A) and anti-Src (B), or nonspecific IgG (C). Membrane proteins were solubilized under nondenaturing (Co-IP) or under denaturing conditions (IP). Proteins were resolved by SDS-PAGE, transferred to nitrocellulose, and analyzed by sequential immunoblotting with anti-NR1, anti-Src, or anti-Kv3.1 as indicated. SM, lanes loaded with 50 μg of solubilized membranes without immunoprecipitation; NS, lane loaded with antibody but no sample to demonstrate the position of the IgG heavy chain. Molecular size standards (in kilodaltons) are indicated on the right. The results are representative of five (A), six (B), or three experiments (C).

To investigate whether synaptic NMDA receptors were regulated by Src (17), we studied miniature excitatory postsynaptic currents (mEPSCs). These mEPSCs showed NMDA- and non-NMDA receptor-mediated components (Fig. 4A) (18). Intracellular administration of EPQ(pY)EEIPIA increased the average NMDA component of the mEPSCs to ∼160% of the initial value (Fig. 4, B and C) (n = 8 cells). During the recording period there was no change in reversal potential, indicating that there was no alteration of driving force (11). In contrast to recordings with EPQ(pY)EEIPIA, the NMDA component was unaffected when EPQYEEIPIA was administered (n = 5 cells). In other recordings, addition of anti-src1 or Src(40–58) significantly reduced the NMDA component of the mEPSCs, whereas sSrc(40–58) had no effect (Fig. 4D). The non-NMDA component of the mEPSCs was unaffected by EPQ(pY)EEIPIA, EPQYEEIPIA, anti-Src1 (n = 6 cells), Src(40–58) (n = 7 cells), or sSrc(40–58) (n = 6 cells). The non-NMDA channels studied in this way were located at the same synapses as the NMDA channels (18); thus, even within the restricted space of a single postsynaptic site, Src appeared to regulate NMDA but not non-NMDA channels.

Fig. 4.

Regulation of synaptic NMDA currents by endogenous Src. (A) Representative recording of averaged mEPSCs before (Control) or during application of APV (100 μM), and after wash; τ, decay time constants of the currents. (Inset) (Top) The I-V plot of the NMDA component of averaged mEPSCs. Each point is the mean of the amplitude of the NMDA component (±SEM; n = 6 cells) relative to those recorded at −60 mV [I(norm)]. (Bottom) An all-points distribution histogram of current 60 to 90 ms after the start of 20 consecutive mEPSCs. The distribution was fit as the sum of two Gaussian curves (solid line). One peak occurred at −0.1 pA and represented background noise. The other peak was at −2.7 pA, from which a single-channel conductance was calculated. (B) A representative recording of averaged mEPSCs (top traces) or NMDA components (INMDA) (bottom traces) compiled immediately after breakthrough (0 to 2 min), 8 to 10 min later, and during bath application of APV (100 μM) when EPQ(pY)EEIPIA (1 mM) was included in the recording pipette. We constructed INMDA by subtracting, from the averaged mEPSC, a current decaying at a single exponential rate equal to the fast component. Bars: time, 25 ms; current, 5 pA (top) and 2 pA (bottom). (C) NMDA and non-NMDA synaptic responses during recordings with EPQ(pY)EEIPIA (1 mM, closed symbols; n = 8 cells) or EPQYEEIPIA (1 mM, open symbols; n = 5 cells). We calculated charge (Q) by integrating currents during NMDA (circles) or non-NMDA (triangles) components of mEPSCs. For each cell, charge during consecutive 2-min periods (Qt) was normalized to that measured during the initial 2-min period (Q0). (D) Mean Q10/Q0 for NMDA or non-NMDA components. Recordings were done with intracellular solution (ICS; n = 5 cells) or intracellular application of EPQ(pY)EEIPIA, EPQYEEIPIA, anti-src1 (1→100 dilution; n = 6 cells), Src(40–58) (0.03 mg/ml; n = 7 cells), or sSrc(40–58) (0.03 mg/ml; n = 6 cells). (C and D) *P < 0.05, paired t test.

Src is expressed at high levels in the CNS and is found preferentially in neurons that express an alternatively spliced form of the enzyme (7). However, the functions of Src in the CNS have until now remained enigmatic (19). These results indicate that Src may be a member of the NMDA channel complex and that one function of Src could be to regulate NMDA channel activity.

NMDA receptor subunits 2A and 2B may be phosphorylated on tyrosine (20), and thus Src might regulate NMDA channels by phosphorylating these subunit proteins. Alternatively, Src might phosphorylate other proteins that are in the NMDA channel complex (15, 21). Tyrosine phosphorylation resulted in an increase in Po, possibly because of an increased probability that the channel will enter long-lived open states. In addition, there is a decreased probability that the channel will be in long shut states. Synaptic NMDA receptor-mediated currents result from single receptor activations that may correspond to clusters or superclusters in single-channel recordings (22). The magnitude of the effects of manipulating Src activity on synaptic NMDA currents corresponded to those of the changes in burst and cluster length more closely than to those of the changes in overall Po. This result is not unexpected given that simulated synaptic NMDA currents change with altered burst and cluster length even without changes in overall Po (23). The changes in single-channel behavior may account for the effects of tyrosine phosphorylation on whole-cell NMDA currents (3, 4), long-term depression in the cerebellum (24), and the effects of Src on the synaptic currents reported here. Our findings also indicate that the unique domain of Src may participate in NMDA channel regulation. This domain is not directly involved in catalysis (25), and thus it is likely that the sequence containing amino acids 40 through 58 participates in protein-protein interactions necessary for coordinating the phosphorylation of the Src target.

NMDA receptors have been implicated in development, plasticity, and pathology in the CNS (1). We postulate that Src, by virtue of its association with and regulation of NMDA receptors, may be important in NMDA receptor-dependent processes. Given that Src and NMDA receptors are widely expressed in the CNS, our results demonstrate a mechanism that may have a general role in regulating excitatory synaptic function in the nervous system.


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
    A concentration of 1 mM was the minimum required to increase Src activity; at concentrations of ≤0.1 mM the peptide had no effect.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 26.
View Abstract

Stay Connected to Science

Navigate This Article