Synergistic Regulation of Glutamatergic Transmission by Serotonin and Norepinephrine Reuptake Inhibitors in Prefrontal Cortical Neurons *

Background: Serotonin and norepinephrine reuptake inhibitors (SNRIs) produce better therapeutic effects than single selective reuptake inhibitors, but the underlying mechanisms are largely unknown. Results: Low dose SNRIs, by acting on 5-HT 1A and (cid:2) 2 -AR, synergistically reduced glutamatergic transmission in prefrontal cortex (PFC). Conclusion: SNRIs exerts a powerful impact on PFC synaptic activity. Significance: Our study provides a mechanism for the therapeutic effects of SNRIs in mental disorders related to PFC dysfunction.

The monoamine system in the prefrontal cortex has been implicated in various mental disorders and has been the major target of anxiolytics and antidepressants. Clinical studies show that serotonin and norepinephrine reuptake inhibitors (SNRIs) produce better therapeutic effects than single selective reuptake inhibitors, but the underlying mechanisms are largely unknown. Here, we found that low dose SNRIs, by acting on 5-HT 1A and ␣ 2 -adrenergic receptors, synergistically reduced AMPA receptor (AMPAR)-mediated excitatory postsynaptic currents and AMPAR surface expression in prefrontal cortex pyramidal neurons via a mechanism involving Rab5/dynamin-mediated endocytosis of AMPARs. The synergistic effect of SNRIs on AMPARs was blocked by inhibition of activator of G protein signaling 3, a G protein modulator that prevents reassociation of G i protein ␣ subunit and prolongs the ␤␥-mediated signaling pathway. Moreover, the depression of AMPAR-mediated excitatory postsynaptic currents by SNRIs required p38 kinase activity, which was increased by 5-HT 1A and ␣ 2 -adrenergic receptor co-activation in an activator of G protein signaling 3-dependent manner. These results have revealed a potential mechanism for the synergy between the serotonin and norepinephrine systems in the regulation of glutamatergic transmission in cortical neurons.
Monoamines, such as serotonin (5-HT) 2 and norepinephrine (NE), play important roles in cognitive and emotional pro-cesses mediated by the prefrontal cortex (PFC) (1,2). Changing 5-HT and NE transmission has been the core mechanism among antidepressants and anxiolytics (3)(4)(5)(6)(7). Because monoamines do not pass the blood-brain barrier, effective drugs act as reuptake inhibitors to raise their synaptic concentrations. Clinical trials suggest that drugs increasing both 5-HT and NE levels at synapses give more robust therapeutic effects than those increasing a single neurotransmitter alone, including faster onset of therapeutic effects and lower relapse rate (4,8,9). However, little is known about the underlying mechanisms.
Both serotonin receptors and adrenergic receptors have multiple subtypes that can be grouped into different classes based on their pharmacological and signaling properties (10,11). Among these receptors, 5-HT 1A and ␣ 2 -AR are G i -coupled G protein-coupled receptors enriched in dendritic spines (12,13) where glutamatergic transmission occurs. Abnormal expression and signaling mediated by 5-HT 1A and ␣ 2 -AR have been highly implicated in psychiatric disorders. For example, genetic knock-out of 5-HT 1A or ␣ 2 -AR leads to an elevated anxiety phenotype (14,15). Post-mortem brains of depressed patients show altered receptor levels or ligand binding (16,17). The 5-HT 1A and ␣ 2 -AR gene polymorphisms or altered receptor functions correlate well with the prevalence of depression in humans (18) as well as the effectiveness of antidepressants (19,20).
In the present study, we identified that AMPAR is the cellular target of monoamines in the prefrontal cortex. A low dose of serotonin and norepinephrine reuptake inhibitors (SNRIs), by simultaneously activating 5-HT 1A and ␣ 2 -AR, synergistically suppressed glutamatergic transmission in PFC pyramidal neurons through a mechanism involving AMPAR internalization. Given the significance of AMPARmediated excitation in PFC-dependent functions (21,22), our results provide a potential basis for the better effects of dual reuptake inhibitors in therapeutic intervention of mental disorders related to PFC dysfunction.

EXPERIMENTAL PROCEDURES
Electrophysiological Recordings in Slices and Cultures-The whole-cell voltage clamp technique was used to record AMPAR-mediated excitatory postsynaptic current (AMPAR-EPSC) in PFC layer V pyramidal neurons from postnatal rats (3-4 weeks) as we described previously (23)(24)(25). Slices (300 m) were perfused with artificial cerebrospinal fluid bubbled with 95% O 2 and 5% CO 2 containing D-2-amino-5-phosphonovalerate (25 M) and bicuculline (10 M). The internal solution contained 130 mM cesium methanesulfonate, 10 mM CsCl, 4 mM NaCl, 1 mM MgCl 2 , 10 mM HEPES, 5 mM EGTA, 2.2 mM QX-314, 12 mM phosphocreatine, 5 mM MgATP, and 0.5 mM Na 2 GTP, pH 7.2-7.3, 265-270 mosM. Neurons were visualized with a 40ϫ water immersion lens and illuminated with near infrared (IR) light. All recordings were performed using a Multiclamp 700A amplifier with neurons held at Ϫ70 mV. Tight seals (2-10 gigaohms) were generated by applying negative pressure followed by additional suction to disrupt the membrane and obtain the whole-cell configuration. EPSCs were stimulated by exciting the neighboring cortical neurons with a bipolar tungsten electrode (FHC, Inc.) located at a few hundred micrometers away from the neuron under recording.
Biochemical Measurement of Surface Receptors-The surface AMPA receptors were detected as described previously (23,24). After treatment, PFC slices were incubated in PBS containing 1 mg/ml Sulfo-NHS-LC-Biotin (Pierce) on ice for 40 min. The slices were then rinsed three times in TBS to quench the biotin reaction followed by homogenization in 500 l of modified radioimmune precipitation assay buffer containing 1% Triton X-100, 0.1% SDS, 0.5% deoxycholic acid, 50 mM NaPO 4 , 150 mM NaCl, 2 mM EDTA, 50 mM NaF, 10 mM sodium pyrophosphate, 1 mM sodium orthovanadate, 1 mM PMSF, and 1 mg/ml leupeptin. The homogenates were centrifuged at 16,000 ϫ g for 30 min at 4°C. The supernatant was collected and incubated with NeutrAvidin-agarose (Pierce) for 2 h at 4°C. Bound proteins were washed three times with radioimmune precipitation assay buffer and subjected to SDS-PAGE. Western blotting was performed on biotinylated (surface) and total proteins using antibodies against GluR1 or GluR2 (both at 1:500; Chemicon).
Immunocytochemistry-After transfection and treatment, neurons were fixed in 4% paraformaldehyde in PBS for 20 min at room temperature and incubated with 5% bovine serum albumin (BSA) for 1 h to block nonspecific staining. Cells were then permeabilized with 0.2% Triton X-100 for 20 min at room temperature followed by incubation with the primary antibody at 4°C overnight. For measuring the levels of total and active p38 MAPK, anti-p38 and anti-Thr-180/Tyr-182 phospho-p38 antibodies (1: 500; Cell Signaling Technology) were used. After washing, neurons were incubated with Alexa Fluor 488 (green)-or Alexa Fluor 594 (red)-conjugated secondary antibodies (1:500; Molecular Probes) for 2 h at room temperature. After washing in PBS three times, the coverslips were mounted on slides with VECTASHIELD mounting medium (Vector Laboratories, Burlingame, CA). Images were captured with identical conditions and analyzed with identical parameters. The intensity of p38 signal was quantified using NIH ImageJ software.
Cloning, Expression, and Purification of Proteins-Purified Rab5 protein was generated using a method similar to that described previously (27). Using mouse brain total RNA, wildtype Rab5 cDNA was cloned by RT-PCR followed by sequence verification. The cDNA was then subcloned into the bacterial expression vector pQE-80 (Qiagen, Valencia, CA), which added a His 6 tag at the N terminus of the protein. Rab5 mutants (S34N and Q79L) were generated by site-directed mutagenesis using the QuikChange kit from Stratagene (La Jolla, CA). Rab5 expression in the M15 strain of Escherichia coli (Qiagen) was induced by adding isopropyl ␤-D-thiogalactoside to 1 mM final concentration for 4 -5 h at 25°C (to minimize the formation of inclusion bodies). Rab5 in cleared E. coli lysate was purified by affinity chromatography using a His Gravitrap column (GE Healthcare) according to the manufacturer's protocol. His 6tagged Rab5 proteins were eluted from the column in a buffer containing 50 mM Tris, 500 mM NaCl, and 300 mM imidazole, pH 7.4. Fractions of eluate were analyzed by SDS-polyacrylamide gel electrophoresis and Coomassie Blue staining to identify the peak fractions containing Rab5 proteins. Western blotting with the polyclonal anti-Rab5 (Santa Cruz Biotechnology) was also performed to verify the expression of purified Rab5 protein. One to two of the most pure fractions (shown as a single band by Coomassie Blue staining) were dialyzed against phosphate-buffered saline before being used in electrophysiological experiments.
Statistics-Data analyses were performed with AxoGraph (Axon Instruments), Kaleidagraph (Albeck Software, Reading, PA), Origin 6 (Microcal Software, Northampton, MA), and Statview (Abacus Concepts, Calabasas, CA). All data are expressed as the mean Ϯ S.E. Experiments with more than two groups were subjected to one-way ANOVA or two-way ANOVA followed by post hoc Tukey tests. Experiments with two groups were analyzed statistically using unpaired Student's t tests.

SNRIs Induce a Synergistic Reduction of AMPAR-mediated Synaptic Transmission in PFC Pyramidal
Neurons-To understand the interactive role of the 5-HT and NE systems in glutamatergic transmission, we recorded AMPAR-EPSC in PFC slices. Reuptake inhibitors fluoxetine and desipramine were used to elevate the synaptic concentration of endogenous 5-HT and NE, respectively. As shown in Fig. 1, A and B, bath application of fluoxetine (10 M) or desipramine (10 M) alone did not significantly alter AMPAR-EPSC (fluoxetine, 7.2 Ϯ 1.7%, n ϭ 5; desipramine, 9.8 Ϯ 1.6%, n ϭ 7; p Ͼ 0.05, ANOVA). However, co-application of both inhibitors induced a substantial reduction of AMPAR-EPSC amplitude by 48.8 Ϯ 4.9% (n ϭ 10; p Ͻ 0.001, ANOVA). Dose-dependent experiments using fluoxetine and/or desipramine are shown in Fig. 1C. AMPAR-EPSC was not sensitive to either inhibitor at a dosage of less than 10 M but was significantly reduced by both inhibitors at low doses (5 M, 23.5 Ϯ 1.8%; 10 M, 49.0 Ϯ 3.9%; p Ͻ 0.001, ANOVA). Because the dual action of fluoxetine and desipramine at 10 M gave a saturating effect on AMPAR currents, we selected this concentration for the remaining experiments.
Next, we tested whether the synergistic reduction of AMPAR-EPSC by SNRIs also occurred in vivo. Animals were intraperitoneally injected with fluoxetine (5 mg/kg; 1 h) and/or desipramine (1 mg/kg; 1 h) before slicing and recording. As shown in Fig. 1, D and E, AMPAR-EPSC was not altered in animals injected with fluoxetine or desipramine alone (saline, n ϭ 12; fluoxetine, n ϭ 11; desipramine, n ϭ 10) but was signif-icantly reduced in those co-injected with desipramine and fluoxetine (desipramine ϩ fluoxetine, n ϭ 10; F 4,192 ϭ 13.8, p Ͻ 0.001, ANOVA). These data suggest that endogenous serotonin To test whether a bona fide SNRI has the same effect as fluoxetine plus desipramine, we examined venlafaxine, an SNRI used for treating certain anxiety disorders with depression. As shown in Fig. 1F, bath application of venlafaxine to cortical slices reduced AMPAR-EPSC in a dose-dependent manner with a significant effect at 50 M (25.5 Ϯ 4.2%; n ϭ 5; F 3,16 ϭ 11.2, p Ͻ 0.001, ANOVA). PFC neurons from animals with an intraperitoneal injection of venlafaxine (8 mg/kg; 1 h) also showed a significant reduction of AMPAR-EPSC (saline, n ϭ 10; venlafaxine, n ϭ 10; F 4,90 ϭ 7.6, p Ͻ 0.001, two-way ANOVA).
The Regulation of AMPARs by SNRIs Involves Dynamin/ Rab5-dependent AMPAR Endocytosis-Because SNRIs reduce surface AMPARs, we tested whether this results from increased internalization of surface AMPARs or from decreased AMPAR transport along cytoskeletons. To do so, we dialyzed neurons with a dynamin inhibitory peptide (50 M) that disrupts amphiphysin-dynamin binding and thus prevents endocytosis through clathrin-coated pits (28). As shown in Fig. 3, A and B, the dynamin inhibitory peptide prevented the reduction of AMPAR-EPSCs by fluoxetine and desipramine (5.8 Ϯ 1.6%; n ϭ 7), whereas a scrambled control peptide was ineffective (46.0 Ϯ 3.3%; n ϭ 6). Dialysis of reagents that perturb microtubule or actin stability failed to alter the effect of SNRIs on AMPAR-EPSC (30 M colchicine, 43.8 Ϯ 2.2%, n ϭ 6; 5 M latrunculin, 41.0 Ϯ 3.2%, n ϭ 6; Fig. 3B), which ruled out the involvement of cytoskeleton-dependent transport of AMPARs. These data suggest that the regulation of AMPARs by SNRIs relies on dynamin-dependent AMPAR endocytosis.
AGS3, a G Protein Modulator, Is Involved in the Synergistic Regulation of AMPARs by SNRIs-We next sought to examine the molecular mechanisms underlying the synergistic regulation of AMPARs by SNRIs. Both ␣ 2 -AR and 5-HT 1A are G i -coupled receptors. G-protein signaling is initiated with the dissociation of ␤␥ and ␣ subunits and is terminated by reassociation of the two. AGS3 binds to ␤␥ subunit and prevents it from reassociating with G i ␣ subunit, resulting in persistent activation of ␤␥-mediated signaling (32). To test whether AGS3 could be the key synergistic modulator in the regulation of AMPARs by SNRIs, we dialyzed neurons with a peptide derived from AGS3 (GRKKRRQRRRPPTMGEEDFFDLLAKSQSKRMDD-QRVDLAK), which interrupts the AGS3-␤␥ binding (33). As shown in Fig. 4, A-C, AGS3 peptide abolished the reduction of AMPAR-EPSC by co-application of fluoxetine and desipramine (AGS3 peptide, 12.2 Ϯ 1.6%, n ϭ 5; control peptide, 40.2 Ϯ 3.5%, n ϭ 5). Moreover, inhibiting intracellular AGS3 function by infusing an AGS3 antibody gave similar blockade (AGS3 antibody, 10.8 Ϯ 1.8%, n ϭ 6; heat-inactivated antibody, 49.3 Ϯ 1.9%, n ϭ 5).
5-HT 1A and ␣ 2 -AR Receptors Synergistically Activate p38 Kinase in an AGS3-dependent Manner-We further investigated the mechanism downstream of AGS3/G protein signaling that leads to the reduction of AMPAR-EPSC by SNRIs. A potential candidate is p38 MAPK, activation of which can induce the removal of synaptic AMPARs (34). To test this, we dialyzed neurons with p38 inhibitor SB2303480 (20 M). The p44/42 MAPK inhibitor PD98059 (40 M) was used as a control. As shown in Fig. 5, A and B, fluoxetine and desipramine lost the capability to reduce AMPAR-EPSC significantly in the presence of SB203480 (11.0 Ϯ 1.7%; n ϭ 6) but not PD98059 (45.6 Ϯ 3.5%; n ϭ 5).

DISCUSSION
Monoamine hypofunction has long been proposed to underlie the pathophysiology of depression and anxiety disorders (6,36). Data suggest that the interplay of NE with 5-HT may exert neurobiological normalization in mental illnesses (37). Dual enhancement of the NE and 5-HT systems produces better therapeutic outcome than targeting only one monoamine system (4,8). The class of SNRIs now comprises three medications, venlafaxine, milnacipran, and duloxetine, all of which are efficacious in treating a variety of anxiety disorders and helpful in relieving chronic pain associated with and independent of depression (38). Evidence from meta-analysis of clinical trials supports the greater efficacy of venlafaxine and duloxetine in moderate to severe depression compared with selective serotonin reuptake inhibitors (9).
Using combined electrophysiological, biochemical, and immunocytochemical approaches, we found that application of low dose SNRI synergistically decreased AMPAR surface expression and function in PFC pyramidal neurons. Because high concentrations of single reuptake inhibitors are toxic and even lethal (39), our results suggest that using a low dose SNRI may provide more effective regulation of synaptic transmission in the prefrontal cortex, which is important for cognitive and emotional processes. Moreover, the SNRI-induced synergistic activation of signaling cascades to alter synaptic function may accelerate the onset of therapeutic effects.
Pharmacological evidence suggests that the synergistic regulation of AMPARs by SNRIs is mediated by 5-HT 1A and ␣ 2 -AR receptors, both of which are enriched at glutamatergic synapses (12,13). An important player involved in the synergistic actions of ␣ 2 -AR and 5-HT 1A receptors is AGS3, which prolongs ␤␥-mediated signaling by preventing the reassociation of G i ␣ subunit (32). Consistent with this finding, AGS3 also mediates the synergy between -opiate and cannabinoid CB1 receptors (40).
Co-stimulation of 5-HT 1A and ␣ 2 -AR receptors induces AGS3-dependent activation of p38 MAPK, probably via G protein ␤␥ subunit, small GTPase Rap1, and MAPK kinases 3/6, similarly to metabotropic glutamate receptors (41). Activation of p38 MAPK was found to be required for the removal of synaptic AMPARs and synaptic depression (34). Our previous study has also shown that G protein-coupled receptors, such as metabotropic glutamate receptors, could induce the activation of p38 MAPK, leading to Rab5-mediated AMPAR endocytosis and long term depression in PFC neurons (42). Here, we have demonstrated that p38 MAPK is the key molecule responsible for SNRI reduction of AMPAR-EPSC. p38 MAPK was found to modulate endocytic trafficking by regulating the functional cycle of Rab5 between the membrane and cytosol that is controlled by guanyl nucleotide dissociation inhibitor (43). Thus, SNRI activation of p38 MAPK may accelerate AMPAR endocytosis by stimulating the formation of Rab5-guanyl nucleotide dissociation inhibitor complex. Another possibility is that the SNRI alters AMPAR recycling because some AMPA receptors undergo constitutive endocytic recycling, and dynamin and Rab5 blockade could impact receptor recycling directly. Taken together, our study has revealed the synergistic regulation of PFC glutamatergic transmission by co-activation of the serotonin and norepinephrine systems, providing a potential basis for the better therapeutic effects of dual reuptake inhibitors in the treatment of mental disorders.