NR2B-NMDA receptor mediated increases in intracellular Ca²⁺ concentration regulate the tyrosine phosphatase, STEP, and ERK MAP kinase signaling

Abstract

NMDA receptors regulate both the activation and inactivation of the extracellular signal-regulated kinase (ERK) signaling cascade, a key pathway involved in neuronal plasticity and survival. This bi-directional regulation of ERK activity by NMDA receptors has been attributed to opposing actions of NR2A- versus NR2B-containing NMDA receptors but how this is implemented is not understood. Here we show that glutamate-mediated intracellular Ca²⁺ increases occur in two phases, a rapid initial increase followed by a delayed larger increase. Both phases of the Ca²⁺ increase were blocked by MK-801, a non-selective NMDA receptor inhibitor. On the other hand selective inhibition of NR2B-NMDA receptors by Ifenprodil or Ro 25-6981 blocked the delayed larger phase but had only a small effect on the rapid initial increase. The rapid initial increase in Ca²⁺, presumably due to NR2A-NMDAR activation, was sufficient to activate ERK, whereas the large delayed increases in Ca²⁺ mediated by NR2B-NMDARs were necessary for dephosphorylation and subsequent activation of STEP, a neuron-specific tyrosine phosphatase that in turn mediates the dephosphorylation and inactivation of ERK. We conclude that the magnitude of Ca²⁺ increases mediated through NR2B-NMDA receptors plays a critical role in the regulation of the serine/threonine and tyrosine kinases and phosphatases that are involved in the regulation of ERK activity.

INTRODUCTION

The tyrosine phosphatase STEP (striatal-enriched phosphatase) is a signaling molecule downstream of both dopamine and NMDA receptors (Paul et al. 2000; Paul et al. 2003). STEP is highly enriched within the medium spiny neurons of the basal ganglia and related structures (Lombroso et al. 1993; Boulanger et al. 1995). There are multiple splice variants of STEP that include both cytosolic (STEP₄₆) and membrane-associated (STEP₆₁) variants (Sharma et al. 1995; Bult et al. 1997). The ability of STEP to bind to its substrates is regulated by phosphorylation of a serine residue located within the kinase-interacting motif or KIM domain. Phosphorylation of this serine residue in response to activation of dopamine/D1 receptors/cAMP-dependent protein kinase-A pathway, decreases the activity of STEP by reducing its affinity for its substrates (Paul et al. 2000). Dephosphorylation of this same serine residue activates STEP and has been shown to occur via calcineurin activity promoted by glutamate/NMDA receptor-mediated Ca²⁺ influx (Paul et al. 2003). Both in vitro and in vivo studies show that the activity of STEP is low in neurons under basal conditions (Paul et al. 2003; Valjent et al. 2005). Once activated, STEP can bind to and down-regulate the activity of extracellular-regulated kinase 2 (ERK2), a key signaling protein involved in NMDA receptor dependent synaptic plasticity and excitotoxicity (Komiyama et al. 2002; Banko et al. 2004; Thomas and Huganir 2004; Kaphzan et al. 2006; Paul et al. 2007). NMDA receptor stimulation by glutamate or NMDA leads to rapid but transient phosphorylation of ERK2. Both phosphorylation and dephosphorylation of ERK2 is Ca²⁺ dependent (Jiang et al. 2000b, a; Chandler et al. 2001; Paul et al. 2003). The dephosphorylation of ERK2 has been attributed to the delayed activation of STEP following NMDA receptor stimulation (Paul et al. 2003) but the mechanisms by which the same messenger, Ca²⁺, is signaling both the phosphorylation and dephosphorylation of ERK2 has not been revealed.

The aim of the present study was to determine the relative contribution of NR2A- and NR2B-NMDARs in mediating intracellular Ca²⁺ increases and examine if the extent of Ca²⁺ increase is a critical determinant of STEP and ERK activity in neurons. Our results show that NMDA receptor stimulation leads to an initial rapid rise in Ca²⁺ followed by a gradual but larger increase in Ca²⁺. The delayed increase in Ca²⁺ is NR2B-NMDAR-dependent. We also show that glutamate-mediated dephosphorylation and subsequent activation of STEP is dependent on this large delayed, NR2B-NMDAR-dependent increase. The preferential activation of STEP by NR2B explains the sub-type specific function of NR2B-NMDARs in the inhibition of ERK activity.

MATERIALS AND METHODS

Materials and reagents

Pregnant female Sprague-Dawley rats (16 day gestation) were from Harlan Laboratories. Antibodies used were as follows: polyclonal anti-NR1 from Abcam, polyclonal anti-NR2B from Millipore, polyclonal anti-NR2A from Covance Research Products, monoclonal anti-STEP from Novus Biologicals, polyclonal anti-phospho-ERK1/2 (TEY^P) from Santa Cruz Biotechnology and polyclonal anti-tubulin from Sigma-Aldrich. All secondary antibodies, Fura 2-AM and Fura FF-AM were from Invitrogen. All other reagents were from Sigma-Aldrich. Approval for animal experiments was given by the University of New Mexico, Health Sciences center, Institutional Animal Care and Use Committee.

Cell culture and stimulation

Primary neuronal cultures were obtained from 16–17 day old rat embryos as described previously (Paul et al. 2003). Briefly, the striatum and the adjoining cortex were dissected, the tissue dissociated mechanically and resuspended in DMEM/F-12 (1:1)-containing 5% fetal calf serum. Cells were plated on poly-D-lysine-coated tissue culture dishes (6 × 10⁶ cells/dish) and grown for 12–14 days at 37°C in a humidified atmosphere (95:5% air:CO₂ mixture). To inhibit proliferation of non-neuronal cells, 10 μM of cytosine D-arabinofuranoside was added to the cultures 72 hr after plating. For receptor stimulation cells were treated with glutamate, NMDA or KCl for the indicated times at 37°C. 1 μM of glycine was added to the media during treatment with glutamate or NMDA. For biochemical experiments NMDA receptor antagonists MK-801, Ifenprodil or Ro 25-6981 (Fischer et al. 1997; Williams 2001; Waxman and Lynch 2005; Kohr 2006) was added 20 min prior to stimulation. The doses selected for MK-801, Ifenprodil and Ro 25-6981 were based on previous studies that have used the compounds to determine the role of NR2B-NMDA receptors in the regulation of NMDA receptor currents, protein phosphorylation, downstream signaling cascades as well as neuronal cell death (Crozier et al. 1999; Tovar and Westbrook 1999; Kim et al. 2005; Waxman and Lynch 2005; Yang et al. 2006; Martel et al. 2009). For Ca²⁺ measurement experiments these antagonist were added at the times indicated.

Immunoprecipitation

Cells were lysed in a buffer containing 50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 50 mM NaF, 10 mM Na₄P₂O₇, 1 mM Na₃VO₄, 0.1% NP-40, 0.1% sodium-deoxycholate and protease inhibitor cocktail (Boehringer). Lysates were centrifuged for 10 min at 14,000 rpm to remove insoluble material, and then pre-cleared with protein G-sepharose for 1 hr. For immunoprecipitation of STEP, samples were incubated overnight with anti-STEP antibody. The immune complexes were incubated with 30 μl of protein G-sepharose for 2 hr at 4°C. Beads were collected by centrifugation at 1000 rpm for 2 min and washed five times with the lysis buffer. Proteins were eluted using SDS-sample buffer and processed for immunoblot analysis with antibodies as described in the individual experiments. Densitometric analysis of immunoblots was performed using Image J software (NIH, Bethesda, MD, USA). Statistical comparison was carried out using One-way analysis of variance (ANOVA, Bonfrroni's multiple comparison test) and differences were considered significant when p < 0.05.

Immunofluorescence

To demonstrate co-expression of STEP and NMDA receptor subunits in neurons, 12–14 days old neuronal cultures grown on poly-D-lysine coated 2-well chamber slides (BD Biosciences) were fixed with 4% para-formaldehyde in phosphate-buffered saline (PBS, pH 7.4) for 10 min, permeabilized with 0.1% Triton-X-100 in PBS for 10 min, blocked with 10% normal goat serum and 1% BSA in PBS for 1 hr followed by incubation with anti-STEP antibody and anti-NR1, -NR2A or -NR2B antibody overnight at 4°C. Cells were then incubated with anti-mouse Alexa 488 and anti-rabbit Cy3 conjugated secondary antibodies, followed by washing with PBS and mounting on glass slides. Imaging was performed with a Zeiss Axiovert 200M fluorescence microscope with attached AxioCam CCD camera using 20× and 63× objective lenses.

Calcium Measurements

Neuronal cultures were grown on poly-D-lysine coated 2-well chambered glass slides (Nunc, New York) for 12 – 14 days. Cytosolic Ca²⁺ determination was performed using 1 of the 2 fluorescent indicators, Fura-2/AM (Kd ~ 0.22 μM) or the lower affinity indicator Fura-FF/AM (Kd ~ 5.5 μM, both from Molecular Probes). Cells were placed in MEM (without glutamic acid and phenol red), containing either Fura-2/AM (5 μM) or Fura-FF/AM (5 μM) for a 30 min loading time. Bathing medium was then changed to one without indicator and cells were given a 20 min post incubation time at 37°C (Connor 1986). Covered culture plates were then placed on temperature-controlled stage (33°C) of a Zeiss IM35 microscope and continuously gassed with pre-moistened 95:5% Air:CO₂ mixture. Fields of 5–20 cells were viewed from underneath using a 40× glycerin objective (Nikon). Excitation light was obtained from a 100 W HBO short arc lamp through 340 and 380 nm narrowband interference filters mounted in a Sutter Instruments Wheel. The standard exposure time was 200 ms for each wavelength. Fluorescence emissions (> 500 nm) were captured using a Photometrices Pixel, cooled CCD, camera (512 × 512 elements). Maximum frame pair rate in was ~ 2/sec. The microscope was also fitted with a Ludl stepper motor stage that allowed precise repositioning of recording locations for multisite recordings during responses. IP Lab software running on a Macintosh G3 computer was used for data acquisition and analysis. Ratio data from regions of interest (neuron somata and processes) were extracted from background-subtracted image pairs (340 and 380 nm excitation). Ratios were converted into estimates of free intracellular Ca²⁺ concentration as described previously (Connor 1986) using the equation developed by Grynkiewicz et al., (Grynkiewicz et al. 1985). Minimum and maximum fluorescence ratios determined by bath indicator loaded cells in saline containing 2 μM ionomycin and either 0 Ca²⁺, 0.5 mM EGTA or 2 mM Ca²⁺.

RESULTS

NR2B-NMDA receptor mediated dephosphorylation of STEP and ERK2

Initial studies examined the expression profile of STEP and NMDA receptor subunits in cultured neurons (12–14 days in vitro) by immunocytochemical staining with antibodies specific for STEP, NR1, NR2A or NR2B (Fig. 1A). As previously reported, STEP is expressed in both the soma and the processes of all neurons in the cultures (Lombroso et al. 1993; Paul et al. 2003). NR1 subunits are heavily expressed in the neuronal processes but are also present in the soma. NR2A and NR2B subunits are abundant in both the cell bodies and the neuronal processes, which is in agreement with previous studies (Petralia et al. 1994; Paquet et al. 1997; Rao et al. 1998; Charton et al. 1999; Tongiorgi et al. 1999; Hallett et al. 2006). Co-expression of STEP and NMDA receptor subunits in the same cells is demonstrated by overlaying the images produced with anti-STEP and anti-NR1, -NR2A or -NR2B antibodies (Fig. 1A, right column). It is also evident from the merged images of the high magnification photomicrographs of neuronal processes that both STEP and NMDA receptor subunits are co-expressed in the same processes. To determine if STEP interacts with the NMDA receptor complex in cultured neurons we immunoprecipitated STEP from neuronal lysates using anti-STEP antibody. Since all functional NMDA receptors are composed of NR1 subunits, co-immunoprecipitation of NMDA receptor complexes with STEP was analyzed using anti-NR1 antibody. The results show that STEP₆₁, the membrane bound isoform of the STEP-family led to co-precipitation of NR1 (Fig. 1B). This is consistent with earlier findings showing that STEP₆₁ is a component of the NMDA receptor complex in vivo (Pelkey et al. 2002; Braithwaite et al. 2006).

Figure 1.

Co-expression of STEP and NMDAR subunits in neurons and NR2B-NMDA receptor-mediated dephosphorylation of STEP and ERK2. (A) Immunocytochemical analysis showing the co-expression of STEP and NMDA receptor subunits NR1, NR2A and NR2B in both neuronal somata and processes. High magnification (63×) images of neuronal processes are shown in the horizontal strips below each lower resolution (20×) panel (B) Immunoblot analysis demonstrating that STEP is part of the NMDAR complex. STEP₆₁ was immunoprecipitated from neuronal lysates with anti-STEP antibody, analyzed by SDS-PAGE and probed with anti-NR1 antibody (upper panel). The blot was re-probed with STEP antibody (lower panel). Immunoprecipitation with Protein G-Sepharose beads alone was included as a control. To ensure the presence of both STEP and NR1 in the total lysate (input lysate used for immunoprecipitation), equal amount of protein from the lysates was processed for immunoblot analysis with anti-NR1 and -STEP antibodies. (C) Immunoblot analysis demonstrating the dephosphorylation of STEP following NR2B-NMDAR activation. Neurons were treated with glutamate for 30 min, in the presence or absence of Ro 25–6981 or Ifenprodil. Equal amount of protein from each sample were analyzed by SDS-PAGE and probed with anti-STEP antibody (upper panel). Total ERK2 was also analyzed to indicate total protein loading (lower panel). (D) STEP is endogenously phosphorylated and glutamate exposure leads to its dephosphorylation. Neurons were treated with or without glutamate for 30 min and STEP₆₁ was immunoprecipitated with anti-STEP antibody. The samples were then processed for immunoblotting with the phospho-specific STEP antibody (left panel). The blot was then re-probed with anti-STEP antibody (right panel). (E) Dephosphorylation of ERK2 correlates with activation of STEP. Neuronal cultures were treated with glutamate for the specified time periods in the presence or absence of Ro 25–6981. Phosphorylated ERK2 was identified using a phospho-specific antibody that recognizes ERK1/2 only when phosphorylated at the regulatory tyrosine residue (TEY^PERK, upper panel). Total ERK2 and STEP was analyzed by probing the membrane with either anti-ERK2 antibody (middle panel) or anti-STEP antibody (lower panel). Quantification of phosphorylated ERK2 (active ERK) and dephosphorylated STEP (active STEP) was done by computer-assisted densitometry. Values are mean ± SEM (n = 3). *Indicates significant difference from 5 min glutamate treatment (p < 0.001). ^#Indicates significant difference from 0 min time point (p < 0.001).

Previous studies showed that glutamate-mediated NMDA receptor stimulation leads to dephosphorylation and subsequent activation of STEP (Paul et al. 2003). To examine the respective role of NR2A- and NR2B-containing NMDA receptors in the dephosphorylation of STEP we treated neurons with glutamate (100 μM, glycine 1 μM) for 30 min in the absence or presence of NR2B-NMDAR antagonists ifenprodil and Ro 25-6981. Cell lysates were then analyzed by immunoblotting with anti-STEP antibody. Under basal condition STEP₆₁ was present as a doublet and the upper band (phosphorylated form) constituted the principal component of the doublet (Fig. 1C, lanes 1 and 5). Incubation of neurons with glutamate alone resulted in a marked decrease in the upper band and the appearance of a predominant lower band, indicating that STEP₆₁ is rapidly dephosphorylated following glutamate treatment (Fig. 1C, lanes 2 and 6). Ro 25-6981 and Ifenprodil both blocked the downward shift in the mobility of the upper band, supporting a role for NR2B-NMDAR functioning in the dephosphorylation of STEP₆₁ (Fig. 1C, lanes 3 and 7). Using a phospho-specific antibody that recognizes STEP only when it is phosphorylated at the protein kinase A site in the KIM domain (Paul et al. 2003) we further demonstrated that STEP₆₁ is basally phosphorylated at this site. A marked decrease in this basal phosphorylation of STEP₆₁ was evident following treatment with glutamate (Fig. 1D, left panel). Re-probing the membrane with anti-STEP antibody showed that glutamate treatment resulted in significant loss of the upper band that was detected with phospho-specific antibody and the appearance of the lower dephosphorylated form of STEP₆₁ (Fig. 1D, right panel).

To examine the role of NR2B-NMDARs in the regulation of ERK2 phosphorylation, neuronal cultures were treated with glutamate (100 μM) in the presence or absence of Ro 25-6981. Immunoblot analysis with the phospho-specific ERK antibody, that recognizes only the tyrosine phosphorylated ERK1/2 (TEY^P-ERK), showed that glutamate stimulated the tyrosine phosphorylation of ERK2 within 5 min with a reduction to near basal levels occurring by 30 min (Fig. 1E, upper panel, lanes 2 and 4). Incubation with Ro 25-6981 prior to treatment with glutamate had no significant effect on the phosphorylation of ERK2 at 5 min (Fig. 1E, upper panel, lane 3) but prevented the decrease in phosphorylation of ERK2 that is normally observed by 30 min (Fig. 1E, upper panel, lane 5). Analyzing the same samples with anti-STEP antibody showed that incubation with Ro 25-6981 also blocked the glutamate-mediated dephosphorylation of STEP (Fig. 1E, lower panel, lane 5) by 30 min. Thus glutamate-mediated NR2B-NMDARs stimulation is responsible for the dephosphorylation and activation of STEP and subsequent tyrosine dephosphorylation of ERK2. After exposure to Ro 25-6981 there was a trend towards decreased active STEP at 5 min from basal levels, which correlates with the trend to increased active ERK at the same time point.

NR2B-NMDAR dependent changes in intracellular Ca²⁺ concentration

Since NR2B-NMDAR antagonists blocked the dephosphorylation of STEP it was important to determine whether these antagonists also produced a significant change in intracellular Ca²⁺ levels in response to glutamate. As glutamate-induced Ca²⁺ increases are often very large (Petrozzino et al. 1995; Hyrc et al. 1997; Cheng et al. 1999; Stout and Reynolds 1999), we used the low affinity indicator Fura-FF for most of these experiments. Figure 2A shows the Ca²⁺ responses (in false color maps) of a small field of cells exposed to 100 μM glutamate, 1 μM glycine. The 30 sec and 3 min frames show that there was a rapid, but relatively small, increase in intracellular Ca²⁺ level at 30 sec but almost no subsequent change at the 3 min time point. This was followed by a progressive increase to a plateau over the next 14 min. The population data in Figure 2B obtained from a different experiment encompassing more cells illustrate more clearly the small, rapid onset increase followed by a slowly developing and much larger increase (n = 10). Individual responses measured in the somata of these ten neurons are shown in Figure 2C, demonstrating the similar qualitative features of the increase and also the considerable dispersion in the time courses over which the large delayed increase developed. The fluorescence ratio for Fura-FF in unstimulated cells was ~ 0.12, but this indicator is very insensitive at Ca²⁺around neuron resting levels owing to its large dissociation constant. Fura-2 measurements in companion cultures showed resting levels to be in the range of 50 –130 nM, which is consistent with measurements from many laboratories (data not shown). The Fura-FF plateau levels reached during glutamate exposure were considerably less than the maximum in situ ratio of 2, produced by ionomycin (see Materials and Methods). Such delayed Ca²⁺ increases were observed in 60 of the 80 cells obtained from six independent preparations. A similar temporal profile of changes in intracellular Ca²⁺ was also observed in neurons loaded with Fura-FF and treated with 100 μM NMDA (data not shown).

Figure 2.

Intracellular Ca²⁺ changes measured with Fura-FF following exposure to glutamate. (A) Initial B&W panel: 380 nm excitation fluorescence picture of the field of cells analyzed. Subsequent panels: false color images showing the fluorescence ratio increases over time following exposure to glutamate. (B) Averaged population data showing the time course of glutamate stimulated intracellular Ca²⁺ increase measured at neuronal somata (mean ± SEM, n = 10 cells). (C) Individual responses of the 10 neurons are shown to illustrate the typical range of responses observed in different cells. Data are expressed as both a background-corrected fluorescence ratio (left abscissa) and the estimated Ca²⁺ (right abscissa). Because of the low sensitivity of Fura-FF, Ca²⁺ estimates below 1 μM are not reliable and should not be extrapolated from the right abscissa.

Although most of the quantifications of Ca²⁺ changes used measurements made at the soma, we were able to analyze a number of cases where the soma and its main dendrite (out to at least 40 μm were identifiable (n = 26). Relative Ca²⁺ increases in these two compartments measured at times < 3 min after agonist exposure fell into 3 categories; dendrite > soma (n = 11), dendrite and soma < 30% difference (n = 10), dendrite < soma (n = 5). In all but 4 cases (one illustrated in Fig. 4) the changes had equalized to levels within 30% at times greater than 5 min. It is worth noting that most previous measurements of Ca²⁺ responses to glutamate have examined much shorter time periods (< 5 min) and have been made using higher affinity indicators. Under these conditions, Ca²⁺ increases in dendrites can be much larger than those in the somata for the first 1 – 2 min. Indeed we observed this to occur in a proportion of the cells here. However, at the longer time periods we have been interested in here, > 5 min, increases at the soma had “caught up” with dendritic increases. Dendrite measurements are also subject to the consideration that fluorescence is relatively weak and during responses where Ca²⁺ levels are very high, the 380 nm excited fluorescence (see Materials and Methods) can drop to values close to background, decreasing the reliability of the measurements. Because difference in the measurable Ca²⁺ values between soma and dendrite are small, and also a large fraction of the total NR2A- and NR2B-NMDA receptors are located at the soma (see Fig. 1), we have employed soma measurements to enable the greatest number of reliable observations. In this context we also note that a previous study (Sinor et al. 2000) has stressed the importance of somatic NR2B receptors in glutamate neurotoxicity.

Figure 4.

Ca²⁺ recovery enabled by late application of NR2B-NMDAR blocker Ro 25–6981. (A) Fluorescence image of neuron showing measurement locations. (B) Time courses of soma and dendrite Ca²⁺ changes determined at the locations (see boxes in A). The arrow indicates approximate time of Ro 25–6981 application.

We next examined the effect of NR2B antagonists, Ro 25-6981 and ifenprodil, on glutamate-induced changes in intracellular Ca²⁺ level. Exposure to Ro 25-6981 before and during glutamate exposure produced only a small decrease in the initial Ca²⁺ response (4.2 μM to 3.3 μM) measured at the 3 min time point (compare Fig. 2B with 3A). In contrast, the large, late developing phase was blocked (compare Fig. 2B with 3A and 3B). Ifenprodil reduced the initial Ca²⁺ increases to a greater extent (Fig. 3A), consistent with its less selective blocking profile (Honda et al. 1988; Contreras et al. 1990; Bath et al. 1996). Figure 3C demonstrates that in the presence of glutamate removal of extracellular Ca²⁺ (no Ca²⁺ added to media + 2 mM EGTA) brought about recovery from an ongoing glutamate response, arguing that the Ca²⁺ increases were due to influx and not to Ca²⁺ release from intracellular stores. MK801, a non-competitive NMDA receptor antagonist, applied before glutamate, blocked nearly all of the Ca²⁺ increases reported by the low affinity indicator Fura-FF (Fig. 3D). This great reduction occurred despite the activation of AMPA receptors which both depolarize neurons and activate voltage gated Ca²⁺ channels, and where Ca²⁺-permeable AMPA receptors are present, directly enable Ca²⁺ influx (Gorter et al. 1997). Again, it is noted that high affinity indicators readily detect the Ca²⁺ increases under these conditions.

Figure 3.

NR2B-NMDA receptor-mediated change in intracellular Ca²⁺ measured with Fura-FF. (A) Population data showing that large, delayed Ca²⁺ increase does not occur in cells pre-incubated with 1μM Ifenprodil or 10 μM Ro 25–6981 before exposure to glutamate (n = 14). (B) Population data showing that 20 μM Ro 256981 even applied after 5 min exposure to glutamate also prevents the delayed Ca²⁺ increase (n = 15). (C) Low Ca²⁺ levels are restored after switch to a Ca²⁺ free medium still containing 100 μM glutamate (n = 15). (D) The NMDAR channel blocker, MK801, eliminates all changes in the Fura-FF signal (n = 9). Values represent mean ± SEM. Similar results to those shown were obtained from at least 3 independent experiments for all treatments.

The finding that initial Ca²⁺ increases were very similar in the absence or presence of the selective NR2B antagonist Ro 25-6981 (compare Fig. 2B with 3A) argues that the late Ca²⁺ increases does not result simply from the loss of regulatory competence of the cells due to the early Ca²⁺ load. Even late addition of Ro 25-6981 after 5 min incubation with glutamate, prevented the delayed Ca²⁺ increase (Fig. 3B) with the cells able to regulate the persisting Ca²⁺ influx for periods of 20 min or more. Moreover the plateau of the late response is well below the saturation level of Fura-FF (see Materials and Methods) indicating that the cells have not simply died. In some instances it was observed that delayed application of the Ro 25-6981 led to an actual decrease in intracellular Ca²⁺ in the continued presence of glutamate. One such example is given in Figure 4 where both soma and dendrite measurements were possible. Following treatment with glutamate Ca²⁺ levels displayed their usual increase (1^st 5 data sets) till Ro 25-6981 was applied approximately at 5.5 min (as shown by an arrow in Fig. 4B). The subsequent measurements show a clear reduction in intracellular Ca²⁺ in both the soma and dendrite. Complete recovery is not expected since other glutamate/depolarization pathways should still be active. The above findings imply that in the minutes following glutamate (or NMDA) exposure, an additional influx pathway for Ca²⁺ mediated by NR2B receptors is recruited which gains importance around 5 min (on average) after the exposure to glutamate or NMDA. This late phase correlates well with both the time course and antagonist sensitivity of STEP dephosphorylation.

The magnitude of intracellular Ca²⁺ changes regulates STEP dephosphorylation

To determine if the glutamate-mediated dephosphorylation of STEP is dose-dependent, neuronal cultures were exposed to either 10 μM or 100 μM glutamate for 5 and 30 min. Immunoblot analysis of cell lysates with anti-STEP antibody showed that treatment with 10 μM glutamate failed to induce any dephosphorylation of STEP (Fig. 5A, lane 3). However, as shown previously, treatment with 100 μM glutamate for 30 min led to significant dephosphorylation of STEP as evident from the downward shift in the mobility of the STEP band (Fig. 5A, lane 5). We next investigated the dose-dependence of glutamate-mediated Ca²⁺ increases. Neurons loaded with Fura-FF and exposed to 10 μM glutamate showed a rapid Ca²⁺ increase of similar magnitude to 100 μM exposure but a much smaller late phase increase (Fig. 5C). The peak Ca²⁺ level was significantly lower for 10 μM glutamate as compared to neurons exposed to 100 μM glutamate (Fig. 2B).

Figure 5.

Dephosphorylation of STEP is dependent on the magnitude of intracellular changes in Ca²⁺. Neuron cultures were (A) treated with 10 μM or 100 μM glutamate, (B) treated with 100 μM glutamate or 60 mM KCl, (C) treated with 10 μM glutamate, and (D) treated with 60 mM KCl for the times indicated. (A and B) Equal protein from each sample were analyzed by SDS-PAGE and probed with anti-STEP antibody (upper panel). The blots were re-probed with anti-β tubulin antibody to indicate total protein loading (lower panel). Quantification of phosphorylated (upper band) and non-phosphorylated (lower band) STEP was done using Image J density analysis of immunoblot data obtained from 3 independent experiments. The data is presented as a percentage of the total. Values are mean ± SEM (n = 3). *Indicates significant difference from corresponding control values at 0 min time point (p < 0.001). (C and D) Cells were loaded with Fura-FF prior to treatment with glutamate (n = 13) or KCl (n = 10). Values represent mean ± SEM. Similar results were obtained from at least 3 independent experiments.

In addition to NMDA receptors L-type voltage-gated Ca²⁺ channels are another principal mediator of Ca²⁺ entry in neurons and are involved in the activation of ERK2 (Sala et al. 2000; Dolmetsch et al. 2001). But as shown in Figure 5B, and as previously reported KCl-mediated membrane depolarization and influx of Ca²⁺ failed to dephosphorylate STEP thereby leading to sustained activation of ERK (Paul et al. 2003). In neurons loaded with Fura-FF and exposed to 60 mM KCl there was a relatively small, transient increase in Ca^2+, which came back almost to the basal level within 3 min and stayed there for the entire period of the experiment (Fig. 5D). Again the peak Ca²⁺ level was far lower than that compared to neurons exposed to 100 μM glutamate (Fig. 2B).

DISCUSSION

Our study shows that the intracellular Ca²⁺ increases occurring after exposure to 100 μM glutamate are very large, requiring the use of a low affinity indicator (here, Fura-FF) to track them, and follow a time course that has two separable phases. In the presence of specific antagonists of NR2B-NMDAR the large delayed increase did not occur, while the smaller, rapid onset phase remained nearly unchanged. In the presence of the non-selective NMDA receptor antagonist MK-801, glutamate application produce no Ca²⁺ increase measurable with Fura-FF (see Figure 3D), indicating a role for the NR2A-NMDARs in the initial rapid increase in Ca²⁺. At present there is no selective NR2A-NMDAR antagonist available to provide a better estimate of the Ca²⁺ increases, due solely to NR2A-NMDAR activation. Zinc has been proposed as a moderately selective and partial blocker for NR2A-NMDAR, but its selectivity is not high enough to make a strong discrimination between the receptor subtypes (Westbrook and Mayer 1987; Chen et al. 1997a; Paoletti et al. 1997; Choi and Lipton 1999; Madry et al. 2007). Evidences also suggest that zinc exposure might up-regulate NMDA receptor activity via activation of Src family kinases (Manzerra et al. 2001; Kim et al. 2002). Besides modulating NMDA receptor function zinc also acts on other post-synaptic channels and receptors including voltage gated ion channels, GABA (γ-aminobutyric acid), dopamine and Trk (tyrosine kinase) receptors (Bitanihirwe and Cunningham 2009) and therefore cannot be used as a selective inhibitor for NR2A-NMDA receptors. More recent studies reported the discovery of a new NMDAR antaogonist, NVP-AAM077 ([(R)-[(S)-1-(4-bromophenyl)-ethylamino]-(2, 3-dioxo-1,2,3,4-tetrahydroquinoxalin-5-yl)-methyl]-phosphonic acid)) (Auberson et al. 2002), which was claimed to display strong selectivity for NR2A-containing NMDA receptors (Liu et al. 2004). However, its selectivity has been debated in several studies (Feng et al. 2004; Berberich et al. 2005; Weitlauf et al. 2005; Frizelle et al. 2006; Neyton and Paoletti 2006; de Marchena et al. 2008). It is also not commercially available.

The rapid onset, but relatively low level Ca²⁺ increases which we hypothesize here are largely due to NR2A-NMDAR activation, are sufficient to activate ERK2 but had no effect on the tyrosine phosphatase STEP, allowing ERK to remain active throughout the time course examined. On the other hand, the delayed larger Ca²⁺ increase resulting from NR2B-NMDAR activation leads to the dephosphorylation and subsequent activation of STEP. As a consequence, activated STEP dephosphorylates ERK thereby limiting the duration of its activation.

The two phase nature of Ca²⁺ increases following exposures to high levels of glutamate can be seen in published work but the phases are much less distinct than we report here and have not been remarked upon (Hyrc et al. 1997; Cheng et al. 1999). A particular contribution of our study is the identification of the late increase being due to NR2B-NMDAR's and not to other factors such as failure of Ca²⁺ regulatory competence. This is not to say that such failure does not occur at later times. The blocker Ro 25-6981 had a very small effect on the early phase of Ca²⁺ increase from which we hypothesize that this initial increase is largely due to NR2A-NMDAR's and voltage-gated channels. The small size of the early Ca²⁺ increase relative to the later NR2B-NMDAR mediated increase is at least consistent with the developmental stage of the cultures used (~ DIV 14). Cheng et al., (1999) have shown that NR2A expression is < 50% of that reached at later times, while NR2B is ~ 150% of later values (Stocca and Vicini 1998). There is a well characterized, Ca²⁺ dependent inactivation of the current carried by NMDA receptors (Krupp et al. 1996; Kyrozis et al. 1996), but the relatively small size of the presumed NR2A dependent Ca²⁺ increase phase cannot be attributed to this phenomenon, since this inactivation is very rapid, occurring over a period of a few seconds and would be missed in the slow, low affinity measurements used here. The reason for the delay in onset of the NR2B mediated response is not known. It certainly cannot be due to channel kinetics since the time scale is minutes. We propose that it reflects the recruitment of NR2B-NMDAR's into a functional state from a pool of non-functional receptors and that investigation of the mechanisms that bring this about is an important topic for future study.

We have regarded the primary underlying cause of the Ca²⁺ increases seen here to be trans-membrane influx of Ca²⁺ rather than the other well studied cause, Ca²⁺ release from intra-cellular stores. Although receptor activated Ca²⁺ increases, due to intracellular release, have been demonstrated in cortical culture preparation as well as in other neuronal preparations (Pozzo-Miller et al. 2000) these increases have been much smaller, in the sub micromolar range, and are transient, lasting a few seconds at most. Changes reported here range from 2–30 μM upward and persist for many minutes, so that while intracellular release may possibly underlie a part of the responses, the contribution is small. Moreover the decrease in intracellular Ca²⁺ level a. lmost to base line following removal of extracellular Ca²⁺ 5 min after glutamate treatment further indicates that the underlying cause of these Ca²⁺ increases to be trans-membrane influx of Ca²⁺. Consistent with this interpretation, electrophysiological studies from other laboratories have shown a similar biphasic change in cationic currents in acutely dissociated hippocampal neurons following exposure to 100 μM NMDA (Chen et al. 1997b; Thompson et al. 2008). The Chen et al., (1997) study was made before NR2B antagonists were available. The Thompson et al., (2008) study demonstrated the response in hippocampal cultures also. Here, there was a rapidly developing inward current, carried partially by Ca²⁺ that would give rise to the rapid Ca²⁺ increase, followed by a larger, slowly developing current that would give rise to the larger, more slowly developing Ca²⁺ increase. In the study of Thompson et al., the slow developing cation current was attributed to the opening of Pannexin-1 hemichannels in the plasma membrane, this late developing current being blocked by either carbenoxolone, or a small peptide inhibitor, or RNAi of Pannexin1. In our study, the NR2B antagonists, ifenprodil and Ro 25-6981 blocked the delayed Ca²⁺ increase but carbenoxolone did not (data not shown). There was also no significant loss of indicator in our studies as occurs with hemichannel opening (Thompson et al. 2008). The straightforward interpretation of the findings is that different mechanisms are operative in different subtype of neurons.

Primary cortical neuronal cultures treated with high concentrations of glutamate (50 –100 μM) are commonly used as a model of neuronal injury. In severe brain disorders such as ischemic stroke and traumatic brain injury a marked increase in the release of glutamate has also been observed and is responsible for NMDAR-mediated brain damage (Simon et al. 1984; Choi and Rothman 1990; Michaels and Rothman 1990; Globus et al. 1995; Rossi et al. 2000). Direct measurement of extracellular glutamate concentration with in vivo microdialysis has shown that experimental ischemia for more than 30 min causes extracellular glutamate levels to rise from under 5 μM to more than 500 μM (Benveniste et al. 1984; Hagberg et al. 1985; Choi and Rothman 1990). Although the origin of the increased extracellular glutamate has not been unequivocally established it has been suggested that reversed operation of both neuronal and glial transporters contribute to the excessive glutamate burden (Rossi et al. 2000). Few studies have also addressed the role of NR2A and NR2B receptor subtypes in glutamate mediated excitotoxicity in vivo (Morikawa et al. 1998; Liu et al. 2007; Tu et al. 2010). The study by Morikawa and colleagues (1998) reported that generating ischemic stroke in mutant mice that lack NR2A resulted in significant reduction in brain damage suggesting that NR2A plays a role in glutamate neurotoxicity. More recent studies, using animal models of ischemic stroke, have reported a role of NR2B-NMDAR in ischemic brain damage (Liu et al. 2007; Tu et al. 2010). The study by Tu et al., (2010) shows that interaction of DAPK1 (death associated protein kinase 1) with NR2B-NMDAR complex mediates brain damage in ischemic stroke and disruption of this association provides neuroprotection. These latter findings imply NR2B-NMDARs are also an important mediator of excitotoxicity not only in neuronal cultures but in mature neurons in vivo as well. Knowledge regarding the specific intracellular signaling pathways leading to glutamate / NMDAR-induced toxicity is incomplete. However, several in vivo studies have shown that ERK1/2 plays a crucial role in ischemic brain injury and specific inhibitors of ERK signaling pathway provides protection against forebrain ischemia (Alessandrini et al. 1999; Irving et al. 2000; Namura et al. 2001). In this context we report that relatively prolonged (> 5 min) treatment with an excitotoxic dose of glutamate (100 μM) leads to NR2B-NMDAR mediated dephosphorylation and subsequent activation of STEP. Since STEP is involved in the inactivation of ERK, the findings suggest that STEP can play an initial role in neuronal cell survival by preventing sustained activation of ERK. However, if the extracellular insult is sustained this protection may be ineffective, since extended exposure to an excitotoxic dose of glutamate may lead to breakdown of STEP and subsequent activation other degenerative processes. Consistent with this hypothesis, a recent study using an animal model of ischemic stroke has shown that STEP is unilaterally down regulated in the stroked hemisphere over time. Furthermore, STEP levels are increased in a neuroprotective-preconditioning paradigm in regions of the brain that are resistant to ischemic brain damage (Braithwaite et al. 2008). Taken together these findings suggest that STEP may be considered as a significant neuroprotectant against brief excitatory insults.

Abbreviations

ERK

extracellular-regulated kinase

MAP

mitogen-activated protein

APV

DL-2-amino-5-phosphonopentanoic acid

CNQX

6-cyano-7-nitroquinoxaline-2,3-dione

STEP

striatal-enriched phosphatase

SDS

sodium dodecyl sulfate

PKA

protein kinase A

DIV

days in vitro