Genetic Inhibition of CaMKII in Dorsal Striatal Medium Spiny Neurons Reduces Functional Excitatory Synapses and Enhances Intrinsic Excitability

Ca2+/calmodulin-dependent protein kinase II (CaMKII) is abundant in striatal medium spiny neurons (MSNs). CaMKII is dynamically regulated by changes in dopamine signaling, as occurs in Parkinson's disease as well as addiction. Although CaMKII has been extensively studied in the hippocampus where it regulates excitatory synaptic transmission, relatively little is known about how it modulates neuronal function in the striatum. Therefore, we examined the impact of selectively overexpressing an EGFP-fused CaMKII inhibitory peptide (EAC3I) in striatal medium spiny neurons (MSNs) using a novel transgenic mouse model. EAC3I-expressing cells exhibited markedly decreased excitatory transmission, indicated by a decrease in the frequency of spontaneous excitatory postsynaptic currents (sEPSCs). This decrease was not accompanied by changes in the probability of release, levels of glutamate at the synapse, or changes in dendritic spine density. CaMKII regulation of the AMPA receptor subunit GluA1 is a major means by which the kinase regulates neuronal function in the hippocampus. We found that the decrease in striatal excitatory transmission seen in the EAC3I mice is mimicked by deletion of GluA1. Further, while CaMKII inhibition decreased excitatory transmission onto MSNs, it increased their intrinsic excitability. These data suggest that CaMKII plays a critical role in setting the excitability rheostat of striatal MSNs by coordinating excitatory synaptic drive and the resulting depolarization response.


Introduction
The striatum is the major input nucleus of the basal ganglia [1]. Dysfunction in this region is associated with drug addiction, Parkinson's disease and other disorders [2,3,4,5,6,7,8]. The striatum is primarily composed of projection GABAergic medium spiny neurons (MSNs) that integrate glutamatergic excitatory transmission with modulatory dopaminergic transmission. Since MSN firing is thought to be driven primarily by excitatory drive, understanding the basic mechanisms of glutamatergic transmission onto MSNs is necessary to understand how the striatum functions in health and disease.
Calcium-calmodulin-dependent kinase II (CaMKII) is a Ser/ Thr kinase that is highly expressed in the striatum, constituting ,0.7% of total striatal protein [9]. CaMKII assembles into dodecameric complexes that in the striatum predominantly contain CaMKIIa and CaMKIIb isoforms [10]. As a major constituent of the postsynaptic density (PSD) in the dorsal striatum [11] as well as other forebrain regions [12,13], CaMKII is activated by N-methyl-D-aspartate-receptor (NMDAR)-mediated calcium influx [14,15,16]. CaMKII is a key modulator of hippocampal and cortical pyramidal cell glutamate synapse function [17,18,19]. CaMKII can phosphorylate many downstream substrates including the ionotropic glutamate receptors NMDARs and a-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptors (AMPARs) [20,21,22,23,24,25]. Indeed, in hippocampal pyramidal cells, CaMKII activation enhances synaptic trafficking of AMPARs and channel function [26,27,28,29]. In addition, a constitutively active form of CaMKII can decrease intrinsic excitability of hippocampal neurons as well as MSNs in the nucleus accumbens shell [30,31]. While much is known about the role of CaMKII at glutamate synapses on glutamatergic projection neurons such as hippocampal and cortical pyramidal neurons, relatively little is known for GABAergic cells. Indeed, little CaMKII is expressed in GABAergic interneurons [32,33,34], making GABAergic projection cells such as MSNs, which are highly enriched in CaMKII, unique targets for studying the role of CaMKII in synaptic transmission and intrinsic excitability.
Previous studies have implicated striatal CaMKII in Parkinson's disease (PD) and addiction. CaMKII is hyperactivated after striatal dopamine depletion, and CaMKII inhibition rescued striatal synaptic plasticity and motor deficits found in animal models of Parkinson's disease [35]. Striatal CaMKII regulates motivational effects of reward cues on goal-directed behaviors [36] as well as curbing D1R-mediated cocaine hyperlocomotion [37] and modulating excitability following chronic cocaine administration [31]. Thus, a better understanding of CaMKII's role in striatal glutamatergic synaptic transmission may suggest new approaches to treat PD and addiction.
In addition to its postsynaptic roles, CaMKII modulates a variety of presynaptic functions, including trafficking of synaptic vesicles [38,39,40,41,42], P/Q type calcium channels [43,44,45], voltage-gated sodium channels [46,47], catecholamine synthesis [48,49] and dopamine transporter function [50,51]. Thus, an investigation of the role of CaMKII within striatal MSNs requires a cell-specific approach. To accomplish this, we generated a transgenic mouse line that expresses a CaMKII inhibitory peptide selectively within dorsal striatal MSNs. Using this line, we found that CaMKII inhibition in dorsal striatal MSNs leads to a loss of functional glutamatergic synapses and an increase in intrinsic excitability. These findings shed light on the neural mechanisms underlying the development of striatal neural circuits, learning and memory, and motor behavior.

Generation of EAC3I-4 transgenic mice
For generation of double transgenic EAC3I-4 X tTA animals, heterozygous transgenic mice carrying the tTA gene driven by an alpha CaMKII promoter fragment were bred to heterozygous mice carrying the EAC3I transgene fused to EGFP driven by the tetO promoter. The CaMKIIa-tTA mice were obtained from Dr. Eric Kandel's lab and maintained at Vanderbilt University. The autocamtide-3 derived inhibitory peptide (EAC3I) sequence (KKALHRQEAVDAL) mimics the autoinhibitory region of the CaMKII regulatory domain (residues 278-290) and acts by competitively binding to the catalytic site. In in vitro biochemical assays AC3-I blocks the phosphorylation of an autocamtide-2 substrate by purified rat CaM kinase with an IC 50 of 3 mM [52,53], with a $100-fold reduced potency toward protein kinase C, CaM kinase I or CaM kinase IV [52,54,55]. EAC3-I is made up of the AC3-I peptide fused N-terminal to enhanced green fluorescent protein (EGFP) to stabilize and mark cellular and tissue distribution. In a previous study, EAC3I was transgenically expressed in the heart and total CaMKII activity in extracts was reduced by <40% [56]. This level of inhibition is likely to be a substantial underestimate of in vivo inhibition, because proteins were diluted upon homogenization due to mosaic transgene expression.
We quantified the level of mosaicism in our EAC3-I mouse by staining with a NeuN antibody (1:1000, Millipore) to label all neurons (See supplemental for detailed immunohistochemical labeling). Manual counts of the number of EGFP positive neurons versus total number of NeuN stained neurons in z-stacks in the dorsal lateral striatum were made in Metamorph (Molecular Devices; Sunny Vale, CA), providing an estimate of the percent of cells expressing the transgene. TetO-linked transgene expression is controlled using mouse chow containing 200 mg/kg Doxycycline (DOX) (Bio-Serv; Frenchtown, NJ). For DOX rescue experiments pregnant dams were fed DOX and weaned pups continued with the same food. At 6 weeks DOX was removed and the transgene was allowed to be expressed for 4-5 weeks. All DOX recordings were made between 10-11 weeks. All mice had been inbred onto a C57BL/6 background for more than seven generations. GluA1 knockout mice (Andrew Holmes Lab) and wildtype littermates 8-16 weeks of age were utilized [57].

Whole-Cell Voltage Clamp Recordings
MSNs of the dorsal lateral striatum were directly visualized with infrared video microscopy (Olympus BX51WI with QImaging Rolera-XP Camera). Only highly expressing EGFP-containing MSNs were selected for study and compared to neighboring MSNs visually devoid of EGFP expression. Recording electrodes (3-6 MV) were pulled on Flaming-Brown Micropipette Puller (Sutter Instruments) using thin-walled borosilicate glass capillaries (WPI). EPSCs were evoked by local fiber stimulation with bipolar nichrome electrodes. Stimulating electrodes were placed on the border of the corpus callosum and dorsal lateral striatum 100-300 mm dorsal to the recorded neuron, and electrical stimulation (5-20 V with 100-150 ms duration, Grass Instruments) was applied at 0.05 Hz unless otherwise noted. This location most likely stimulates both cortical and thalamic glutamatergic axons onto MSNs. Recording electrodes (3-6 MV) were filled with (in mM) Cs+ gluconate (117), HEPES (20), EGTA (0.4), TEA (5), MgCl 2 (2), ATP (4), GTP (0.3) pH 7.35, 285-290 mOsm. Series resistance averaging 16 MV (ranging 8-30 MV) was monitored and experiments with changes greater than 20% were omitted. AMPAR EPSCs and sEPSCs were isolated by adding 25 mM picrotoxin and recording at a holding potential of 270 mV in normal ACSF. To isolate mEPSCs 1 uM TTX was added in addition to sEPSCs recording conditions. In all experiments a time period of at least 5 minutes post break in was allowed for internal solution exchange and stabilization of membrane properties. GluA1KO sEPSC recordings were conducted similarly except recording ACSF contained 50 mM picrotoxin and internal solution contained (in mM) 120 CsMeSO 3 , 5 NaCl, 10 TEA-Cl, 10 HEPES, 5 QX-314, 1.1 EGTA, 0.3 Na-GTP, and 4 Mg-ATP, 295-300 mOsm. GluA1 WT and KO littermate mice were 8-16 weeks of age at time of recording. In PPR experiments evoked 100-200 pA responses were elicited with the interstimulus interval set at 40 ms, 50 ms and 60 ms. For MK-801 experiments NMDA currents were pharmacologically isolated (25 mM picrotoxin, 10 mM NBQX) and held at +40 mV while a stable ten minute baseline (0.05 Hz) was acquired. After a stable baseline was acquired the stimulator was switched off and 10 mM MK-801 was washed on for seven minutes. Following the wash-in period the stimulator was turned on (0.1 Hz) and the time constant for decay of the integral of the NMDA current was calculated using nonlinear regression one-phase decay. For baclofen modulation of PPR, 20 sweeps of approximately 200 pA EPSCs (0.05 Hz) were collected to establish a baseline. Following a wash-in of 10 uM baclofen for 8 minutes, an additional 20 sweeps were collected. Additional 20 sweeps were taken at 10 minutes and 20 minutes post washout. CV was calculated by dividing the SD of the amplitude of the evoked EPSCs by the mean. For rectification experiments, evoked AMPAR-mediated EPSCs were isolated in 25 mm picrotoxin and 100 mm DL-APV containing aCSF while the voltage was stepped from 270 mV to +40 mV in 10 mV steps. 0.1 mM spermine was included in the standard cesium internal solution to avoid dialysis of endogenous polyamines. The rectification index (RI) was calculated as the ratio of the amplitude of AMPAR-mediated currents evoked at 270 mV over +40 mV. All signals were acquired via a Multiclamp 700B amplifier (Axon Instruments), digitized at 10 kHz, filtered at 2 kHz and analyzed via pClamp 10.2 software (Axon Instruments). Holding current and series resistance were all monitored continuously throughout the duration of experiments. Experiments in which changes in series resistance were greater than 20% were not included in the data analysis.
Statistical analyses were performed using Graphpad Prism 5.04. Two-tailed unpaired Student's t-test (t) were used unless variance differed significantly (Bartlett's test for equal variances) then nonparametric Mann-Whitney (U) tests were used. One or Two-way analysis of variance (ANOVA) (F) were used when indicated with Neuman-Keuls Multiple Comparison post hoc test. Non-parametric Kruskal-Wallis tests (H) were used with Dunn's Multiple Comparison post hoc test when variances differed significantly (Bartlett's test for equal variances). All values given are presented as average 6 SEM. Cumulative probability plots were analyzed with Kolmogorov-Smirnov (KS) test.

Whole cell current clamp recordings
Slices were prepared as before, but perfused with ACSF containing (in mM): NaCl (124), NaH 2 PO 4 (1.25), KCl (2.5), CaCl 2 (2.5), MgSO 4 (2), NaHCO 3 (26), Glucose (11) pH = 7.35, 300-305 mOsm. Recording electrodes were filled with in (mM): K+ gluconate (120), NaCl (4), HEPES (10), Mg-ATP (4), Na-GTP (0.3), KCl (20), Na+ Phosphocreatine (10) pH = 7.3, 285-290 mOsm. MSNs were identified by their intrinsic membrane properties (i.e. resting membrane typically more negative than 280 mV, inward and outward rectification in response to somatic positive and negative current injections, and a long depolarizing ramp to delayed first spike discharge [58]. Recordings were rejected if the initial Vm was more positive than 275 mV. Resting membrane potential or zero current potential was determined right upon break-in. Spontaneous excitatory postsynaptic potentials after 5 minutes post break in were recorded for 4 minutes with the cell dynamically current clamped at 285 mV. Resting membrane potential was monitored and current was injected to maintain the resting potential at 285 mV. For current-voltage (IV) relationships positive or negative current injections were given in 20 pA steps until the cell reached threshold and fired a single AP. Five additional current injections steps (20 pA each) were given above threshold. Input resistance was monitored throughout the experiment and the cell was rejected if the input resistance changed by more than 20%. Healthy cells showed APs that crossed +30 mV and stable resting membrane potentials.
Lucifer yellow intracellular fills, confocal imaging and measurement of dendritic structure Transgenic EAC3I mice (3-4 months) were perfused with 10 ml room temperature phosphate buffered saline (PBS) followed by 100 ml of 4% paraformaldehyde solution delivered over 10-20 minutes. Brains were post-fixed in ice cold paraformaldehyde solution for an additional hour before 200 mm thick coronal sections of the precommissural striatum were prepared on a vibrating microtome (Leica; Buffalo Grove, IL). MSN cell bodies in dorsolateral striatum were visualized at 406. Randomly selected EAC3I-expressing (EGFP-positive) MSNs and non-fluorescent (EGFP-negative) MSNs in the dorsal lateral striatum were iontophoretically filled with an 8% solution of Lucifer yellow (LY; Sigma-Aldrich, in 50 mM Tris-HCl, pH 7.4) using hyperpolarizing current (3-5 nA for 8-10 minutes). Slices were fixed in 4% paraformaldehyde/PBS overnight at 4uC and then coverslipped using Prolong Gold mounting solution (Invitrogen; Grand Island, NY).
Digital images of MSN dendritic segments located 80-100 mm distal to the cell soma were captured using a Zeiss LSM 710 confocal microscope with 663 oil immersion objective and 62.5 digital zoom. Spine density of three to four dendritic segments emanating from different primary dendrites was averaged to yield a mean spine density value per MSN. A total of 17 EAC3Inegative MSNs and 16 EAC3I-positive MSNs were analyzed; these were obtained from four different mice of each genotype. After the confocal images were coded by someone not involved in the study, another person unaware of the animal or genotype of the MSN being examined used Imaris (Version 5.5; Bitplane) to quantify dendritic spine density. A three-dimensional perspective in ''surpass'' mode of the software package was generated and images were processed with background subtraction thresholding and smoothed with a Gaussian filter. Dendritic segments were modeled with the largest and smallest diameters set at 2 mm and 1 mm, respectively. Florescence in each dendritic segment was thresholded manually to capture all dendritic spines. The minimum terminal point spine diameter was set at 0.143 mm and the florescence contrast threshold was set at 1. Identified spines were counted and marked in 3D on a rotating version of the image. Finally, each structure identified as a spine by the Imaris software was visually verified. In order to determine total dendritic length and generate a Sholl analysis (number of dendritic branches intersecting circular rings drawn around the soma every 20 mm distal to the soma), we used Neurolucida Explorer (MicroBright-Field; Willington, VT).

Characterization and in vivo localization/expression of CaMKII inhibitory (EAC3I) peptide
To determine the role CaMKII plays in modulating glutamatergic transmission onto MSNs in the dorsal lateral striatum, we generated a transgenic mouse model with striatally-enriched expression of a CaMKII inhibitory peptide fused to enhanced green fluorescent protein (EGFP) referred to hereafter as EAC3I [52]. The EAC3I peptide inhibits all isoforms of CaMKII, as well as both calcium-dependent and independent forms of the kinase, and the fusion with EGFP allows for visualization of the regional and cellular distribution of the transgenically expressed protein.
EAC3I was previously utilized in another transgenic line to examine the role of CaMKII in the heart [56]. To spatially and temporally regulate EAC3I expression, the tetracycline transactivator (tTA) is driven by a CaMKIIa promoter fragment with the tTA gene product driving expression of EAC3I ( Figure 1A). Constitutive expression of the EAC3I transgene has no overt effect on viability or on gross brain morphology, and is silenced by including doxycycline (DOX, 200 mg/kg) in the animals' chow for two weeks ( Figure 1B). The alpha CaMKII promoter normally restricts transgene expression to the forebrain, but this founder line exhibits enrichment of EAC3I expression in the MSNs of the dorsal striatum ( Figure 1C, 1D), presumably due to integration site-dependent effects. Little to no expression of the inhibitor was seen in cortex and thalamus ( Figure 1C, 1D). The EAC3I peptide is expressed in a mosaic pattern throughout the dorsal striatum, with 3464% of neurons containing the inhibitory peptide ( Figure 1D). The mosaicism observed is common in transgenic animals and different founder lines utilizing the alpha CaMKII promoter fragment to drive transgene expression show differing expression patterns throughout the forebrain [59,60]. However, we observed no overlap of signal between EGFP positive neurons and striatal cholinergic and GABAergic interneuron markers such as ChaT, parvalbumin, NPY, and calretinin ( Figure S1). At higher magnification the EAC3I inhibitory peptide was observed in the MSN cell soma, dendrites and dendritic spines ( Figure 1E). Dense expression of EAC3I was detected in both the globus pallidus and in the substantia nigra pars reticulata ( Figure 1C, 1F, 1H), but higher magnification images revealed the staining was localized to axons ( Figure 1F), demonstrating that both indirect and direct pathway MSNs contain the inhibitory peptide ( Figure 1C, 1F, 1H).
In adult mice no differences in sEPSC amplitudes were observed between MSNs from the four groups described above (in pA, Wt Figure 2B, 2E], suggesting that this effect is independent of presynaptic excitability. The use of the tTA expression system provides DOX-dependent control of EAC3I expression ( Figure 1B). We suppressed expression of the transgene by supplementation of food with DOX (200 mg/kg) for both the dam and her litter until 6 weeks of age, and then recorded from EAC3I-expressing MSNs 4 weeks after removal of DOX. Under these conditions, the global expression of EAC3I was markedly lower than in non-DOX exposed mice, as has been previously noted with the tTA expression system [61].  Figure 2C, 2F]. These data indicate that the sEPSC frequency phenotype observed in no-DOX EAC3I MSNs is not likely due to an insertion site artifact.

MSN CaMKII inhibition reduces excitatory transmission independently of changes in release probability
The canonical interpretation of a reduction in s/mEPSC frequency is via a reduction in the probability of glutamate release or the number of release sites/number of synapses. In order to better understand the mechanism(s) underlying the reduction in s/ mEPSC frequency in EAC3I MSNs we first examined pairedpulse ratios (PPR) of evoked EPSCs, a measurement that inversely correlates with neurotransmitter release probability [62].  Figure 3A, 3B, 3C]. To show that we could predictably manipulate PPR we used the GABA B R agonist baclofen (10 mM). Baclofen acts presynaptically to reduce the probability of release and therefore increase PPR at a number of CNS synapses [62,63]. Baclofen increased the PPR in EAC3I MSNs to a similar extent as in control MSNs suggesting that release probability was modifiable in the CaMKII-  Figure 3F].
Changes in s/mEPSC frequency could also theoretically be produced by changes in synaptic glutamate concentration. To address this possibility, we utilized MK-801, an uncompetitive, activity-dependent and irreversible antagonist of NMDARs [64]. After obtaining a stable baseline of evoked NMDAR-mediated EPSCs at +40 mV in the presence of picrotoxin and NBQX, the stimulator was switched off and 10 mM MK-801 was applied to the slice. After eight to ten minutes to allow the drug to equilibrate in the bath the stimulator was turned back on and the subsequent rate of inhibition of the NMDAR-mediated EPSC was calculated. The rate of inhibition was not significantly different between non-EGFP and EAC3I MSNs (tau in seconds, NON EGFP 80.8269.65; EGFP 87.0768.59) [t (9) = 0.6469; p = NS; Figure 4A, 4B], suggesting that the levels of glutamate at both synapses is not significantly different. Together, the lack of effect of CaMKII inhibition on PPR, baclofen modulation or MK-801 rate of blockade point to a postsynaptic mechanism for the decreased sEPSC frequency in the CaMKII-inhibited cells, which we interpret as a decrease in the number of functional synapses.

CaMKII inhibition does not alter dendritic spine density, but reduces dendritic length and complexity
We next examined dendritic spine density, dendritic length and branching complexity in CaMKII-inhibited versus neighboring non-inhibited cells. Previous work has suggested a correlation between changes in mEPSC frequency and dendritic spine density [65,66,67,68], although this is not always the case [69]. Additionally, CaMKII has been shown to modulate dendritic length in the hippocampus [70]. There was no difference in spine density related to expression of the transgene (NON EGFP 18.0360.52 spines per 10 mm vs.; EGFP 17.9560.85) [t (31) = 0.087; p = NS; Figure 5A, 5B]. However, we observed a significant reduction in total dendritic length (NON EGFP 1538699 mm vs.; EGFP 1134678) [t (31) = 3.158; p = 0.0035; Figure 5C, 5D]. Sholl analysis revealed a significant overall decrease in dendritic branching (F (1,31) = 28.55; p,0.0001), with Bonferroni post-hoc analyses revealing specific significant decreases at 40 and 60 mm distal to the cell soma ( Figure 5E).

GluA1 KO mimics EAC3I decrease in sEPSC frequency
The AMPAR GluA1 subunit is critical for activity-dependent postsynaptic strengthening of excitatory synapses, and is inserted into the synaptic membrane in a CaMKII-dependent process in hippocampal neurons [26]. Mice lacking the GluA1 subunit of the AMPAR have deficits in CA1 LTP [57] (but see [71]) and deficits in learning and memory [72,73,74,75]. Additionally, reductions in CaMKIIa mRNA and protein levels are seen in the hippocampus of GluA1 KO animals [76]. If the CaMKII inhibition-dependent reduction in functional glutamatergic synapses on striatal MSNs is due to defects in synaptic GluA1 insertion, then we predicted that GluA1 knockout mice should mimic EAC3I mice in terms of sEPSC frequency and amplitude. Thus, we measured sEPSC    Figure 6A, 6C]. The similarity in synaptic outcomes of the GluA1 KO and EAC3I expression reinforces the idea that a common pathway has been affected.

MSN CaMKII inhibition leads to enhanced intrinsic excitability
To further examine the impact of CaMKII inhibition on physiological responses of dorsal striatal MSNs, we next examined excitatory drive and excitability of these cells under current clamp conditions. As expected from our voltage clamp experiments with sEPSCs, we also observed a robust decrease in the frequency of sEPSPs in the CaMKII inhibited cells versus control neurons in current clamp (in Hz, Wt 3.  Figure 7F]. This suggests that CaMKII inhibition moves the resting membrane potential closer to firing threshold and increases membrane resistance to enhance the propagation of depolarizing current from distal MSN dendrites. Next we examined the voltage responses to differing current injections. While injecting minimal current to maintain resting membrane potential at 285 mV, a series of hyperpolarizing and depolarizing current injections were given in 20 pA steps. The threshold for 1 st AP or rheobase current injection was significantly lower in the CaMKII inhibited MSNs versus control (in pA, Wt 357632; tTA 285628; NON EGFP 357630; EGFP 194617) [F (3, 78) = 9.393; p,0.0001; Figure 7G]. The firing threshold was not significantly different amongst groups (in mV, Wt 236.461.3; tTA 235.361.5; NON EGFP 233.460.8; EGFP 237.060.9) [F (3, 78) = 2.516; p = 0.0649], suggesting that input resistance changes are a major contributor to changes in rheobase. Also CaMKII inhibited cells exhibited increased spiking over a range of suprathreshold current injections [F (3,60) = 5.425; p = 0.0023; Figure 7H] versus control, reflecting a decrease in the interspike interval. Taken together, these data show that CaMKII inhibition enhances the intrinsic excitability of MSNs.

Discussion
We present converging lines of evidence that dorsal striatal MSN CaMKII inhibition decreases functional synapse number and increases intrinsic excitability. Inhibition of CaMKII in MSNs leads to a decrease in sEPSC frequency, without a change in release probability, glutamate levels at the synaptic cleft or dendritic spine density. These observations are consistent with a decrease in the number of functional synapses. In addition to changes in excitatory transmission, inhibition of CaMKII leads to an enhancement of MSN intrinsic excitability. These data suggest that CaMKII coordinates opposing regulation of excitatory transmission and intrinsic excitability in MSNs, serving as a cellular rheostat.
CaMKII inhibitors such as KN62 and KN93 have been useful tools in probing CaMKII functions, but these drugs also inhibit voltage-gated K + and Ca 2+ channels [77,78], and do not inhibit the autonomous activity of Thr286-autophosphorylated CaMKII [79,80]. In addition, in dendritic spines where the concentration of calmodulin and CaMKII are extremely high (,100 mM) [81,82], KN-62 (10 mm) only partially decreases CaMKII activity [18]. The EAC3I peptide we used inhibits all isoforms of CaMKII, including CaM-stimulated and autonomous activity, with low micromolar potency [52,56,83]. EAC3-I is $100-fold selective for CaMKII over protein kinase C, CaM Kinase I or CaM kinase IV [52,54]. The AC3-I peptide sequence differs from another highly selective CaMKII-inhibitor, AIP, by only one amino acid residue [55]. It is also important to consider the localization of CaMKII inhibition when interpreting these results. CaMKII is highly expressed in dopamine terminals, which densely innervate the striatum, where it stimulates dopamine efflux via the dopamine transporter in the presence of amphetamine [51]. In addition, CaMKII is present in glutamatergic projections, which form the presynaptic terminal onto MSN spines and dendrites [33,51], where it may modulate release events [39,43,44,45]. Our transgenic strategy resulted in the selective expression of the CaMKII inhibitor in the postsynaptic MSN, where it cannot directly affect the function of the glutamatergic and dopaminergic terminals, consistent with the lack of change in glutamate release parameters in EAC3I MSNs (Figures 3, 4).
We demonstrated that CaMKII inhibition decreases s/mEPSC frequency with no changes in presynaptic function. The lack of a bimodal distribution in the s/mEPSC frequency data suggests that CaMKII inhibition similarly affects both direct and indirect pathway MSNs. Changes in s/mEPSC frequency are traditionally interpreted as alterations in presynaptic quantal content; the product of changes in release probability or synapse number. However, multiple lines of evidence suggest that presynaptic function is unaltered. Together these data suggest that CaMKII inhibition decreases functional synapse number. One possibility is that CaMKII inhibition produces a loss of functional synaptic connections. Alternatively, there could be an increase in the number of silent synapses which contain NMDARs but no AMPARs and are typically abundant early in development [84,85,86,87,88,89]. A further possibility is that CaMKII inhibition could increase the numbers of silent modules of synapses or increase the number of AMPAR-lacking subregions of the synapse due to local nature of basal synaptic transmission [90,91]. In the hippocampus, silent synapses can be unsilenced following NMDAR activation by the introduction of new AMPAR to the synapse, underlying a common mechanism of LTP of synaptic glutamatergic transmission [26,27,92,93]. Previous research has suggested a direct role for CaMKII in the unsilencing of synapses in the hippocampus [94,95,96,97]. Additionally, CaMKII is required for the formation of new synapses and/or morphological growth following hippocampal LTP induction [18,98,99,100].
Striatal CaMKII inhibition did not alter dendritic spine density, suggesting that the decrease in s/mEPSC frequency could best be explained by some spines lacking active presynaptic terminals or increased numbers of silent synapses. However, the significant increase in the input resistance of EAC3I-positive MSNs should enhance sampling of mEPSCs from more distal dendritic sites, potentially increasing s/mEPSC frequency. These data together suggest that decreases in s/mEPSC frequency in EAC3-I MSNs are potentially underestimated. Alternatively, some of these effects may be due to decreased dendritic length and complexity seen in EAC3I-positive MSNs. However, it is not clear whether distal MSN synapses are sampled in our s/mEPSC analyses due to cable filtering effects previously reported [101]. Hippocampal CaMKIIb has been shown to modulate dendritic length and branching as well as synapse number [70]. These results suggest that CaMKII plays important roles in modeling MSN dendritic morphology. The CaMKII alpha promoter fragment that drives EAC3I expression reportedly turns on around P5 [102,103,104,105]. This would lead to inhibition of CaMKII in early postnatal development and continuing into adulthood. The effects of this longer term genetic CaMKII inhibition contrasts with the typically minimal effects of acute, short term application of a related CaMKII inhibitor peptide, AIP, on basal glutamatergic transmission in the CA1 and CA3 region of the hippocampus [106,107,108,109]. Conversely, expression of CaMKII inhibitor peptides CaMKIIN or AIP over 2-6 days reduced hippocampal CA1 AMPAR-mediated, but not NMDAR-mediated EPSCs [110]. Another report suggests a CaMKII inhibitor peptide, CN19, persistently decreased hippocampal CA1 field EPSPs or EPSCs amplitudes at higher concentrations disrupting the CaMKII/NMDAR complex [111]. These data suggest that disruption of the CaMKII/NMDAR complex, a complex that increases following strong synaptic stimulation [112] and is necessary for LTP [113,114], may offer an alternative mechanism underlying our observed effects.
Endogenous CaMKIIa levels peak around 3 weeks postnatally, which is a crucial time for synaptogenesis and synapse maturation [104,105,115]. The apparent decrease in functional synaptic connections in CaMKII inhibited MSNs in adulthood may have four possible explanations: 1) CaMKII activity is necessary for the normal unsilencing of synapses in the adult; 2) Ongoing CaMKII activity is required to maintain functional synapses; or 3) CaMKII is needed early in synaptogenesis to turn initially silent synapses into functional ones; or 4) the CaMKII-mediated decrease in dendritic length may underlie the reduction in total synapse number. The decrease in sEPSC frequency in EAC3I MSNs was detected as early as three weeks postnatally, suggesting a large proportion of synapses were never unsilenced or perhaps never formed. In the hippocampus both electrophysiological and anatomical studies at light and electron microscopic levels suggest that in the first few weeks of life many synapses start as NMDARonly synapses or silent synapses [116,117,118,119,120]. Synapse unsilencing involving the trafficking of new AMPARs (GluA4containing) to the synapse in early postnatal development is dependent on activity, but is independent of CaMKII [121,122]. Instead, PKA plays an important role early in postnatal development (,P9) in plasticity in the hippocampus being necessary and sufficient for GluA4 incorporation, but requiring additional CaMKII activity for GluA1 receptor incorporation [122,123,124]. These data suggest that the similar synaptic phenotypes of GluA1KO and EAC3I mice arise from disruption of a common mechanism. However, it is interesting that the phenotype of the EAC3I cells is also virtually identical to that recently reported for MSNs in SAPAP3 knockout mice [7,125]. Intriguingly, SAPAP3 may be phosphorylated by CaMKII, possibly assisting in synaptic targeting of GluA1R [126]. Thus, it will be important for future studies to directly identify downstream MSN proteins regulated by CaMKII.
It is also possible that differences in the relative innervation of MSNs by cortical and thalamic inputs impacts the synaptic phenotypes. Indeed, differences in release probability have been observed between the two inputs [127,128,129]. Our measures of sEPSC and mEPSC frequency are likely comprised of both cortical and thalamic-mediated glutamate release, yet our data do not rule out the possibility that inhibition of MSN CaMKII may have a greater influence on one of these excitatory synapses over the other.
CaMKII inhibition also leads to alterations in intrinsic excitability that may serve to broadly counteract the reduced s/ mEPSC frequency. Although we expected a decrease in sEPSP frequency in current clamp based on the sEPSC results, we observed significantly larger sEPSP amplitudes. This is likely due to enhanced intrinsic excitability. In CaMKII-inhibited neurons we observed more depolarized resting membrane potentials, increased membrane resistance, decreased rheobase current injection, and increased firing frequency. Autonomously active CaMKII has been shown to suppress neuronal excitability by increasing cell-surface expression of an A-type K + channel, Kv4.2, via phosphorylation [30,130,131]. In addition, CaMKII inhibition in medial vestibular nucleus neurons increased intrinsic excitability via a reduction in BK-type calcium activated potassium currents [132]. Changes in Kv4.2 or BK activity following CaMKII inhibition could account for the differences in firing that we observed in EAC3I cells. However, modulation of Kir 2 channels may be responsible for the changes in input resistance and resting membrane potential [133]. A recent study showed that acute CaMKII inhibition in cortical cultures leads to increased excitability, but also increased cell death [134]. We did not note increased cell death, nor did membrane properties hint at unhealthy EAC3I-expressing cells. The differences in these studies may be attributed to the fact that MSNs are GABAergic cells bypassing potential excitotoxicity vulnerabilities seen with recurrent excitatory connections in the cortex. Alternatively, cortical cultures -which are often more excitable -may have a more difficult time regulating extracellular glutamate levels, something that is not as problematic in ex vivo slice preparations.
The opposing regulation of excitatory transmission and excitability observed in these studies suggests that CaMKII may serve as a molecular fulcrum to counterbalance changes in enhanced excitatory input with decreases in excitatory output. It is important to note in the present dataset we cannot rule out that possibility that one of these adaptations is compensatory to the other, rather than both being directly initiated by CaMKII inhibition. Regardless, this likely has important implications for the modulation of basal ganglia circuitry underlying habit learning, addiction and neurodegenerative disease. CaMKII plays a role in setting the number of functional synapses and therefore may provide a substrate for experience dependent plasticity in the striatum. Dorsal striatal CaMKII may be crucial early in postnatal development as well as in adulthood entraining new motor repertoires and refining synaptic connections as those motor skills are refined into habits later in life. With the inhibition of CaMKII leading to a decrease in the number of functional contacts, CaMKII may function in the dendritic and synaptic maturation processes, from nascent filopodia to mature dendritic spine [18,99], or alternatively be important in maintaining existing synaptic connections. Further investigation will be needed to determine the precise role of CaMKII in striatal synaptic maturation.