Integrated regulation of PKA by fast and slow neurotransmission in the nucleus accumbens controls plasticity and stress responses

Cortical glutamate and midbrain dopamine neurotransmission converge to mediate striatum-dependent behaviors, while maladaptations in striatal circuitry contribute to mental disorders. However, the crosstalk between glutamate and dopamine signaling has not been entirely elucidated. Here we uncover a molecular mechanism by which glutamatergic and dopaminergic signaling integrate to regulate cAMP-dependent protein kinase (PKA) via phosphorylation of the PKA regulatory subunit, RIIβ. Using a combination of biochemical, pharmacological, neurophysiological, and behavioral approaches, we find that glutamate-dependent reduction in cyclin-dependent kinase 5 (Cdk5)-dependent RIIβ phosphorylation alters the PKA holoenzyme autoinhibitory state to increase PKA signaling in response to dopamine. Furthermore, we show that disruption of RIIβ phosphorylation by Cdk5 enhances cortico-ventral striatal synaptic plasticity. In addition, we demonstrate that acute and chronic stress in rats inversely modulate RIIβ phosphorylation and ventral striatal infusion of a small interfering peptide that selectively targets RIIβ regulation by Cdk5 improves behavioral response to stress. We propose this new signaling mechanism integrating ventral striatal glutamate and dopamine neurotransmission is important to brain function, may contribute to neuropsychiatric conditions, and serves as a possible target for the development of novel therapeutics for stress-related disorders.


Edited by Roger Colbran
Cortical glutamate and midbrain dopamine neurotransmission converge to mediate striatum-dependent behaviors, while maladaptations in striatal circuitry contribute to mental disorders. However, the crosstalk between glutamate and dopamine signaling has not been entirely elucidated. Here we uncover a molecular mechanism by which glutamatergic and dopaminergic signaling integrate to regulate cAMP-dependent protein kinase (PKA) via phosphorylation of the PKA regulatory subunit, RIIβ. Using a combination of biochemical, pharmacological, neurophysiological, and behavioral approaches, we find that glutamate-dependent reduction in cyclindependent kinase 5 (Cdk5)-dependent RIIβ phosphorylation alters the PKA holoenzyme autoinhibitory state to increase PKA signaling in response to dopamine. Furthermore, we show that disruption of RIIβ phosphorylation by Cdk5 enhances cortico-ventral striatal synaptic plasticity. In addition, we demonstrate that acute and chronic stress in rats inversely modulate RIIβ phosphorylation and ventral striatal infusion of a small interfering peptide that selectively targets RIIβ regulation by Cdk5 improves behavioral response to stress. We propose this new signaling mechanism integrating ventral striatal glutamate and dopamine neurotransmission is important to brain function, may contribute to neuropsychiatric conditions, and serves as a possible target for the development of novel therapeutics for stress-related disorders.
A fundamental function of brain circuitry is to impart emotional salience upon sensory input so that environmental experiences may be acted upon with appropriate behavioral responses. This requires the integration of fast ligand-gated ionotropic synaptic signals with those mediated by G-protein-coupled neurotransmitter receptors. Subserving these synaptic receptors, Ca 2+ and cAMP are two principle second messengers that invoke overlapping intracellular signaling networks which compute the appropriate responses in the form of altered excitability and synaptic strength upon which learned behavioral responses are based. Understanding of the molecular mechanisms by which this is achieved is incomplete.
The nucleus accumbens (NAc, ventral striatum) is a central input structure within the basal ganglia and mediates reinforced learning and behavioral response processing. The medium spiny neurons (MSNs) in the NAc receive glutamatergic input from the medial prefrontal cortex, hippocampus, and other regions as well as dopaminergic input from the ventral tegmental area (1,2). Coincidental detection of these two inputs is critical to reward, aversion, motor learning, and other aspects of sensory and emotional integration and behavioral flexibility (3,4). The contributions of glutamate and dopamine to these critical brain functions are mediated by the interplay of key signaling pathways, such as the cAMP-dependent protein kinase (PKA) cascade (5)(6)(7).
The tetrameric PKA holoenzyme consists of two catalytic subunits (PKA cat : Cα, Cβ, or Cγ isoforms) bound to a homodimer of regulatory subunits (RIα, RIβ, RIIα, or RIIβ isoforms). When RI subunits are incorporated, it is classified as Type I PKA, when bound to RII subunits as Type II PKA. R subunits possess an N-terminal dimerization domain (D/D), responsible for both dimerization and subcellular localization via binding to members of the A-kinase anchoring protein (AKAP) family of scaffolding molecules (8). Each R subunit monomer contains two C-terminal cyclic nucleotide-binding domains. Upon cAMP binding, PKA is activated by unleashing the catalytic activity from R dimer inhibition (9). Between the D/D and cyclic nucleotide-binding domains is a flexible linker region containing the inhibitor sequence (IS) responsible for occupying the C subunit active site in the inhibited state under basal conditions. The linker imparts an important structural distinction between Type I and Type II PKA, with functional implications for the holoenzyme. While the IS in Type I R subunits is a pseudosubstrate for PKA, the Type II IS acts as a true substrate for the C subunits at Ser95 on RIIα and Ser114 on RIIβ (10). Phosphorylation at the PKA site on RII subunits has been suggested to slow the reassociation rate of the R and C subunits (11,12).
PKA activation engages many downstream effectors and signaling cascades, and many pathways impinge upon cAMP/ PKA signaling. The cAMP/PKA cascade may be activated by either ligand-gated ionotropic or G s protein receptor-coupled metabotropic neurotransmission (6,13). Previously, we showed that constitutive phosphorylation of DARPP-32 (14,15) by cyclin-dependent kinase 5 (Cdk5) serves as an important mechanism by which glutamate and dopamine interact to regulate PKA signaling. Also, we reported that PKA-dependent activation of cAMP-specific phosphodiesterases is modulated though Cdk5-dependent phosphorylation of PDE4 family members (16). However, direct regulation of the PKA holoenzyme has not previously been shown to fall under the coordinated control of excitatory and metabotropic neurotransmission. We hypothesized that control of PKA activation/inhibition could serve as a critical point of integration for glutamate and dopamine inputs. Here we report a novel mechanism by which these two modes of neurotransmission are integrated to mediate ventral striatal circuitry function in plasticity and stress-response behavior.

Results
The PKA regulatory subunit RIIβ is a substrate for Cdk5 Messenger RNA transcripts encoding all four isoforms of the PKA regulatory subunit are present in brain, with expression of the RIIβ isoform enriched in the central nervous system (17)(18)(19). Examination of protein expression (Fig. 1A) indicates that the RIIβ subunit is most selectively expressed in brain compared to the other regulatory isoforms. Moreover, its brain region tissue distribution mimics the expression pattern of Cdk5. For both RIIβ and Cdk5, highest levels occur in limbic regions including prefrontal cortex, dorsal and ventral striatum, and hippocampus.
Immunostains of ventral striatum showed RIIβ in the soma and neuropil of MSNs where it colocalized with its scaffolding protein, AKAP150 (20, 21) (Fig. 1B). RIIβ colocalized with Cdk5 in dissociated rat striatal neurons grown in culture (Fig. 1C) and within cells and neuropil of rat striatal tissue. Furthermore, analysis of a rat NAc single nucleus RNA (snRNA-seq) library (22) revealed that RIIβ (i.e., Prkar2b) and Cdk5 genes are co-expressed within subclusters of rat NAc cell populations including D1 and D2 dopamine expressing MSNs (D1-MSN, D2-MSN) and GABAergic and glutamatergic neurons (Fig. S1, A-E). Interestingly, RIIβ expression was absent from populations of astrocytic and oligodendritic cells (Fig. S1, B and D). Similarly, Cdk5 expression occurred in populations of D1-MSN, D2-MSN, GABAergic, and glutamatergic neurons and was also absent in astrocytic and oligodendritic cell clusters (Fig. S1, C and E). Thus, Cdk5 and RIIβ are coexpressed within subcellular populations of the rat ventral striatum.
The amino acid sequence of RIIβ includes a proline-directed Cdk5 consensus motif (S/TPXK/H/R) within the N-terminal linker region predicting Thr69, a site unique to this particular R subunit, as a potential Cdk5 phosphorylation site. Indeed, Cdk5 phosphorylated pure recombinant RIIβ in vitro, achieving a maximal stoichiometry of 0.6 mol PO 4 /mol of substrate under saturating conditions (Fig. 1, E and F). Mass spectrometry positively identified Thr69 as the site of phosphorylation, and the T69A site-directed mutant form of RIIβ was not phosphorylated by Cdk5 at all. To determine if this phosphorylation occurred in brain, a phosphorylation statespecific antibody for phospho-Thr69 RIIβ was derived (Fig. S1, F-H). Phospho-Thr69 RIIβ was detected within MSNs throughout the striatum by immunostaining (Fig. 1G). Together these data demonstrate that the RIIβ regulatory subunit of PKA is phosphorylated at Thr69 by Cdk5 and that this phosphorylation event occurs in vivo within striatal neurons.

Phosphorylation of RIIβ by Cdk5 affects PKA holoenzyme autophosphorylation kinetics
Both genetic deletion and pharmacologic inhibition of Cdk5 increases phosphorylation of PKA substrates in striatum (16). Thus, we hypothesized that Cdk5-mediated phosphorylation of Thr69 RIIβ could have an inhibitory effect on PKA cat activity. However, in vitro analysis indicated that Thr69 phosphorylation did not alter the inhibition of PKA catalytic activity by RIIβ (Fig. S1I). Furthermore, phospho-memetic T69D mutation had no effect on cAMP-dependent activation of PKA in the RIIβ/PKA holoenzyme complex (Fig. S1J). As an alternative, we considered that Cdk5-dependent phosphorylation of Thr69 RIIβ could alter structure/function aspects of the holoenzyme inhibitory state. As mentioned above, the RIIβ D/D forms a four-helix bundle that docks with AKAPs through hydrophobic interactions (23, 24) ( Fig. 2A). The Cdk5 phosphorylation site on RIIβ is encompassed within the flexible linker region that occurs between the D/D and the cAMP binding domains. This site is proximal to the IS in Type II R subunits, which contains the PKA phosphorylation site Ser114 (Fig. 2B). Therefore, we assessed the effect of phospho-memetic T69D mutation on PKA-dependent phosphorylation of Ser114 RIIβ (Fig. 2C). Interestingly, T69D phospho-mimetic mutation attenuated the efficiency of Ser114 phosphorylation by PKA in comparison to WT RIIβ, significantly reducing the maximum velocity of the reaction under linear conditions. Thus, the phosphorylation state of RIIβ at Thr69 may govern PKA "autophosphorylation" at Ser114.
To determine if this functional relationship between Thr69 and Ser114 RIIβ phosphorylation occurs ex vivo, acute striatal brain slices were treated with the Cdk5 inhibitor, CP681301 (Fig. 2D). Cdk5 inhibition attenuated the phosphorylation of phospho-Thr69, with corresponding increase in PKAdependent phosphorylation of Ser114. Together, these data reveal a novel intramolecular mechanism by which RIIβ phosphorylation is reciprocally regulated at two proximal sites in vitro and ex vivo.

Mechanistic function and regulation of RIIβ/PKA in ventral striatum
PKA regulation via RII subunit autophosphorylation has not previously been examined as a neuronal signaling mechanism. This may be because phosphorylation of RII by PKA characterizes the inhibited state of type II PKA holoenzymes (10). However, early studies suggested that phosphorylation at the PKA site could slow RIIβ reassociation with PKA cat following holoenzyme dissociation (11).
Consistent with this mechanistic function, phospho-mimetic S114D mutation of RIIβ reduced the PKA inhibition potency over 4-fold (IC 50 = 1.5 ± 0.3 nM, WT; 6.1 ± 1.1, S114D) (Fig. 3A). Thus, the level of PKA-dependent phosphorylation of Ser114, which occurs with formation of the PKA/RIIβ holoenzyme inhibitory state, may be controlled by the phosphorylation state of Thr69 RIIβ and may determine PKA inhibition potency by RIIβ. Therefore, neuronal activity which regulates Thr69 levels could modulate the efficacy of activators of G protein-coupled receptors that invoke the cAMP/ PKA signaling cascade.
To further assess this possibility, the regulation of phospho-Thr69 and Ser114 RIIβ was explored in striatal slices. Acute N-methyl-D-aspartate (NMDA) treatment (50 μM, 5 min) strongly reduced the phosphorylation state of Thr69 (Fig. 3B). Concomitantly, PKA-dependent phosphorylation at Ser114 was increased, consistent with the reciprocal modulation of these sites detected in vitro and in response to Cdk5 inhibition in intact brain tissue. The effect of NMDA on phospho-Thr69 appeared to be mediated by the activation of the Ca 2+dependent serine/threonine protein phosphatase, PP2B (calcineurin), as the NMDA-induced reduction was blocked by the PP2B inhibitor cyclosporin A (Fig. S2A). Inhibition of protein phosphatases PP1 and PP2A by 1 μM okadaic acid also attenuated the effect of NMDA. Moreover, the basal phosphorylation state of phospho-Thr69 was increased by PP2B inhibition with cyclosporin A, PP1 inhibition by 1 μM okadaic acid, or selective PP2A inhibition with 200 nM okadaic acid (Fig. S2B). Together these data indicate that phospho-Thr69, and thereby phospho-Ser114, are regulated by activation of ionotropic NMDA class glutamate receptors. This is mediated, at least in part, through Ca 2+ -dependent protein phosphatases including PP2B and 2A. These phosphatases together with PP1 also determine the basal phosphorylation state of phospho-Thr69 RIIβ. The diversity in protein phosphatases dephosphorylating RIIβ may not be surprising as both RII subunits interact with the serine/threonine phosphatase calcineurin (PP2B), Cdk5 complexes with PP2A (25), and AKAP signalosomes cluster protein phosphatases (26). Phosphorylation at the Thr69 site modulates phosphorylation at Ser114, which in turn affects PKA inhibition by dissociated RIIβ. Therefore, NMDA-dependent modulation of Thr69 has the potential to impart effects upon PKA activation.
To test this, ventral striatal slices were treated with NMDA to activate fast neurotransmission, dopamine to activate slow neurotransmission, or both. As both NMDA and dopamine receptor activation are essential to the induction of striatal long-term potentiation (LTP) (27), these treatments were conducted under conditions similar to those used to induce striatal plasticity (Fig. 3C). Specifically, dopamine (10 μM) was bath-applied for 15 min with NMDA (25 μM) added for the last 5 min in reduced Mg 2+ buffer with a 30 min delay following treatments. Effects on phospho-Thr69 RIIβ compared with PKA activity, as indicated by changes in PKAdependent phosphorylation states of two key effectors, Ser845 GluA1 and Thr34 DARPP-32 (Fig. 3D). Under these conditions, NMDA again induced a decrease in phospho-Thr69 RIIβ, without significantly altering PKA activity. Dopamine alone also did not sustain PKA activity for 30 min. However, phospho-Thr69 reduction by combined dopamine and NMDA treatment corresponded to increased PKA activity 30 min after the treatments were removed (Fig. 3D). These data suggest a compound mechanism in which relatively high levels of phospho-Thr69 occur under basal conditions. Robust excitatory glutamatergic neurotransmission reduces phospho-Thr69 via protein phosphatase activation, resulting in increased PKA-dependent phosphorylation of Ser114 RIIβ. In the context of dopamine receptor activation, this leads to sustained elevated PKA activity involving downstream effector signaling pathways known to mediate synaptic plasticity and striatal function. Data are normalized means ± S.D., ****p < 0.0001 for P-T69, **p < 0.01 for P-S114 RIIβ, unpaired t test, n = 8 to 9. AKAP, A-kinase anchoring protein; Cdk5, cyclin-dependent kinase 5; PKA, cAMP-dependent protein kinase.

RIIβ phosphorylation controls ventral striatal plasticity
To directly assess the contribution of the RIIβ regulatory phosphorylation mechanism to striatal function, a small interfering peptide (siP) was developed which corresponded to 18 amino acid residues from the RIIβ linker region encompassing Thr69 and included a membrane-permeabilizing penetratin tag (Fig. 4A, top). This Thr69 RIIβ siP (RIIβ siP) potently inhibited Cdk5-dependent phosphorylation of RIIβ (IC 50 = 7.14 ± 2.28 μM) (Fig. S3A). Treatment of ventral striatal slices with this peptide caused reduction in phospho-Thr69 RIIβ but had no effect on Cdk5-dependent phosphorylation of Thr75 DARPP-32 (Fig. 4A, bottom). RIIβ siP/ dopamine cotreatment of ventral striatal slices resulted in a sustained increase in PKA activity, similar to the previously observed effects of NMDA/dopamine treatment (Fig. S3B). Consequently, we established the means to selectively modulate phospho-Thr69 RIIβ levels in intact brain tissue and facilitate prolonged PKA activation.
With this reagent in hand, we chose to assess the role of RIIβ phosphorylation in cortico-ventral striatal plasticity as one measure of physiological function. Cortical afferents within corpus callosum of sagittal brain slices were stimulated, and field excitatory postsynaptic potentials (fEPSPs) were recorded in ventral striatum core (Fig. S4A). LTP was induced by high-frequency stimulation (HFS) and could be enhanced by 15 min bath application of the D1-type dopamine receptor agonist SKF81297 (Fig. S4B), as previously observed for dorsolateral striatum (27). To assess the contribution of RIIβ Recovery period 30 45 Time ( phosphorylation to LTP, equilibrated slices were treated with either the RIIβ siP or a scrambled siP (control, 10 μM) for 15 min, followed by drug washout for 25 min prior to HFS (Fig. 4, B and C). The RIIβ siP had no effect on synaptic excitation (input-output, I/O) or paired pulse response ratios. RIIβ siP caused a transient reduction in basal fEPSP amplitude but significantly enhanced LTP compared to scrambled control siP, causing elevations in fEPSP amplitudes that were maintained through 50 min following HFS (0.31 ± 0.03 mV scrambled siP versus 0.45 ± 0.06 mV for RIIβ siP). Interestingly, the enhancement of LTP by the RIIβ siP was partly occluded in slices pretreated with SKF-81297 (2 μM) prior to HFS (Fig. S4, C and D).
To better understand this effect, whole cell patch-clamp recordings were made of D1R-expressing neurons (28). As observed in field recordings, the RIIβ siP induced a significant increase in EPSP amplitude and markedly elevated LTP levels in comparison to control treated neurons (Fig. 4D). This effect was attributable to increased α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptor (AMPAR)-mediated function, as AMPA/NMDA ratios were increased by the RIIβ siP (Fig. 4E), consistent with elevated phospho-Ser845 GluR1, which is linked to increased synaptic surface AMPAR trafficking (29) Thus, direct siP targeting of Thr69 RIIβ phosphorylation enhances cortico-ventral striatal plasticity, likely by increasing PKA activity. These findings together with the slice signaling data (see Fig. 3) are consistent with the modulation of RIIβ phosphorylation and PKA activity by NMDA and dopamine as an important mechanism mediating ventral striatal plasticity.

RIIβ regulation is altered by stress and can be targeted to affect stress-related behaviors
Synaptic plasticity and cognition are affected by stress (30)(31)(32). While acute stress can improve cognitive performance, exposure to persistent stress causes cognitive impairment (33,34). To determine if stress could alter the signaling state of RIIβ phosphorylation, rats were exposed to a stressful environment either once (acute stress, one forced swim) or chronically (chronic unpredictable stress, two stressors daily for 14 days). Interestingly, 1 h following exposure to acute stress, phospho-Thr69 RIIβ was significantly decreased (Fig. 5A). In contrast, chronic exposure to stress caused elevation in the phosphorylation state of this site. Thus, environmental exposure to stressful conditions can alter the basal signaling state of RIIβ, raising the intriguing possibility that the regulation of RIIβ phosphorylation may contribute to behavioral responses induced by stress. Additionally, this mechanism may be targeted to modify behavioral responses to stress.
To examine this possibility, bilateral intra-accumbens infusions of the RIIβ siP via cannula were conducted. Delivery of FITC-tagged RIIβ siP to ventral striatal core MSNs was confirmed by immunostains (Fig. 5B). RIIβ siP infusion induced increased PKA-dependent phosphorylation of Ser114 RIIβ, Ser845 GluA1, and Thr34 DARPP-32 in vivo (Fig. 5C), similar to its effects in slices. Incorporating this approach with behavioral studies, animals were subjected to the 2-day forced swim test (FST. Fig. 5D). On day 1, animals that had undergone intra-accumbal cannulation were made to swim for 8 min. The following day, subjects were infused with either the RIIβ siP or a scrambled control peptide, then re-exposed to the stress of forced swimming. RIIβ siP-infused subjects spent significantly less time immobile (Fig. 5E, left). The RIIβ siP effect was completely blocked by co-administration of the PKA inhibitor, PKI. No changes in immobility time were detected when PKI alone was infused. Moreover, none of these infusions had a significant effect on latency to immobility in the FST or locomotor activity (Fig. S5, A and B). Immobility was significantly less for animals that received the RIIβ siP during the 4 to 6 and 6 to 8 min periods of forced swimming compared to controls receiving scramble peptide (Fig. S5C). Also, the RIIβ siP's effects closely mimicked those of the rapidacting antidepressant, ketamine, administered at subanesthetic levels via bilateral intra-accumbal infusion. These findings demonstrate that RIIβ phosphorylation in the NAc mediates behavioral reactions to a stressful environment and that these responses may be altered by pharmacological targeting of this integrative mechanism.
Together these findings support a signaling scheme (Fig. 5F) where the PKA/RIIβ holoenzyme can be shifted to a form where Thr69 RIIβ is less phosphorylated in response to excitatory glutamatergic neurotransmission and more phosphorylated at Ser114. In this state, upon activation of Gscoupled D1 dopamine receptor signaling, PKA activation may be prolonged, thereby contributing to synaptic plasticity and behavioral responses. Thus, the equilibrium between different regulatory states of PKA may be an important determinant of behavioral responses to stress and overall mood.

Discussion
Emotional salience is imparted upon environmental experiences through integration of cortical glutamatergic and midbrain dopaminergic neurotransmission. The striatum integrates these inputs to mediate reward, motivation, and behavioral flexibility, allowing for learned behavioral responses to rewarding or aversive stimuli (35). The predominant second messengers that mediate dopaminergic and excitatory glutamatergic activity are cAMP and Ca 2+ , respectively. These signaling modes can act antagonistically by activating protein phosphorylation or dephosphorylation of the same substrate post-LTP induction. Summary data (right) show effect of RIIβ siP (10 μM, 1 h) versus control (scrambled siP). Data represent means ± S.D.; **p < 0.01 compared to control baseline, *p < 0.05, compared to RIIβ siP baseline, #p < 0.05, compared to LTP in controls, Student's t test, n = 4 to 6. E, effect of RIIβ siP on EPSP amplitude and response to selective ionotropic glutamate receptor antagonists in D1R neurons. Control and RIIβ siP-treated NAc slices underwent NMDAR antagonist APV (20 μM) treatments. Subsequent treatments with the AMPAR antagonist NBQX (10 μM) ablated EPSP responses (see inset). EPSP recordings (30 min after HFS, LTP induction) and APV effects are shown. Insets show traces for D1R neurons before (1) and after APV (2), and combined APV/NBQX (3) treatments. Summary data are presented (right) for AMPA/NMDA ratios. Data are means ± S.D., *p < 0.05, Student's t test n = 4 to 6. fEPSP, field excitatory postsynaptic potential; HFS, high-frequency stimulation; LTP, long-term potentiation; NAc, nucleus accumbens.
site. However, they may function most critically when working synergistically to invoke downstream effectors that alter neuronal excitability (36). For example, integration may occur upstream of PKA, at the level of cAMP metabolism, as both Ca 2+ /CaM-dependent adenylyl cyclase (37,38) and phosphodiesterases (39) can control cAMP/PKA signaling. cAMP/PKA and Ca 2+ cascades can also converge at key effectors downstream of PKA such as DARPP-32 (40) and CREB (41). Here, we reveal convergence directly upon the RIIβ/PKA cat holoenzyme complex as a fundamental mechanism of glutamate and dopamine integration.
The novel mechanism by which phosphorylation of the RIIβ regulatory subunit governs PKA holoenzyme activation involves the ability of Cdk5-dependent phosphorylation of Thr69 to reduce PKA autophosphorylation of Ser114 RIIβ. Phosphorylation at Ser114 has the net effect of increasing PKA activity by reducing the ability of RIIβ to inhibit PKA cat once activated. Both Thr69 and Ser114 are located within the nonstructural linker region of RIIβ that occurs between the AKAP binding/D/D and cAMP-binding portion of RIIβ. Despite the high level of flexibility in this region, the linker has been demonstrated to play an important role in the overall conformation of the PKA holoenzyme. Holoenzymes containing RIIβ are the most structurally compact compared to those of other regulatory isoforms despite possessing the longest linker sequence of any R subunit (42). This feature is conferred by the unique sequence of the RIIβ linker region itself, as substitution of the RIIβ linker into RIIα renders holoenzyme structural density similar to that of PKA cat /RIIβ (43). Phosphorylation of the proline-directed Thr69 site may also impart constraints or conformational changes in RIIβ secondary structure that prevent S114 from optimally occupying the catalytic pocket of PKA. Ongoing studies are aimed at solving the structural basis for these effects and better understanding how Ser114 phosphorylation attenuates PKA inhibition. Detailed understanding of the interactions between RIIβ, PKA cat , and other protein kinases and phosphatases in the context of AKAP signalosomes (44) which contribute to synaptic plasticity (45) also remains an important area of investigation.
D1-dopamine receptor and glutamatergic NMDAR coactivation enhances striatal plasticity (5) and is required to achieve cortico-striatal LTP in dorsolateral striatum (27). PKA phosphorylation of DARPP-32 (46,47) and GluA1 (48)  phosphorylation either via NMDAR activation or with an RIIβ siP resulted in PKA-dependent phosphorylation of GluA1, DARPP-32, and likely other downstream effectors. Importantly, selective reduction of phospho-Thr69 RIIβ markedly improved cortico-striatal LTP in ventral striatum, consistent with the known functional contributions of these pathways. Ventral striatal plasticity is thought to be crucial for coordination of behavioral responses to stressful environments through reinforcement learning and the encoding of rewarding versus aversive stimuli (49). Proper integration of fast and slow neurotransmission may be required for accurate assignment of emotional salience to stimuli and appropriate goal-directed behavioral responses, such as avoidance (50). Here, acute stress caused dephosphorylation of Thr69 RIIβ, and intraaccumbens reduction of Thr69 RIIβ increased struggle in response to re-exposure to forced swim stress. Thus, the regulation of RIIβ phosphorylation and its consequent effects on PKA activity may be an important mechanism mediating stress response behavior.
Chronic exposure to stress caused an increase in the phosphorylation state of Thr69 RIIβ. Acute stress can improve behavioral and mental acuity (51). However chronic and unpredicted stress can result in a range of deleterious effects including cognitive impairment (33) and is a major predisposing factor for mental illnesses such as major depressive disorder (52). Elevated levels of stress hormones have been linked to suppression of PKA signaling (53). Furthermore, dysregulation of the PKA pathway has been implicated in psychiatric disorders including depression (54). Thus, cAMP/ PKA signaling presents an important target for drug therapy. Indeed, antidepressant medications often act either directly or indirectly upon G protein-coupled catecholamine receptors that modulate cAMP/PKA signaling. However, therapies that act at the receptor level often cause unwanted side effects. Thus, more direct manipulation of cAMP/PKA signaling is an attractive alternative approach to antidepressant treatment. Here we show that targeting of PKA signaling via an RIIβ siP to alter plasticity and behavioral function. It will be interesting to further explore the mediators of this effect including possible contributions from cFos and δFosB as well as less well-defined mechanisms. If RIIβ phosphorylation can be targeted by small compounds to achieve similar effects as those demonstrated here using the RIIβ siP, perhaps improved health care for mood disorders may be achieved.

Experimental model and subject details
Ten-to 14 week-old male Sprague-Dawley rats (Charles River, strain code 400) were used for most animal experiments and maintained on a 12 h light/dark cycle, single-or doublehoused (chronic stress paradigm only) in standard cages, with food and water available ad libitum unless otherwise specified (e.g., food restriction in chronic stress paradigm). For single unit patch clamp recordings, Drd2-EGFP mice in an FVB background were used. All manipulations were approved by UT Southwestern Medical Center and University of Alabama at Birmingham Institutional Animal Care and Use Committees and conducted in accordance with the NIH Guide for the Care and Use of Laboratory Animals.

Antibodies and reagents
All chemicals were purchased from Sigma-Aldrich unless otherwise specified. The sequence of the penetratin-tagged RII β-siP and penetratin-tagged scramble control peptides were RQIKIWFQNRRMKWKK-AGAAAGGGTPSKGVNF AEE and RQIKIWFQNRRMKWKK-AAGAGSGAVAFKGA NGEA or RQIKIWFQNRRMKWKK-AGAAAGGGAPR KGVNFAEE, respectively. The FITC-tagged siP was synthesized with an N-terminal FITC label.

Protein purification, in vitro phosphorylation, and phosphorylation site identification
Rat recombinant RIIβ was derived as a C-terminal 6XHis fusion by Ni-NTA affinity purification using standard pET vector methods (Novagen). RIIβ was phosphorylated in vitro by Cdk5 or PKA using standard conditions (55). Cdk5/p25 was from Millipore, and PKA was from NEB. 32P incorporation was quantitated by PhosphorImager analysis. All kinetic analyses were conducted under empirically defined linear conditions. PKA activity was measured by phosphorylation of recombinant inhibitor-1 (56) or DARPP-32 (14). For phosphorylation site identification, preparatively phosphorylated RIIβ was subjected to nano-liquid chromatography tandem mass spectrometry as previously described (57). Digests were redissolved in 0.1% trifluoracetic acid and loaded onto a Zip-Tip with C 18 resin (Millipore) for purification. Analysis was conducted using nanoelectrospray-QSTAR Pulsar I quadrupole time-of-flight tandem mass spectrometry (MDS Proteomics). Structural model of AKAP/RIIβ dimer interaction was based on published data of AKAP (PDB: 2HWN) (24) and RIIβ (PDB: 1CX4) (58).
For staining of primary cultured neurons, embryonic striatal neurons (E18) from Long Evans rats (Charles River Labs) were cultured 14 to 21 days in vitro on 12-mm coverslips, fixed, immunostained for Cdk5 and RIIβ using the above indicated antibodies and imaged as previously described (61).

Rat NAc single cell library analysis
Using an open-source software of single-nucleus RNA sequencing data from the rat ventral striatum (Ratlus) (22) Cdk5 and RIIβ were searched and segmented for adult-male Sprague Dawley control condition rats. Uniform Manifold Approximation Projections were generated across all cell types previously assessed (22). Feature plots displaying genes of interest across all cell types were generated, and violin plots indicating gene expression across six cell classes were exported for comparative analysis.

Electrophysiology
Adult rats were anaesthetized by isoflurane (Piramal Healthcare), and brains were placed in ice-cold sucrose saline solution ( (22-25 C) and allowed 1 h recovery. Slices were transferred to perfusion chambers placed within an upright microscope stage and visualized by infrared differential interference microscopy and a CCD Super Low Luminance camera (KT&C, Co., Ltd). Slices were perfused continuously with oxygenated saline solution (2-3 ml/min).
Extracellular recordings: fEPSPs were obtained at 25 C in the presence of the GABA A antagonist (SR95531, Gabazine, 2 μM) and were evoked by square current pulses (0.2 ms) at 0.033 Hz with a bipolar stimulation electrode (FHC) placed at the border of corpus callosum separated by 300 to 500 μ from the recording electrode. The perfusion saline solution was partially modified (in mM): 0.5 MgCl 2 , 1 CaCl 2 , 3.5 KCl. Results were obtained using a stimulus intensity to induce 60% of the maximal fEPSP amplitude taken from the I/O curve of each slice. The paired pulse ratio (PPR) paradigm was applied using different inter-stimulus intervals and PPR changes were measured as PPR = second fEPSP amplitude/first fEPSP amplitude. After recording stable baseline for at least 10 min, LTP was induced by HFS (4 trains, 100 Hz, 1 s duration, separated by 20 s). fEPSP amplitude was monitored for at least 50 min after HFS to evaluate LTP. All recordings were performed using a Multiclamp 700B amplifier and filtered at 4 kHz and digitized with a Digidata 1440 with pClamp 10 software for data acquisition (Axon, Molecular Devices, LLC). Recording pipette was filled with the same extracellular solution as the perfusion bath (2-4 MΩ resistance).

Stereotaxic surgeries
Rats were anesthetized with isoflurane and placed in a stereotaxic frame (Kopf Instruments). Bilateral cannulas (3 mm center-to-center distance, cut 8 mm below pedestal, Plastics One) were placed in NAc core (anterior-posterior to bregma +1.5 mm, medial/lateral ±1.5 mm, depth −6.8 mm). Peptides were bilaterally infused 1 week later in awake and mobile animals. Injection volume was 1 μl of peptide (100 μM in PBS, 10% DMSO, per cerebral hemisphere). Animals were allowed to recover for 1 h prior to behavioral testing. Accuracy was confirmed by infusion of methylene blue dye or direct visualization of cannula tracts postbehavioral testing.

Behavioral analysis
The Porsolt FST was conducted as described previously (64). Briefly, one-trial forced swim was used for acute stress. Nonsurgerized rats underwent an 8 min swim in an 8-gallon transparent water filled cylinder at 24 to 25 C. Animals recovered in their home cages for 1 h prior to brain dissection. The NAc was then microdissected for immunoblotting. For two-trial FST, rats recovered from surgical cannulation procedures for 7 days prior to behavioral testing. On day 1, animals were placed in the testing room for 1 h to acclimate. They then underwent an 8 min preswim. Video recording was performed from the swimming tank side to observe front and hind limb motion. Awake and mobile animals were bilaterally infused 24 h later with scrambled peptide or siP and/or PKI (1 μl,100 μM), or ketamine (2 μg in 1 μl) at a rate of 0.1 μl/min. They were then placed in a locomotor chamber, and locomotor activity was monitored for 30 min in circular donut-shaped locomotor chambers in the dark with a computer-monitored infrared photobeam system (MED-PC, Med Associates). Locomotor counts were defined as sequential adjacent beam breaks. Locomotor testing was followed by a 30 min recovery in home cages and then 8 min test swims. The last 6 min of the test were scored for latency to immobility and time immobile, with immobility defined as the time animals spent completely motionless.
The chronic unpredictable stress paradigm was performed as adapted from Willner (65). Animals were pair-housed for 14 days prior to testing. Four pairs of control animals received only regular cage cleaning and handling, while the four pairs of test animals were exposed to two stressors per day, chosen pseudorandomly, for 14 days. Stressors included: Cage crowding (4/cage, 2 h), overnight isolation (1/cage, 12 h), overnight food deprivation (12 h), overnight water deprivation (12 h), cage placed at 4 C (45 min), cage placed on orbital shaker (1 h), forced swimming (4 min), reverse lighting for dark cycle (12 h), lights off mid-cycle (3 h), tilted cage (45 , 12 h), and aversive odor exposure (cat urine, 12 h). Animals did not experience any stressor more than four times during the testing period, and at least 3 days were allowed between any one specific stressor. Brains were harvested 12 h after final stressor.

Statistical analysis
Data are reported as mean or normalized mean ± standard deviation, unless specified otherwise. Statistical analysis comparing two means was performed using the Student's t test. In addition, for some comparisons the Wilcoxon signedrank test was used where indicated. In some cases, the means for more than one group were compared by ANOVA with the appropriate post hoc comparison as specified in the figure legends using GraphPad Prism 6.0 unless stated otherwise. Where specified, two categorical variables were compared using a two-way ANOVA. For all experiments, *p <0 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 were considered significant. No statistical methods were used to predetermine sample sizes, but sample sizes are similar to those generally employed in comparable studies. Statistical analysis was conducted with n ≥ 4 except for in vitro biochemistry (n = 3).

Data availability
Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Dr James A. Bibb (jbibb@uab.edu).