Protection of α-CaMKII from Dephosphorylation by GluN2B Subunit of NMDA Receptor Is Abolished by Mutation of Glu96 or His282 of α-CaMKII

Interaction of CaMKII and the GluN2B subunit of NMDA receptor is essential for synaptic plasticity events such as LTP. Synaptic targeting of CaMKII and regulation of its biochemical functions result from this interaction. GluN2B binding to the T-site of CaMKII leads to changes in substrate binding and catalytic parameters and inhibition of its own dephosphorylation. We find that CaMKIINα, a natural inhibitor that binds to the T-site of CaMKII, also causes inhibition of dephosphorylation of CaMKII similar to GluN2B. Two residues on α-CaMKII, Glu96 and His282, are involved in the inhibition of CaMKII dephosphorylation exerted by binding of GluN2B. E96A-α-CaMKII is known to be defective in GluN2B-induced catalytic modulation. Data presented here show that, in both E96A and H282A mutants of α-CaMKII, GluN2B-induced inhibition of dephosphorylation is impaired.


Introduction
Many mechanisms exist for the stable storage of information in living systems. Activity dependent strengthening of neuronal synapse leads to long term potentiation (LTP), a cellular mechanism required for learning and memory. The activation of NMDA receptor (NMDAR) and GFP-α-CaMKII expressed in HEK-293 cells was also used in the experiments. WT and mutants, H282A and K21A, of GFP-α-CaMKII were used. The cells were lysed in RIPA buffer or in solution containing 50 mM Pipes (pH 7.0), 2 mM DTT, 1 mM PMSF and protease inhibitor cocktail.

Expression of fusion proteins
CaMKIINα inhibitor protein, S1291A-GluN2A mutant (with amino acid sequences 1265-1301of GluN2A) and S1303A-GluN2B mutant (with amino acid sequence 1271-1311of GluN2B) were expressed in BL21(DE3)pLys strain of E. coli cells as GST fusion proteins. S1291A-GluN2A and S1303A-GluN2B mutants are the non-phosphorylatable forms of GluN2A and GluN2B sequences. These non-phosphorylatable forms are used to rule out any substrate phosphorylation occurring during the reaction and interfering with the binding. The expression and purification were carried out as published earlier [27,38]. The expressed proteins were purified using glutathione-sepharose column and were subsequently subjected to gel filtration to remove glutathione. S1 Fig shows the preparations of GST-S1291A-GluN2A, GST-S1303A-GluN2B and GST-CaM-KIINα. GluN2B sequence containing the amino acids 1271-1311 was cloned into pET-32a vector to obtain His-GluN2B with His tag expressed at the C-terminal.
The concentrations of proteins were estimated using BCA (Bicinchoninic acid) method [39].

GST pulldown assay
CaMKII was incubated with either GST-GluN2B or GST-CaMKIINα or GST-GluN2A in binding buffer (50 mM Pipes, pH 7.0, 0.1% BSA, 150 mM NaCl and 0.1% Tween-20) either in presence or absence of 1 mM CaCl 2 and 1 μM calmodulin for 1 hour at 4°C. The reaction mix was allowed to bind for 1 hour at 4°C to glutathione-sepharose beads which were washed in 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). Alternatively, PBS washed glutathione-sepharose beads were first incubated with the GST-fusion proteins in PBS for 1 hour at 4°C, washed 4 times with PBS and were subsequently incubated with CaMKII in binding buffer for 1 hour at 4°C. After incubation, the glutathione-sepharose beads were again washed with PBS four times and were suspended in SDS sample buffer. The samples were heated in a boiling water bath for 4 minutes and were subjected to SDS-PAGE, followed by Western blot analysis using anti-α-CaMKII antibody and anti-GST antibody [40,41]. Blots were developed either by colour reaction of alkaline phosphatase conjugated secondary antibody or by enhanced chemiluminescence using horse radish peroxidase conjugated secondary antibody. Densitometric quantitation of the blots were done after capturing images of the blots, using BioRad Quantity One software. In competition experiments, CaMKII was incubated with His-GluN2B and GST-CaMKIINα simultaneously.
In the pulldown assay of WT and E96A of α-CaMKII purified after expression in insect cells and WT, H282A and K21A expressed as GFP-α-CaMKII in HEK-293 cells, GST-GluN2B was first bound with washed glutathione-agarose beads. The GST-GluN2B bound beads were washed 4 times with PBS. Equal amounts of GST-GluN2B bound beads were allowed to bind to the WT or mutants of CaMKII. The autophosphorylated CaMKII samples for pulldown were prepared as described in the section 2.5 below except for using nonradioactive ATP. The pulldown was done under the same reaction conditions as that of the dephosphorylation experiments.

Autophosphorylation of Thr 286 -α-CaMKII and its dephosphorylation
Autophosphorylation of CaMKII was carried out in a reaction mix containing 50 mM Tris, pH 8.0, 10 mM MgCl 2 , 10 mM DTT, 1 μM γ 32 P-ATP, 3 mM CaCl 2 , 9 μM calmodulin and 0.164 mg/ml of purified WT-α-CaMKII. The reaction mix was preincubated at 30°C for 1 min and autophosphorylation was carried out for 30 seconds after initiating the reaction with ATP addition. The reaction was stopped with either SDS-PAGE sample buffer or with 10 μM staurosporine. For dephosphorylation experiments, the reaction mix after staurosporine addition was incubated with either 10 μM GST or 5.7 μM GST-CaMKIINα in presence of 22 μM calmodulin and 1 mM CaCl 2 for 20 minutes at 4°C. GST was used as control for GST-CaMKIINα. The dephosphorylation was carried out at 30°C in presence of reaction buffer containing 50 mM HEPES, 100 mM NaCl, 2 mM DTT, 0.01% Brij 35, pH 7.5, 1 mM MnCl 2 and 0.001 mg/ml PP1 for 20 minutes. The dephosphorylation was stopped by adding SDS sample buffer. The concentration of CaMKII after staurosporine addition was 0.162 mg/ml, after GST-CaMKIINα addition was 0.082 mg/ml and before PP1 addition was 0.065 mg/ml. Staurosporine concentration after adding GST or GST-CaMKIINα to the reaction mix was 4.9 μM and before adding PP1 was 4 μM.
In the experiments to find the rate of dephosphorylation, autophosphorylation of CaMKII was carried out in a reaction mix containing 50 mM Tris, pH 8.0, 10 mM MgCl 2 , 10 mM DTT, 1 μM γ 32 P-ATP, 3 mM CaCl 2 , 9 μM calmodulin andCaMKII. For CaMKII expressed in HEK-293 cells 22 μM calmodulin was used. The concentrations of purified CaMKII used for autophosphorylation were 0.0081 mg/ml and 0.0087 mg/ml respectively for WT and E96A. The concentrations of HEK-293 cell lysates used were 2.21 mg/ml for WT, 1.46 mg/ml for H282A and 1.35 mg/ml for K21A. The reaction mix was preincubated at 30°C for 1 min and autophosphorylation was carried out for 5 minutes for WT and 20 minutes for E96A expressed in insect cells. For GFP-α-CaMKII expressed in HEK-293 cells, reaction was carried out for 5 minutes each for WT, H282A and K21A. The reaction was stopped with 10 μM staurosporine and non-radioactive ATP was added to a final concentration of 1 mM. The reaction mix was incubated with either 10 μM of GST-S1291A-GluN2A or GST-S1303A-GluN2B in presence of 22 μM calmodulin and 1 mM CaCl 2 for 20 minutes at 4°C. For dephosphorylation, a buffer of pH 7.5 was added which contributed the following components at the indicated final concentrations: 50 mM HEPES, 100 mM NaCl, 2 mM DTT, 0.01% Brij 35 and 1 mM MnCl 2 . PP1 was added to a final concentration of 0.002 mg/ml and the reaction mix was incubated for the indicated times. For CaMKII expressed in HEK-293 cells, 0.0013-0.0033 mg/ml PP1 was used for dephosphorylation. The dephosphorylation was stopped by adding SDS sample buffer. The concentrations of the enzymes before PP1 addition were 0.0041 mg/ml and 0.0043 mg/ml for WT and E96A respectively and 1.11 mg/ml, 0.73 mg/ml, 0.68 mg/ml respectively for WT, H282A and K21A. Staurosporine concentration during incubation with GST-S1291A-GluN2A or GST-S1303A-GluN2B was 6.2 μM and before adding PP1 to the reaction mix was 5 μM. The reaction mix composition after addition of staurosporine became 49.5 mM Tris, 9.9 mM MgCl 2 , 9.9 mM DTT, 2.97 mM CaCl 2 ; after addition of non-radioactive ATP became 48

GST-CaMKIINα inhibits dephosphorylation of CaMKII
As reported earlier [34,42] we also find that CaMKII binds to GST-CaMKIINα inhibitor protein similar to its binding to GST-GluN2B ( Fig 1A). We find that the binding of GST-CaMKIINα to CaMKII after 20 min incubation is tight and irreversible as His-GluN2B, even after one hour of incubation, could not replace the bound GST-CaMKIINα ( Fig 1B). In a competition experiment, His-GluN2B competes with GST-CaMKIINα for the T-site of CaM-KII ( Fig 1B). Since CaMKIINα binds to the T-site of CaMKII similar to GluN2B, we checked whether it influences dephosphorylation of phospho-Thr 286 of α-CaMKII. Phosphorylation at Thr 286 and its dephosphorylation by PP1 were demonstrated under our experimental conditions using antibody specific to phospho-Thr 286 -α-CaMKII (S2 Fig). For dephosphorylation experiments, autophosphorylation of CaMKII, carried out using γ 32 P-ATP, was stopped with staurosporine to prevent any further kinase reaction. It was found that the dephosphorylation of CaMKII by PP1 was reduced in presence of GST-CaMKIINα inhibitor protein (Fig 2).

Glu 96 and His 282 are key residues involved in mediating the GluN2B induced effects on CaMKII dephosphorylation
The dephosphorylation of CaMKII had been shown to be inhibited by GST-S1303A-GluN2B [9]. Here we show that the rate of dephosphorylation of WT-CaMKII in presence of GST-S1303A-GluN2B was slower compared to that in presence of GST-S1291A-GluN2A ( Fig  3 and S5 Fig). The rate of dephosphorylation is calculated from the slope of the plot of phosphorylation level against the time. About three fold decrease in the rate of dephosphorylation of WT-CaMKII was observed in presence of GST-S1303A-GluN2B. GSTpulldown assay was performed as described in methods. Lane 1: GST-pulldown of CaMKII with GST-CaMKIINα. Lane 2: CaMKIINα was allowed to bind to CaMKII first, followed by incubation with His-GluN2B. Lane 3: GST-CaMKIINα and His-GluN2B were allowed to bind to CaMKII simultaneously. C: The mean ± standard deviation of densitometric values of the band intensities obtained from three independent experiments as shown in B are plotted. The binding of CaMKII with GST-CaMKIINα taken as the control is considered as 100% binding. The binding of CaMKII in 2 and 3 are presented as the relative binding with respect to control in the bar graphs. The p value for the difference between 1 and 3 is less than 0.05 in twotailed student's t-test.
The position of Glu 96 of α-CaMKII or its analogous residues in other isoforms of CaMKII in the active and autoinhibited states suggests that this residue is involved in ATP binding. The movement of the D-helix in the active structure swings Glu 96 towards ATP binding site which is positioned away from the ATP binding site in the autoinhibited state [43,44]. In the CaM-KIIδ/Ca 2+ /CaM complex Glu 97 (Glu 96 in α-CaMKII) is suitably positioned to coordinate with ATP [44]. We have previously shown that in the E96A-α-CaMKII mutant, GluN2B bindinginduced modulation of catalysis that is seen in WT-α-CaMKII, is impaired [27] showing that Glu 96 has the additional role of mediating the GluN2B induced structural changes. E96A-α-CaMKII binds to GluN2B although the extent of binding was modestly different compared to WT (S6 Fig). The binding to GluN2B by E96A was comparable to WT after autophosphorylation at Thr 286 (S7 Fig). When the dephosphorylation of E96A-α-CaMKII was analysed, it was found that inhibition of dephosphorylation by GluN2B seen in WT was almost completely absent in the E96A mutant indicating that the GluN2B binding induced alterations in the structure of CaMKII did not occur in the mutant (Fig 3 and S5 Fig). This suggested that the same structural changes mediated by Glu 96 of α-CaMKII upon binding of GluN2B may be responsible for catalytic modulation [27] as well as reduction of Thr 286 dephosphorylation rate.
We checked the dephosphorylation rate of another mutant, H282A-α-CaMKII. His 282 is a residue involved in anchoring the D-helix of α-CaMKII [43]. H282A mutation is reported to bring about interesting biochemical consequences in CaMKII [45,46]. Since His 282 is also involved in maintaining the structure of D-helix similar to Glu 96 , we anticipated functional anomalies in the H282A-α-CaMKII mutant similar to the E96A mutant. We found that H282A-α-CaMKII exhibited Ca 2+ -independence for its autophosphorylation activity (S2 Fig) as well as for binding to GluN2B (data not shown). However its GluN2B binding showed specificity when compared to GluN2A (S3 Fig). Moreover, the mutant was comparable to WT for its binding to GluN2B whether it was autophosphorylated or not (S6 and S7 Figs). When Thr 286 dephosphorylation was studied, it was found that H282A-α-CaMKII also behaved like E96A mutant and showed similar rate of dephosphorylation with or without GluN2B (Fig 4  and S5 Fig). Lys 21 is a residue in the ATP binding region. GluN2B binds to K21A-α-CaMKII mutant in its autophosphorylated and non-autophosphorylated states as seen in the pulldown assay (S4, S6 and S7 Figs). The K21A-α-CaMKII mutant showed decrease in the rate of dephosphorylation in presence of GluN2B compared to that in presence of GluN2A (Fig 4 and  S5 Fig) showing that Lys 21 does not appear to be involved in the GluN2B-mediated effects and that the effects seen with Glu 96 and His 282 are specific.

Discussion
CaMKII functions as a molecular switch supporting memory in PSD and a fine balance between the kinase and phosphatase is required for optimal functioning of the switch [13][14][15]. As reported earlier, we see that CaMKIINα binds to the T-site of CaMKII [34]. We also show that the binding of CaMKIINα inhibitor protein is irreversible and persistent even after calcium removal. GluN2B sequence inhibits the binding of CaMKIINα to CaMKII indicating that both could be interacting at the T-site. CaMKIINα may be involved in a feedback mechanism in CaMKII signaling in memory formation [47].
Lisman et al hypothesized that CaMKII-PP1 system may be acting as a bistable molecular switch forming the basis for molecular memory [13]. Both LTP and LTD, the two forms of synaptic plasticity are mediated by phospho-Thr 286 -CaMKII autonomous activity [48]. CaMKII--GluN2B interaction is required for LTP induction as well as maintenance of synaptic strength [49]. GluN2B interaction induced catalytic modulation of CaMKII stabilizes its activity against fluctuations in ATP concentrations [9,27]. Restrained dephosphorylation of phospho-Thr 286 -CaMKII when it is bound to GluN2B in vitro or in PSD [9,50] is considered as a fundamental characteristic of the CaMKII-GluN2B complex that allows it to support the maintenance of LTP [30]. In the crystal structure of the truncated CaMKII, regulatory region of one subunit binds to the active site of the neighbouring subunit. The N-terminal portion of the regulatory region (amino acids 276-291) of one subunit interacts with the C-lobe of the neighbouring kinase subunit. In the active state conformation, Arg 283 -Leu 291 of the regulatory region of the subunit acting as substrate binds to the active site of the enzyme subunit [51]. Arg 283 interacts electrostatically with an acidic region in the enzyme subunit positioning Thr 286 into the active site [51]. It has been proposed that Thr 286 of α-CaMKII in the phosphorylated state might also occupy the active site of the neighbouring subunit. When phospho-Thr 286 is bound to the active site it could be inaccessible to phosphatase. GluN2B interaction might facilitate this binding thereby reducing dephosphorylation by PP1 [30]. His 282 being close to Arg 283 , is likely to be involved in electrostatic interactions with docking sites in the acidic region of the enzyme subunit. Mutation of His 282 to Ala might destabilize these interactions thereby exposing phospho-Thr 286 to the action of phosphatase. Glu 96 mutation to Ala might also perturb this interaction and expose phospho-Thr 286 to phosphatase consistent with our data presented here. Indeed PP1 has been shown to be a suppressor of learning and memory [52]. Hence in addition to GluN2B, other endogenous T-site binding proteins that may also regulate the accessibility of the phosphatase to CaMKII could play a supportive role for the proper functioning of the CaMKII-PP1 switch. Our finding reported in the current manuscript that CaMKIINα inhibits dephosphorylation of phospho-Thr 286 -CaMKII suggests that CaMKIINα may thus support the CaMKII-PP1 molecular switch.
The dual consequences of the binding of GluN2B to the T-site of CaMKII described above, i.e. catalytic modulation and inhibition of dephosphorylation, could be the result of the same structural mechanism. We have previously shown that in the E96A mutant of α-CaMKII, the catalytic modulation is impaired eventhough the mutant binds GluN2B [27] indicating that Glu 96 could be playing a role in the associated structural changes. The present study shows that this mutant is defective in exhibiting the GluN2B-induced reduction in dephosphorylation also (Fig 3) suggesting that the same structural mechanisms could be responsible for both the effects. In the inactive state of CaMKII the regulatory segment interacts to the helix αD which positions Glu 96 away from the ATP binding site [43]. In the active state as the regulatory segment disengages, helix αD moves towards ATP so that Glu 96 (Glu 97 in δ-CaMKII) interacts with the hydroxyl groups of ribose rings of ATP [44]. We hypothesise that Glu 96 of CaMKII might be participating in essential interactions that stabilize the GluN2B-induced altered conformation. His 282 is also involved in anchoring D helix of α-CaMKII [43]. As expected, H282A-α-CaMKII showed similar results like E96A mutant in the regulation of dephosphorylation. It appears that modulation of catalysis of CaMKII as well as inhibition of dephosphorylation of phospho-Thr 286 -CaMKII are both consequences of the same set of structural changes in CaMKII that are induced by binding of ligands at the T-site.
This study reveals the commonality and cooperation of the functions of two key residues of α-CaMKII, Glu 96 and His 282 and also provides insights for further investigations on the regulation of CaMKII functions by GluN2B. The data on CaMKIINα also suggests additional physiological roles for this inhibitory protein of CaMKII. We note that the band intensity in the CaMKII blot for the mutant is lesser than expected. For non-autophosphorylated samples ATP was added after SDS sample buffer addition. ATP-dependent autophosphorylation was observed. The lower panel shows the same blot probed with antiα-CaMKII antibody. There is only slight band shift between autophosphorylated and non-autophosphorylated samples indicating that no sites other than Thr 286 site is autophosphorylated under the reaction conditions. E96A-α-CaMKII band was seen at slightly higher position than WT-α-CaMKII. B: The upper panel shows the Western blot of the non-autophosphorylated and autophosphorylated HEK-293 cell lysates expressing GFP-WT-α-CaMKII (28 μg), GFP-K21Aα-CaMKII (35 μg) and GFP-H282A-α-CaMKII (28 μg) probed with anti-phospho-Thr 286 -α-CaMKII antibody. Since HEK-293 cells have endogenous ATP as reported before [53] negative control for autophosphorylation was carried out without Ca 2+ . The lysates without any incubation were also directly loaded. H282A-α-CaMKII shows Ca 2+ independent activity. The lower panel shows the same blot probed with anti-α-CaMKII antibody. C: Protein phosphatase 1 dephosphorylates phospho-Thr 286 -α-CaMKII in vitro. Western blot shows CaMKII-Thr 286 autophosphorylation and its dephosphorylation by PP1. Data represents two experiments.