Binding of CaMKII (Ca2+/calmodulin-dependent protein kinase II) to the NR2B subunit of the NMDAR (N-methyl-D-aspartate-type glutamate receptor) in the PSD (postsynaptic density) is essential for the induction of long-term potentiation. In this study, we show that binding of NR2B to the T-site (Thr286-autophosphorylation site binding pocket) of CaMKII regulates its catalysis as reflected in the kinetic parameters. The apparent S0.5 (substrate concentration at half maximal velocity) and Vmax values for ATP were lower for phosphorylation of a GST (glutathione transferase)-fusion of NR2B(1271-1311) (with the phosphorylation site Ser1303) when compared with phosphorylation of the analogous sequence motif from NR2A. The co-operative behaviour exhibited by the CaMKII holoenzyme towards ATP for phosphorylation of GST–NR2A was significantly altered by the interaction with GST–NR2B. Disrupting the T-site-mediated binding by mutagenesis of either NR2B or CaMKII abolished the modulation of CaMKII activity by NR2B. The active site residue of α-CaMKII, Glu96, participates in effecting the modulation. The CaMKII-binding motif of the Drosophila voltage-gated potassium channel Eag interacted with the T-site of CaMKII with lower affinity and caused catalytic modulation to a lesser extent. The kinetic parameters of ATP for the Thr286-autophosphorylation reaction of CaMKII were also altered by NR2B in a similar manner. Interestingly, the NR2B sequence motif caused increased sensitivity of CaMKII activity to ATP, and saturation by lower concentrations of ATP, which, in effect, resulted in a constant level of activity of CaMKII over a broad range of ATP concentrations. Our findings indicate that CaMKII at the PSD may be regulated by bound NR2B in a manner that supports synaptic memories.
CaMKII [Ca2+/CaM (calmodulin)-dependent protein kinase II] and NMDAR (N-methyl-D-aspartate receptor) are involved in LTP (long-term potentiation) at neuronal synapses that underlies learning and memory . Upon postsynaptic Ca2+ influx, CaMKII gets activated and translocates to the PSD (postsynaptic density) , where it binds to the NR2B subunit of NMDAR [3,4]. This binding anchors CaMKII to the PSD and helps to maintain a separate pool of CaMKII that is located proximal to the neurotransmitter receptors .
NR2B binds to CaMKII at two sites, one of them being the active site, resulting in phosphorylation of Ser1303 of NR2B , and the other is the T-site (Thr286-autophosphorylation site binding pocket), which lies outside the active site . The voltage-gated potassium channel of Drosophila, Eag, which is localized in synapses as well as in axons also interacts with the T-site of CaMKII . The presence of these ligands at the T-site prevents the autoinhibitory domain from occupying the active site, thereby rendering the enzyme autonomously active [7,8].
We have previously shown that the kinetics of phosphorylation of NR2B-Ser1303 (Figure 1) are different to that of classical substrates of CaMKII such as syntide-2 . We have also reported that the NR2B substrate sequence modulates the kinetic parameters of CaMKII for ATP . A previous report has also suggested that the NR2B sequence modulates the catalysis of CaMKII by inhibiting its activity . The present study shows that the catalytic function of CaMKII is modulated by binding of NR2B or Eag to the T-site of CaMKII and involves alterations in the mode of binding of ATP substrate and phosphorylation kinetics in a way that appears to suit the role of CaMKII in synaptic plasticity.
Sequence alignment of NR2A, NR2B and Eag
Insect cell culture medium (IPL-41) was from Sigma Chemicals (St. Louis, MO, U.S.A.). Glutathione–Sepharose and Reacti-Bind™ Glutathione Coated Clear Strip Plate were from Amersham Pharmacia Biotec or from Pierce. Bac-to-Bac baculovirus expression kit was from GIBCO-BRL/Invitrogen. QuikChange Site-Directed Mutagenesis Kit was from Stratagene. Oligonucleotides were obtained from either GIBCO-BRL or from Sigma Genosys. [γ-32P]ATP was from Bhabha Atomic Research Centre (Trombay, Mumbai, India). The monoclonal antibody against the α-subunit of CaMKII was from Affinity Bioreagents (Golden, CO, U.S.A.) or from Sigma Chemicals. Monoclonal anti-phospho-α-CaMKII (pThr286) antibody was from Sigma Chemicals. Anti-GST (glutathione transferase) antibody was from Santa Cruz Biotechnology (Santa Cruz, CA, U.S.A.). The cDNA encoding the α-subunit of CaMKII was a gift from Professor Mary B. Kennedy (Division of Biology, California Institute of Technology, Pasadena, CA, U.S.A.). The cDNAs encoding NR2A and NR2B were gifts from Professor S. Nakanishi (Graduate School of Medicine and Faculty of Medicine, Department of Biological Sciences, Kyoto University, Kyoto, Japan). The cDNA encoding the sequence of the Drosophila voltage-gated potassium channel Eag was a gift from Professor Leslie Griffith (Department of Biology, Brandeis University, Waltham, MA, U.S.A.).
Construction of mutants
Site-directed mutagenesis was carried out using the QuikChange Site-Directed Mutagenesis Kit based on the method of Kunkel . The cDNA of α-CaMKII present in the expression vector pFastBac1-α-CaMKII  that was used for expression of the WT (wild-type) enzyme served as the template to generate mutants. The NR2A/2B sequences present in the pGEX-2T-NR2A/2B expression vectors  were also used as templates to generate mutants. DNA sequencing was carried out by the dideoxy chain termination method in an automated DNA sequencer (ABI Prism 310).
Expression of GST-fusion proteins of NR2A, NR2B and Eag substrate sequences in Escherichia coli
The fusion proteins GST–NR2A (NR2A amino acid residues 1265–1301), GST–NR2B (NR2B amino acid residues 1271–1311) (Figure 1) and their mutants were expressed using the respective construct in the pGEX-2T vector in E. coli [BL21 (DE3) pLys strain] by inducing with 1 mM IPTG (isopropyl β-D-thiogalactoside). The fusion protein GST–Eag (Eag amino acid residues 731–803) (Figure 1) was expressed using the respective construct in pGEX-2T vector in E. coli [BL21 (DE3) pLys strain] by inducing with 0.2 mM IPTG. The cell pellet was lysed by sonication (3×10 s pulses with 10 s intervals) in a lysis buffer containing 40 mM Tris/HCl, pH 7.6, 1 mM DTT (dithiothreitol), 1 mM EDTA, 1 mM EGTA and the protease inhibitors 1 mM PMSF and 10 μg/ml leupeptin or 1×protease inhibitor cocktail (Sigma) and 0.1% Triton X-100. The cell lysate was centrifuged at 9500 g for 10 min. The supernatant, which had a single band on Western blot (Supplementary Figure S1A at http://www.BiochemJ.org/bj/419/bj4190123add.htm) was used as the source of GST-fusion proteins for pull-down assays as well as for phosphorylation reactions.
When required, GST-fusion proteins were purified by affinity chromatography using glutathione–Sepharose 4B in 50 mM Tris buffer (pH 7.5). Proteins were eluted with 50 mM Tris buffer containing 10 mM reduced glutathione (pH 7.5). The purified fusion proteins were nearly homogenous when analysed by SDS/PAGE (Supplementary Figure S1B). Yields of up to 45% could be achieved for purification of fusion proteins.
Total protein in the samples was quantified using the BCA (bicinchoninic acid) method . For quantification of the concentration of the fusion protein, samples with fusion proteins were run on SDS/PAGE  and were stained with Coomassie Blue R-250. The band corresponding to the fusion protein was quantified by densitometry using QuantityOne software (Bio-Rad) and the amount of protein present in the band was determined by comparing with known amounts of bovine serum albumin run on the same gel as standards.
Expression of α-CaMKII using the baculovirus/Sf21 cell system
The recombinant virus containing the cDNA of α-CaMKII was prepared using the Bac-to-Bac Expression system according to the manufacturer's protocol. For purification of the enzyme, monolayer cultures of Sf21 cells in 175 cm2 flasks were infected with viral stocks and were harvested 60–72 h post infection.
Preparation of crude and purified forms of α-CaMKII expressed in the baculovirus/Sf21 cell system
The crude insect cell lysate expressing α-CaMKII and purified α-CaMKII were prepared as described previously  with modifications. The crude lysate, which had a single band of α-CaMKII on Western blot  (Supplementary Figure S2A at http://www.BiochemJ.org/bj/419/bj4190123add.htm), was used as the enzyme source for GST pull-down experiments.
Purification of the enzyme was carried out at 4 °C throughout. The crude lysate was loaded on to a phosphocellulose column, pre-equilibrated with equilibration buffer (50 mM Pipes, pH 7.0, 100 mM NaCl, 1 mM EGTA, 0.5 mM DTT and 1×protease inhibitor cocktail) and was incubated for 30 min. The bound proteins were eluted with 5 times the column volume (approx. 25 ml) of elution buffer (50 mM Pipes, pH 7.0, 500 mM NaCl, 1 mM EGTA, 0.5 mM DTT and 1×protease inhibitor cocktail). The eluate containing CaMKII activity was used for affinity purification on a CaM–Sepharose column as described previously . The purified enzyme preparation showed a single major band of the expected size on SDS/PAGE (Supplementary Figure S2B) and on Western blot (Supplementary Figure S2C). The yield of purified protein in a typical purification was 65%. The specific activity of purified WT α-CaMKII for phosphorylation of the GST–NR2B substrate varied from 0.014 to 0.13 μmol·min−1·mg−1 of protein for the different preparations used in the study. When necessary, the final purified sample was concentrated using Amicon ultracentrifugal devices with a 100 kDa molecular weight cut off by spinning at 5000 g to get the required concentration.
GST pull-down assay
Glutathione–Sepharose 4B beads were washed thoroughly with PBS (10 mM disodium hydrogen phosphate, 1.8 mM potassium dihydrogen orthophosphate, 0.14 M sodium chloride, 2.7 mM potassium chloride, pH 7.4) and were mixed separately with each of the GST-fusion proteins in the presence or absence of Triton X-100 (1% final concentration). The mix was incubated at room temperature (25 °C) for 30 min or at 4 °C for 1 h. The beads were then washed four times with PBS and were resuspended in PBS. Fusion proteins bound to glutathione–Sepharose beads were quantified by SDS/PAGE (10 or 12% gels) as described above. Equal quantities of the different fusion proteins bound to the beads were separately incubated with equal concentrations of α-CaMKII in binding buffer (50 mM Pipes, pH 7.0, 0.1% BSA, 150 mM NaCl and 0.1% Tween 20) with or without 1 mM CaCl2 and 3 μM CaM for 1 h at 4 °C. Beads were then washed four times with PBS, resuspended in SDS sample buffer and were kept in a boiling water bath for 3 min. The beads with bound complexes in SDS sample buffer were then directly loaded on to a gel and were subjected to Western blotting using anti-α-CaMKII antibody and anti-GST antibody. Under our experimental conditions, the binding of α-CaMKII was specific to GST–NR2B and was also Ca2+-dependent (Supplementary Figure S3 at http://www.BiochemJ.org/bj/419/bj4190123add.htm).
Glutathione-coated plates (Reacti-Bind™ Glutathione Coated Clear Strip Plate) from Pierce were also used for pull-down assays instead of glutathione–Sepharose 4B beads. Equal volumes of fusion proteins were added to prewashed wells and were incubated for 1 h at room temperature with shaking. Unbound fusion proteins were removed and the wells were washed three times with PBS. Equal concentrations of crude lysates of insect cells expressing WT α-CaMKII in binding buffer were then added to the wells and were incubated for 1 h at 4 °C with occasional mixing. The unbound protein was removed and the wells were washed with PBS. SDS sample buffer was then added to the wells to retrieve the bound proteins which were then subjected to Western blotting.
Fusion protein phosphorylation
GST fusion proteins of the CaMKII phosphorylation site sequence on NR2B, NR2A and Eag (Figure 1) were used as substrates for CaMKII in an In vitro assay. Reactions were carried out in 20 μl volumes. Each assay tube contained 50 mM Tris/HCl, pH 8.0, 10 mM MgCl2, 12 μM CaM, 0.2 mg/ml BSA, CaCl2-EGTA (0.9 mM CaCl2 in excess of EGTA), various concentrations of [γ-32P]ATP (1000–3000 cpm/pmol) with or without 10 mM DTT and GST–NR2B, GST–NR2A or GST–Eag. In the case of ATP kinetics, the concentration of ATP was varied from 1–400 μM with saturating concentrations of GST–NR2A/NR2B/Eag substrates. The reaction mix was incubated for 1 min at 30 °C. The reaction was initiated by adding the purified enzyme and after 1 min the phosphorylation was terminated by the addition of 5 μl of 5× SDS sample buffer. The sample was denatured, loaded on to SDS/PAGE (10% gel), and the gel was dried and exposed to a phosphor screen. The image obtained by scanning the phosphor screen (Bio-Rad Personal Molecular Imager FX) showed a single major band corresponding to the phosphorylated fusion protein (Supplementary Figure S1C). The image was used to quantify the band intensities using QuantityOne software. The band intensity values showed a linear relationship with the amount of phosphorylated fusion protein (Supplementary Figure S4 at http://www.BiochemJ.org/bj/419/bj4190123add.htm). To quantitatively equate the band intensities with radioactivity in cpm, the following procedure was used. From the stock solution of radioactive ATP that was used for the assay, three different volumes were spotted, after appropriate dilutions, on to Whatman paper as standards. The paper was subjected to Phosphoimager analysis along with the gel. The radioactivity of the standards was then measured in a liquid scintillation counter by Cerenkov counting. Volumetric analysis carried out using Quantity One software gave intensity values for radioactive spots after deducting background. A standard curve was plotted between the values measured using the scintillation counter and those obtained from the phosphoimager analysis. The slope of the resulting straight line was used as the factor for converting band intensity to radioactivity in cpm.
The data were analysed using a Eadie–Hofstee plot and Hill plot. The plots were made in Microsoft Excel software and the values for Vmax, Km, S0.5 and Hill coefficient were obtained from the equation for the best fit straight line generated by the software. The final values for Vmax, Km, S0.5 (substrate concentration at half maximal velocity) and Hill coefficient are presented as the means±S.D. of data from at least three independent experiments. The data were also fitted to the Hill equation using Origin Software which also yielded similar kinetic parameters.
Autophosphorylation of CaMKII
ATP kinetics of Thr286-autophosphorylation of CaMKII was carried out as described before  with different [γ-32P]ATP (1000–5200 cpm/pmol) concentrations ranging from 0.005 to 200 μM. Each assay tube contained 50 mM Tris/HCl, pH 8.0, 10 mM MgCl2, 0.4 mM EGTA, 1.3 mM CaCl2, 17 μM CaM, 0.2 mg/ml BSA and 0.34 μM GST–(S1291A)-NR2A or 0.27 μM GST–(S1303A)-NR2B (both purified) or control (no GST-fusion protein) in a total volume of 20 μl. The assay was performed by first pre-incubating the purified WT α-CaMKII enzyme along with the fusion protein containing the assay mixture for 2 min at 30 °C followed by a 30 s phosphorylation reaction, initiated with the addition of [γ-32P]ATP. The reaction was stopped by the addition of 5×SDS sample buffer and the samples were run on SDS/PAGE. After the staining and destaining processes, the gels were dried and were subjected to autoradiography. The gel bands in the autoradiogram corresponding to phosphorylated CaMKII were quantified as described above. Autophosphorylation at Thr286 was confirmed by Western blotting using an anti-phospho-Thr286-α-CaMKII antibody (Supplementary Figure S5 at http://www.BiochemJ.org/bj/419/bj4190123add.htm) as well as by the appearance of Ca2+-independent activity (results not shown).
Interaction of CaMKII with ATP is influenced by the nature of the protein substrate
The kinetic parameters of CaMKII for ATP were measured for phosphorylation of GST–NR2A and GST–NR2B, two protein substrates that are known to differ in their interaction with CaMKII. The apparent S0.5 and Vmax values for ATP were lower when GST–NR2B was used as the protein substrate (Table 1). However, there was no significant change in the Vmax/S0.5 value. More interestingly, the saturation pattern of ATP for phosphorylation of GST–NR2A was sigmoidal, whereas in the case of phosphorylation of GST–NR2B it was hyperbolic (Figure 2A). Hill coefficient values for ATP also reflected this change (Table 1 and Figure 2B). The Hill coefficient value was approaching 2.0 with GST–NR2A as the protein substrate indicating co-operative behaviour, whereas the value close to 1 with GST–NR2B as the protein substrate indicated non-co-operative behaviour. These results show that the catalytic parameters of CaMKII for ATP differ for phosphorylation of different protein substrate sequences.
|Protein substrate||S0.5 (μM)||Vmax (μmol·min−1·mg−1 of protein)||Vmax/S0.5||Hill coefficient|
|Protein substrate||S0.5 (μM)||Vmax (μmol·min−1·mg−1 of protein)||Vmax/S0.5||Hill coefficient|
Estimation of kinetic parameters of ATP for phosphorylation of GST–NR2A and GST–NR2B substrates by CaMKII
Mutations that affect the T-site-mediated interaction between CaMKII and GST–NR2B also affect catalysis
CaMKII is reported to employ an ordered substrate-binding mechanism where ATP binds first to the free catalytic site of the enzyme followed by protein substrate [17,18,19]. Hence differences in the interaction of the two protein sequences at the catalytic site are unlikely to be the cause for the observed differences in kinetic parameters for ATP. We investigated whether the binding of NR2B to the T-site of CaMKII could be causing a modulation of the interaction of ATP with the enzyme. To test this, we measured the kinetic parameters for ATP after disrupting the T-site-mediated interaction between the CaMKII and NR2B sequence. The phosphorylation reaction was conducted with mutants of GST–NR2B or CaMKII that were defective in the T-site-mediated interaction. The mutants used were L1298A-NR2B  and I205K-α-CaMKII . Using the GST pull-down assay, which represents binding at the T-site, we demonstrated that binding exists between WT forms of CaMKII and GST–NR2B or GST–Eag (Figure 3A), whereas it was defective in the case of GST–NR2A (Figure 3A), as well as for the mutants (Figures 3B and 3C). Both the mutants, GST–(L1298A)-NR2B (Figure 3B) and I205K-α-CaMKII (Figure 3C) exhibited significantly attenuated binding to their WT binding partner in the pull-down assay. When GST–(L1298A)-NR2B was phosphorylated by WT α-CaMKII, the kinetic parameters for ATP exhibited a shift towards those observed for phosphorylation of GST–NR2A by WT α-CaMKII. It was observed that the values of S0.5 and Vmax were higher compared with phosphorylation of WT GST–NR2B (Figure 4). The Hill coefficient value was also higher, showing resemblance with that of GST–NR2A (Figure 4). Similar changes were observed when phosphorylation of GST–NR2B by I205K-α-CaMKII was compared with that by WT α-CaMKII (Figure 5). The apparent values of S0.5 and Vmax showed increases along with an increase in the Hill coefficient value when phosphorylation of GST–NR2B was carried out by I205K-α-CaMKII. Interrupting the T-site-mediated stable association either by mutating the NR2B sequence or by mutating CaMKII thus seemed to result in the shifting of kinetics of phosphorylation towards that observed for GST–NR2A. This suggested that binding of NR2B at the T-site of CaMKII exerts a modulatory effect resulting in the altered kinetic parameters observed for phosphorylation of GST–NR2B.
GST pull-down of α-CaMKII
Phosphorylation of WT or mutant L1298A of GST–NR2B by WT α-CaMKII
Phosphorylation of GST–NR2B by WT or mutant I205K of α-CaMKII
Modulation by NR2B is mediated through ATP-binding residues
A conserved glutamate residue, Glu96 of α-CaMKII, has been attributed a key role in binding ATP based on molecular simulation studies using the crystal structure of the catalytic domain of CaMKII . We found that mutation of Glu96 to alanine impaired the modulation of catalysis by NR2B (Table 2), even though its binding to the T-site of CaMKII is not affected by the mutation, as seen by binding in the pull-down assay (Figure 3D), indicating that Glu96 is involved in mediating the modulation of catalysis by NR2B.
|Protein substrate||S0.5 (μM)||Vmax (μmol·min−1·mg−1 of protein)||Vmax/S0.5||Hill coefficient|
|Protein substrate||S0.5 (μM)||Vmax (μmol·min−1·mg−1 of protein)||Vmax/S0.5||Hill coefficient|
In WT CaMKII, modulation of catalysis due to the binding of NR2B sequence to the T-site results in the relatively higher value for the ratio GST–NR2A/GST–NR2B for S0.5 and Vmax values (Table 2). However, in the case of E96A mutant, there was a clear reduction in the GST–NR2A/GST–NR2B ratios for S0.5 and Vmax values (Table 2). The S0.5 value for ATP with GST–NR2B was higher for the E96A mutant compared with WT α-CaMKII, although it was largely unaffected with GST–NR2A. The Vmax values for phosphorylation of GST–NR2A and GST–NR2B were similar with the E96A mutation. These changes indicate that modulation of catalysis by GST–NR2B bound at the T-site is impaired in the E96A mutant, suggesting that Glu96 plays a role in the modulation of catalysis by the NR2B sequence.
NR2B modulates kinetics of Thr286-autophosphorylation
We studied the effect of NR2B sequence on the Thr286-autophosphorylation of α-CaMKII. Interestingly, we find that ATP saturation does not show co-operativity in the case of Thr286 autophosphorylation of CaMKII (Table 3), indicating major differences between the kinetic mechanisms of the autophosphorylation and substrate phosphorylation reactions. The kinetic parameters for the autophosphorylation reaction also differed significantly from those of the substrate phosphorylation reaction. The kinetic parameters for ATP were measured in the presence of GST–(S1303A)-NR2B and were compared with the reaction carried out in the presence of GST–(S1291A)-NR2A or in the absence of any protein substrate modulators. Although the presence of NR2A sequence did not seem to affect the kinetic parameters, the presence of NR2B sequence exerted a significant influence. It could be seen that the apparent values of Km and Vmax were decreased in the presence of GST–(S1303A)-NR2B, whereas the Vmax/Km value showed an increase (Table 3). These effects on Km and Vmax were similar to the observations made in the case of substrate phosphorylation (Table 1).
|Modulator||Km (μM)||Vmax (μmol·min−1·mg−1 of protein) (×10−3)||Vmax/Km (×10−3)||Hill coefficient|
|Modulator||Km (μM)||Vmax (μmol·min−1·mg−1 of protein) (×10−3)||Vmax/Km (×10−3)||Hill coefficient|
Activity of CaMKII is enhanced by NR2B at low ATP concentrations
Interestingly, when the ATP concentration dependence of the activity of α-CaMKII was closely examined, it was observed that at low ATP concentrations the activity of CaMKII was higher in the presence of NR2B than in the presence of NR2A. This could be observed in the case of both substrate phosphorylation as well as autophosphorylation. The non-phosphorylatable GST–(S1303A)-NR2B was first allowed to bind to α-CaMKII by incubating them together in the presence of Ca2+/CaM and the complex was then used to phosphorylate GST–NR2A. As seen in Figure 6, the activity of α-CaMKII was higher when it was pre-treated with GST–(S1303A)-NR2B compared with the control in which the enzyme was pre-treated with GST–(S1291A)-NR2A. When phosphorylation of GST–NR2A and GST–NR2B were compared, it could be seen that the activity was higher for phosphorylation of GST–NR2B at concentrations of ATP below 5 μM (Supplementary Figure S6 at http://www.BiochemJ.org/bj/419/bj4190123add.htm). In the presence of GST–NR2B, the enzyme is saturated at relatively lower concentrations of ATP and the activity stays constant for a large range of higher ATP concentrations. In contrast, the phosphorylation of GST–NR2A continues to increase to much higher levels as the concentration of ATP is increased. The autophosphorylation activity was also higher in the presence of GST–(S1303A)-NR2B compared with the activity in the presence of GST–(S1291A)-NR2A at ATP concentrations lower than 1 μM (Supplementary Figures S7 and S8 at http://www.BiochemJ.org/bj/419/bj4190123add.htm). In the presence of GST–(S1303A)-NR2B, the autophosphorylation activity also stayed constant over a broader range of ATP concentrations (Supplementary Figure S9 at http://www.BiochemJ.org/bj/419/bj4190123add.htm). These results very clearly showed that the NR2B sequence can enhance the activity of α-CaMKII at low concentrations of ATP and maintain it at a constant level over a wide range of ATP levels.
Modulation of the substrate phosphorylation activity of CaMKII by the NR2B sequence motif
Eag potassium channel sequence exerts a lesser degree of modulation on α-CaMKII compared with NR2B
A sequence motif in the Eag potassium channel of Drosophila that is homologous to the phosphorylation site on NR2B (Figure 1) is known to bind α-CaMKII in the GST pull-down assay (Figure 3A), similar to GST–NR2B . To see whether binding of Eag to the T-site of α-CaMKII also causes modulation of the activity of CaMKII, we investigated the kinetics of phosphorylation of GST-Eag by WT α-CaMKII. The kinetic parameters for phosphorylation of GST-Eag showed moderate differences compared with those of GST–NR2B (Figure 7 and Supplementary Table S1, Experiment A, at http://www.BiochemJ.org/bj/419/bj4190123add.htm). The Hill coefficient value for ATP was 1.3. The values of S0.5 and Vmax were also marginally higher compared with GST–NR2B. The binding of Eag to the T-site of Drosophila CaMKII is known to be disrupted by mutating Ile206 to lysine (equivalent to the I205K mutation in rat α-CaMKII) . The kinetic parameters for phosphorylation of GST–Eag by I205K-α-CaMKII were higher compared with WT α-CaMKII (Supplementary Table S1). The increased S0.5 and Hill coefficient values for I205K-α-CaMKII indicated a loss of catalytic modulation, since T-site binding was defective. This showed that binding at the T-site by the Eag sequence also can modulate catalysis by α-CaMKII.
Phosphorylation of GST–NR2B and GST–Eag by WT α-CaMKII
We compared the kinetic parameters of ATP for phosphorylation of GST–NR2B and GST–Eag by I205K-α-CaMKII (Figure 8 and Supplementary Table S1, Experiment B). Except for the apparent Vmax value, the Hill coefficient and S0.5 values showed very little differences between GST–NR2B and GST–Eag. This showed that, when binding in the GST pull-down assay is defective, as in the case of I205K-α-CaMKII, the two substrates did not modulate the kinetic parameters for ATP.
Phosphorylation of GST–NR2B and GST–Eag by I205K-α-CaMKII
The affinities of GST–NR2B and GST–Eag towards the T-site of α-CaMKII were compared by competition experiments. Binding of GST–NR2B and GST–Eag to α-CaMKII were subjected to competition by a peptide with the sequence of the binding site of NR2B (NR2B 17-mer, Figure 1). As seen in Figure 9(A), the peptide competes out GST–Eag more effectively than GST–NR2B. The IC50 value (concentration required for 50% inhibition) of the NR2B 17-mer was measured (Figure 9B) and was found to be lower in the case of pull-down with GST–Eag. This indicated that the NR2B sequence has a higher affinity for α-CaMKII than Eag.
NR2B binds to the T-site of CaMKII with higher affinity than Eag
Interaction of CaMKII with NR2B has been suggested to regulate the enzymatic activity of CaMKII by making it Ca2+/CaM independent . It has also been reported that the NR2B sequence causes uncompetitive inhibition of CaMKII with respect to ATP substrate . Our results bring out further novel features of the interaction between CaMKII and NR2B. We show that interaction of NR2B sequence at the T-site of CaMKII causes a shift in the saturation pattern of ATP from co-operative to non-co-operative. Although the Vmax and S0.5 values for ATP were decreased by the presence of NR2B, there were no significant changes in the observed Vmax/S0.5 ratio, a parameter which could be taken as a measure of catalytic efficiency (Table 1).
Both ordered and random substrate-binding mechanisms have been proposed for CaMKII [17,18,19,22], although ordered binding is more favoured . Comparing the results obtained for WT and mutant forms establishes that the kinetics of substrate phosphorylation is altered by the binding of the protein substrate to the T-site of CaMKII (Figures 4 and 5). The mutations that we used, L1298A in NR2B and I205K in α-CaMKII, might have additional influences on the catalytic parameters of phosphorylation. For example, the residue Leu1298 has been reported to be a specificity determinant for the catalytic site of CaMKII . When this amino acid is mutated to a less hydrophobic residue, the recognition by the catalytic site could also become hampered in addition to the loss of binding at the T-site. Therefore the observed Vmax would be a result of both of these events, leading to a decrease in the Vmax/S0.5 ratio (Figure 4), whereas the Vmax/S0.5 ratio of GST–NR2A was closer to GST–NR2B (Table 1). The I205K mutant of α-CaMKII exhibited a higher Vmax/S0.5 ratio compared with WT α-CaMKII (Figure 5), which indicates that the mutation may have additional effects on catalysis other than disrupting the binding of NR2B at the T-site.
The effect of the E96A mutation (Table 2) shows that the catalytic modulation by NR2B is mediated in a major way through Glu96. In WT α-CaMKII, the observed lowering of Km and Vmax values consequent to binding of GST–NR2B to the T-site might require reorientation of Glu96 to a position favourable for interaction with ATP. Absence of this mechanism due to lack of Glu96 in the E96A mutant may result in a loss of regulatory control, which leads to escalation of the Km and Vmax values. The mutation causes a decrease in the Vmax value for GST–NR2A, indicating a marginal impairment of the catalytic process of CaMKII. On the whole, the present data on the E96A mutant shows that in the CaM-activated state, binding of NR2B at the T-site further modulates catalytic parameters of ATP through Glu96. This reveals a novel structural mechanism involving Glu96, and probably helix-αD that harbours Glu96, to regulate ATP binding in CaMKII.
We find that kinetic parameters for autophosphorylation at the Thr286 residue of α-CaMKII differ significantly from those of exogenous substrate phosphorylation (Tables 1 and 3). The binding of ATP to the enzyme was non-co-operative for the autophosphorylation reaction, whereas it was co-operative for substrate phosphorylation in the absence of any T-site modulators. Since in the autophosphorylation reaction two adjacent CaM-bound subunits of CaMKII form an enzyme–substrate pair, it is possible that the ATP-binding characteristics of the enzyme could differ from those of the substrate phosphorylation reaction. The autophosphorylation reaction is also modulated by the NR2B sequence, as seen by the changes in kinetic parameters (Table 3). Since CaMKII present in the PSD can be in complex with NR2B , it is likely that its activity for both substrate phosphorylation, as well as autophosphorylation, may be modulated, as observed in our In vitro experiments.
Although the changes in the apparent S0.5 and Vmax values observed for substrate phosphorylation (Table 1) as well as for Thr286 autophosphorylation (Table 3) appear to resemble uncompetitive inhibition by GST–NR2B with respect to ATP, the effect on Hill coefficient for substrate phosphorylation indicates that the observed phenomenon is not classical uncompetitive inhibition. Moreover, in uncompetitive inhibition, the enzyme activity measured in the presence of the inhibitor will always be less than that in the absence of the inhibitor . However, we would like to highlight the fact that both the Thr286-autophosphorylation activity as well as substrate phosphorylation activity were consistently higher in the presence of the NR2B sequence than in its absence for the lower range of ATP concentrations used in our study (Figure 6, Supplementary Figures S6, S7 and S8). These data suggest that, at low concentrations of ATP, binding of NR2B to the T-site of CaMKII facilitates its interaction with ATP probably by increasing its affinity for ATP. This phenomenon could result from allosteric modulation. Since the NR2B-modulated enzyme is saturated at comparatively lower concentrations of ATP (Supplementary Figure S6), the activity remains constant over a broad range of ATP concentrations compared with activity in the absence of NR2B. A mechanistic explanation of such a regulation of CaMKII would require further investigation. An earlier study reported that a peptide with the NR2B sequence causes uncompetitive inhibition of CaMKII with respect to ATP . The fusion protein GST–NR2B, used in the present study, had additional sequences compared with the peptide used in the earlier study and might differ in its interaction at the T-site of CaMKII, probably due to difference in the three-dimensional structure attained by its binding motif.
Although modulation of CaMKII by NR2B at the PSD may be necessary to support synaptic plasticity, CaMKII bound to Eag on axons may conceivably serve physiological functions of a different nature. This is probably being reflected by the difference in the extent of allosteric modulation exerted on CaMKII by NR2B and Eag. Our experiments show that the Eag sequence binds to CaMKII with a lower affinity compared with NR2B (Figure 9). This could explain the minor differences in kinetic parameters such as S0.5 and Hill coefficient between NR2B and Eag (Figure 7 and Supplementary Table S1, Experiment A). The Vmax obtained for GST–NR2B was significantly higher compared with GST–Eag for I205K-α-CaMKII-mediated phosphorylation (Figure 8 and Supplementary Table S1, Experiment B), indicating that the NR2B sequence is a better substrate for phosphorylation at the catalytic site. This characteristic of the NR2B sequence was probably not revealed in the WT enzyme due to the unique modulation caused by the binding of NR2B at the T-site. This becomes relevant in kinetic analyses of mutated substrate sequences aimed at identifying specificity determinants for the CaMKII catalytic site. The outcome of such studies might be affected by the ability of those sequences to bind the T-site, which in turn would influence the kinetic parameters.
The crystal structure of the CaMKII catalytic domain dimer is consistent with the observed co-operative mode of binding of calmodulin . The co-operative saturation by calmodulin could in turn induce co-operativity in the saturation of ATP, based on earlier reports that ATP and calmodulin reciprocally enhance the binding of each other . This, however, is unlikely since we have used calmodulin at saturating concentrations for all the kinetic experiments. Interestingly, the crystal structure of CaMKII also suggests that the ATP-binding pocket has features suitable for allosteric modulation . Co-operative binding of ATP  suggests that there could probably be additional events of structural communication between subunits of a dimer upon binding of ATP. The modulatory effects of binding of NR2B at the T-site perhaps bypass these co-operative interactions between the subunits by elevating them all to a state of higher affinity for ATP.
Dendritic spine heads, especially in the cortex, have very few mitochondria to provide ATP. Hence ATP requirement in the PSD is suggested to be met by glycolytic source endogenous to PSD which is regulated by the levels of NAD+, glyceraldehyde-3-phosphate and nitric oxide . Consequently, fluctuations in ATP levels in the PSD are likely to occur. The autophosphorylated state of CaMKII in the CaMKII/PP1 (protein phosphatase 1) switch that has been proposed as the storehouse of synaptic memory [24,27], has to be maintained against the action of phosphatases by continuous autophosphorylation at more or less constant rates even when the concentration of ATP varies. The enhanced activity of NR2B-bound CaMKII at low concentrations of ATP and its saturation by a minimal increase in concentration of ATP ensure a constant level of activity of CaMKII over a wide range of ATP concentrations (Supplementary Figures S6 and S9). Moreover, the reduced activity of NR2B-bound CaMKII at higher ATP concentrations (Supplementary Figures S6 and S9) helps to maintain the stable state of the CaMKII/PP1 switch in an energy efficient manner by consuming minimal ATP as proposed previously . Thus the modulation of CaMKII activity by NR2B reported in the present study could aid in the role of CaMKII as a molecular switch in synaptic memory.
We thank Professor Mary B. Kennedy, Professor S. Nakanishi, Professor Leslie Griffith, Dr K. Santhosh Kumar (Department of Molecular Medicine and Cancer Biology, Rajiv Gandhi Centre for Biotechnology, Kerala, India) and Dr S. Leena (Department of Molecular Medicine and Cancer Biology, Rajiv Gandhi Centre for Biotechnology, Kerala, India) for providing vectors and peptides. We thank Dr R. V. Thampan, Dr M. R. Das, Dr Moinak Banerji and Mr Manoj, P. for help during the course of this work.
Ca2+/calmodulin-dependent protein kinase II
protein phosphatase 1
substrate concentration at half maximal velocity
Thr286-autophosphorylation site binding pocket
This work was supported by the Rajiv Gandhi Centre for Biotechnology, Department of Science and Technology, Government of India and Council of Scientific and Industrial Research, Government of India. P. K. K., R. K. R. and P. M. received fellowships from the Council of Scientific and Industrial Research of the Government of India. J. C. and S. P. S. received fellowships from the University Grants Commission of the Government of India.
These authors contributed equally to this work.