Neurogranin stimulates Ca2+/calmodulin-dependent kinase II by suppressing calcineurin activity at specific calcium spike frequencies

Abstract

Calmodulin sits at the center of molecular mechanisms underlying learning and memory. Its complex and sometimes opposite influences, mediated via the binding to various proteins, are yet to be fully understood. Calcium/calmodulin-dependent protein kinase II (CaMKII) and calcineurin (CaN) both bind open calmodulin, favoring Long-Term Potentiation (LTP) or Depression (LTD) respectively. Neurogranin binds to the closed conformation of calmodulin and its impact on synaptic plasticity is less clear. We set up a mechanistic computational model based on allosteric principles to simulate calmodulin state transitions and its interactions with calcium ions and the three binding partners mentioned above. We simulated calcium spikes at various frequencies and show that neurogranin regulates synaptic plasticity along three modalities. At low spike frequencies, neurogranin inhibits the onset of LTD by limiting CaN activation. At intermediate frequencies, neurogranin facilitates LTD, but limits LTP by precluding binding of CaMKII with calmodulin. Finally, at high spike frequencies, neurogranin promotes LTP by enhancing CaMKII autophosphorylation. While neurogranin might act as a calmodulin buffer, it does not significantly preclude the calmodulin opening by calcium. On the contrary, neurogranin synchronizes the opening of calmodulin's two lobes and promotes their activation at specific frequencies. Neurogranin suppresses basal CaN activity, thus increasing the chance of CaMKII trans-autophosphorylation at high-frequency calcium spikes. Taken together, our study reveals dynamic regulatory roles played by neurogranin on synaptic plasticity, which provide mechanistic explanations for opposing experimental findings.

Fig 1. Reaction Diagram involving CaM and its binding partners.

Binding interactions and state transitions involving calcium, CaM and CaM-binding proteins. For simplicity, only the binding of the first calcium ion is included. The reactions shown here are applied to all other calcium-bound forms, with any combination of filled calcium-binding sites. a and b represents calcium binding on CaM N lobe; c and d represent calcium binding on the C lobe. R indicates open lobes; T closed lobes. Dissociation coefficients for calcium at specific binding sites are specified, as well as state transition constants (ratios between opposite state transition rates). Identical dissociation constants for protein binding CaM are only written once. Parameter values are included in Table 1.

Fig 2. Saturation curves of CaM by calcium.

Values were obtained by scanning a range of initial calcium concentrations and running simulations until steady-state is reached, using initial conditions described in Hoffman et al. 2014 [26]. a) shows only the saturation of C lobe (within the context of the entire CaM) while b) shows the whole CaM, without any substrates (grey), with Ng (blue), non-phosphorylated CaMKII (orange), CaN (green) or Thr286-phospho-CaMKII (red). The x-axis shows the free concentration of calcium and not the initial concentration used for the scan. [CaM] = 5 μM; [Ng/CaMKII/CaMKIIp/CaN] = 50 μM. Solid line: simulation results; dots: experimental observations.

Fig 3. Kinetics of calcium dissociation from CaM.

Calcium dissociation from the C lobe of CaM was simulated upon mixing calcium chelator EGTA (10⁻² M) after the model simulation reached steady states with pre-mixed CaM (10 μM) and calcium (100 μM) with (cyan) or without (salmon) Ng (50 μM). Solid line: simulation results; dots: experimental observations [26].

Fig 4. Free calcium concentration at various input frequencies.

Intracellular free calcium elevation simulated with a single calcium input (a) or a train of inputs (b) at 1 Hz (red), 5 Hz (blue) and 10 Hz (grey). Each calcium spike represent the addition of 1926 ions over 8 ms.

Fig 5. Calcium binding and CaM conformational changes in response to calcium input frequencies.

Calcium binding CaM was followed during simulations using trains of 300 spikes at different frequencies. Plots show calcium saturation levels at the C and N lobe and the corresponding CaM conformational changes, with CaM on its own, or in the presence of Ng. Red dots represent the end of calcium stimulation. [CaM]_tot = 40 μM, [Ng]_tot = 40 μM, basal [Ca] = 0.08 μM.

Fig 6. AUCs of CaM conformations in response to calcium input frequencies.

Conformational changes of CaM (40 μM) without Ng (a) or with Ng ([Ng]_tot = 40 μM) (b). Basal activities were subtracted before AUCs were calculated.

Fig 7. Change of protein activity in response to calcium spikes.

Simulations in absence (a,b) or presence (c,d) of Ng, from equilibrium and during stimulation by 300 calcium spikes at 10 Hz (a,c) and 30 Hz (b,d). The protein activities were defined as follow: fraction of CaMKII monomers bound to CaM and/or phosphorylated multiplied by CaMKII kcat for GluR1, fraction of CaN bound to CaM multiplied by CaNA kcat for GluR1, fraction of Ng bound to CaM. [CaM]_tot = 40 μM, [CaMKII]_tot = 80 μM, [CaN]_tot = 8 μM, and [Ng]_tot = 40 μM; kcat_CaMKII = 2 s^-1, kcat_CaN = 0.5 s^-1. All the concentrations of perspective CaM binding proteins are assigned according to the proteomic study in the hippocampus CA1 region [64].

Fig 8. CaMKII and CaN activation in response to repeated high-frequency calcium spikes.

Computational models were simulated, with (a) or without (b) Ng, by 300 calcium spikes organized into three 100-spike discrete bursts at 100 Hz each, separated by 10 min intervals. Each calcium input was as described in Fig 3. Both protein activities were normalized to their total concentration, then multiplied by their catalytic constant for GluR1. [CaM] = 40 μM, [Ng] = 40 μM, [CaN] = 8 μM, [CaMKII] = 80 μM; kcat_CaMKII = 2 s^-1; kcat_CaN = 0.5^-1.

Fig 9. Activity of CaMKII and CaN in response to calcium spike frequencies.

Protein activities, as described in Fig 7, were integrated over time. The effect on synaptic plasticity was calculated by subtracting CaN’s activated area from CaMKII (a). Each protein’s activated area was also plotted as a function of calcium input frequency (b). Models were stimulated either with (grey circle) or without Ng (black cross). All protein concentrations were as described in Fig 7 and in Table 1.

Fig 10. CaN mediates Ng’s effect on CaMKII activation.

Models were simulated without CaN (a) or without CaN’s phosphatase activity (kcat = 0)(b) or with double amount of CaN (c). The combined response of CaMKII and CaN to calcium spike frequencies were quantified as described in Fig 9: with Ng (grey circle) and without (black cross). Simulation conditions and concentrations are as described in Fig 7. Double amount of CaN equals to 16 μM.

Fig 11. CaN concentration affects Ng regulation of CaMKII activation.

Different CaN concentrations were applied in the presence and absence of Ng. The difference between CaMKII’s activated areas reached in response to a 100 Hz calcium spike frequency in presence and absence of Ng were plotted against CaN concentration. The relationship was further fitted into a polynomial function (method loess() from ggplot2) and plotted as the blue line. Red points are the values coming from our simulations. Shaded light grey area represents the 95% confidence interval for the fitting.

Fig 12. Ng concentration affects synaptic plasticity.

Computational models with various Ng concentrations were stimulated by 300 calcium spikes at various frequencies. The concentrations of proteins, except Ng, were listed in Table 1 and remain unchanged. In particular, the concentration of CaM is 40 μM. The activated areas of CaMKII and CaN were calculated and their net effects on AMPA receptor phosphorylation were calculated as described in Fig 7, and plotted as a function of calcium spike frequency.