|Home | About | Journals | Submit | Contact Us | Français|
Mood disorders and antidepressant therapy involve alterations of monoaminergic and glutamatergic transmission. The protein S100A10 (p11) was identified as a regulator of serotonin receptors, and has been implicated in the etiology of depression and in mediating the antidepressant actions of selective serotonin reuptake inhibitors (SSRIs). Here we report that p11 can also regulate depression-like behaviors via regulation of a glutamatergic receptor in mice. p11 directly binds to the cytoplasmic tail of metabotropic glutamate receptor 5 (mGluR5). p11 and mGluR5 mutually facilitate their accumulation at the plasma membranes, and p11 increases cell surface availability of the receptor. While p11 overexpression potentiates mGluR5 agonist-induced calcium responses, overexpression of mGluR5 mutant, which does not interact with p11, diminishes the calcium responses in cultured cells. Knockout of mGluR5 or p11 specifically in glutamatergic neurons in mice causes depression-like behaviors. Conversely, knockout of mGluR5 or p11 in GABAergic neurons causes antidepressant-like behaviors. Inhibition of mGluR5 with an antagonist, MPEP, induces antidepressant-like behaviors in a p11-dependent manner. Notably, the antidepressant-like action of MPEP is mediated by parvalbumin-positive GABAergic interneurons, resulting in a decrease of inhibitory neuronal firing with a resultant increase of excitatory neuronal firing. These results identify a molecular and cellular basis by which mGluR5 antagonism achieves its antidepressant-like activity.
Major depressive disorder (MDD) is a devastating psychiatric disease with an estimated lifetime incidence of more than 12 % in men and 20% in women in the United States1. Although dysfunction of monoaminergic and glutamatergic transmission has been implicated in MDD1, the pathogenic mechanisms underlying the disease are not well understood. Currently available antidepressant drugs, which influence monoaminergic systems, are effective only for 30%–50% of patients and have critical drawbacks such as delayed onset of therapeutic efficacy. Thus, the development of improved therapies is necessary.
p11 has been implicated in the etiology of depression and in the mechanism of action of selective serotonin reuptake inhibitors (SSRIs)2. p11 was initially identified as a binding protein for serotonin receptors 1B, 1D and 4 (5-HTR 1B, 1D and 4) using a yeast two hybrid screen3, 4. The levels of p11 mRNA and protein in the brain are down-regulated in depressed humans and in a mouse model of depression3, 5. In contrast, the levels of p11 are increased by electroconvulsive therapy or chronic administration of monoaminergic antidepressants including SSRIs2, 3, 6. p11 null mice showed depression-like behaviors, and reduced neurogenic and behavioral responses to the SSRIs2, 3, 7, 8. Conversely, mice overexpressing p11 showed antidepressant-like behaviors3. Recently we have found a chromatin-remodeling factor, called SMARCA3, as a binding partner of p11 and showed a central role for SMARCA3 in p11-dependent neurogenic and behavioral responses to the SSRIs6.
Metabotropic glutamate receptors (mGluRs), a sub-family of glutamate receptors, are G protein-coupled receptors. mGluRs are divided into three groups based on the modes of G-protein coupling and signaling pathways: group I (mGluR1 and 5), group II (mGluR 2 and 3), and group III (mGluR4, 6, 7 and 8)9. Previous studies have shown that antagonists acting on mGluR5 or mGluR2/3 exert antidepressant-like activities in mice10–12. We became particularly interested in mGluR5 because we have found a potential p11 binding motif in the cytoplasmic tail of mGluR5. Thus, we aimed at investigating the potential interaction between mGluR5 and p11, and a possible role for p11 in antidepressant-like activities of an mGluR5 antagonist. In this study, we have performed a variety of biochemical, cell biological, imaging, electrophysiological and behavioral experiments, and elucidated a mechanism by which mGluR5 antagonism achieves its antidepressant-like activity.
All procedures for biochemical and behavioral experiments involving animals were approved by the Rockefeller University Institutional Animal Care and Use Committee and were in accordance with the National Institutes of Health guidelines. In vivo electrophysiology experiments were conducted in accordance with the directives of the Animal Care and Use Committee of the Korea Advanced Institute of Science and Technology (KAIST, Korea). p11 null mice 3, floxed p11 line 13, and floxed mGluR5 line 14 were generated previously. EMX-Cre (stock# 005628), GAD-Cre (stock# 10802), and PV-Cre (stock# 008069) lines were purchased from The Jackson Laboratory (Bar Harbor, Maine, USA). We produced the progeny for each line by in vitro fertilization (IVF) and embryo transfer (ET) techniques to produce a number of animals with the same age sufficient for the behavioral tests (Transgenic facility, Rockefeller University). 10–16 week old male mice were used for behavioral tests, and housed 2–4 per cage with a 12:12-hr light/dark cycle and ad libitum access to food and water. Mice were assigned to experimental groups based on their genotype. Selection of animal for different drug-treatment was performed randomly and in a blinded fashion. mGluR5 antagonist MPEP (Tocris, Minneapolis, MN, USA) or mGluR2/3 antagonist LY341495 (Tocris) was IP injected at 30 min (for FST )15, 16 and 24 hr (for NSF)17 prior to behavioral tests.
pGEX-6P1-p11, pGEX-6P1-p11-AnxA2 fusion protein, pIRESneo-Flag/HA-EGFP expressing p11-Flag/HA (WT p11) and pIRESneo-Flag/HA-EGFP expressing p11C83Q-Flag-HA (C83Q p11) were previously generated 6. pRK5-Myc-mGluR5 and pRK5-cytoplasmic tail of mGluR5 (aa 826–1171) were generated previously 18. pRK5-cytoplasmic tail of mGluR5 was used as a template to generate p11 binding-defective mutants by site-directed mutagenesis (Genewiz, South Plainfield, NJ, USA). pRK5-Myc-mGluR5 was digested with NheI (1704 of mGluR5) and XbaI (in vector) and the digested region was replaced with synthetic DNA fragment containing p11 binding site mutations (2440-GCTGCTGCTGCTGCT-2454) (Genewiz, , South Plainfield, NJ, USA).
GST or peptide pull down assay was performed as described previously 6. To evaluate the SSV motif as a p11 binding motif, we used a peptide walking method 19. A series of overlapping synthetic peptides (15-mer) covering the SSV sequence were used for the interaction assay with GST-p11-AnxA2 fusion protein. 15-mer peptides coupled with a biotin tag at the C-terminus were immobilized on streptavidin magnetic beads (Dynabead M-280 Streptavidin, Invitrogen, Grand Island, NY, USA) and incubated with purified GST-p11-AnxA2 fusion protein. The bound protein samples were subjected to SDS-PAGE and western blot analysis.
COS7 cells (CRL-1651, ATCC) were co-transfected with pRK5-Myc-mGluR5 plus control RNA duplex (Stealth RNAi™ siRNA Negative Control Med GC Duplex #2, Cat. No 12935-112, Life Technologies, Grand Island, NY, USA) or p11 siRNA (Stealth p11 siRNA, Cat. No MSS276909, Life Technologies), or plus control empty vector or pIRESneo-p11-Flag-HA. For the study of surface expression of WT and MT mGluR5, COS7 cells were transfected with pRK5-Myc-WT-mGluR5 or pRK5-Myc-MT-mGluR5. The analysis of cell surface expression with biotinylation was performed as described previously3. The synaptosomal fraction was prepared from the prefrontal cortex or hippocampus as described previously20.
HEK293 cells (CRL-1573, ATCC) were transfected with pIRESneo-p11-Flag-HA (p11-WT) and pIRESneo-p11C83Q-Flag-HA (p11 C83Q mutant). Immunoprecipitation was performed with anti-Flag affinity gel (A2220, Sigma, St Louis, MO, USA) as described previously6. Immunoblotting was performed with a standard protocol using the following antibodies: mGluR5 (rabbit monoclonal (1:5000), Abcam, Cambridge, MA, USA), AnxA2 (mouse monoclonal (1:1000), sc-28385, Santa Cruz), p11 (for human p11, mouse monoclonal (1:1000), BD Bioscience, San Jose, CA, USA; for mouse p11, goat polyclonal (1:200), R&D systems, Minneapolis, MN, USA), PSD95 (mouse monoclonal (1:2500), clone K28/43, Millipore, Billerica, MA, USA), mGluR2/3 (rabbit polyclonal (1:5000), Millipore) and GAPDH (mouse monoclonal (1:2500), clone 6C5, Chemicon, Billerica, MA, USA). Immunoblot analysis was performed with Odyssey infrared imaging system (LI-COR, Lincoln, NE, USA).
HEK293 cells were seeded on poly-D-lysine (Sigma, St Louis, MO, USA) coated optical glass bottom chamber slides (Thermo Fisher Scientific, Waltham, MA, USA). Transient transfection was performed at ~50% confluence with mGluR5 plasmid DNA complexes using Lipofectamine LTX with PLUS reagents (Invitrogen, Grand Island, NY, USA) according to the manufacturer’s instructions. Twenty four hours later, the cells were fixed with 4% paraformaldehyde in phosphate buffered saline (PBS) for 15 min at room temperature, rinsed with Dulbecco's phosphate buffered saline (DPBS), and then permeabilized with 0.5% Triton X-100 in DPBS. After rinsing the cells in DPBS, cells were incubated with 3% BSA in DPBS for 1 hr at room temperature. Cell samples were subsequently incubated with primary antibody (rabbit monoclonal anti-mGluR5 (1:1000; Abcam, Cambridge, MA, USA); monoclonal anti-p11 (1:500; BD Biosciences, San Jose, CA, USA)) overnight at 4°C, followed by incubation with fluorophore-conjugated secondary antibodies (Alexa Fluor 488 anti-mouse and Alexa Fluor 568 anti-rabbit (1:1000; Invitrogen, Grand Island, NY, USA)) with gentle shaking for 1 hr at room temperature. Confocal images were obtained on a Zeiss LSM 710 confocal imaging system (Carl Zeiss Microscopy, Thornwood, NY, USA) using a 100×/1.4 numerical aperture oil-immersion objective lens. Images were analyzed using the Zen 2010 software (Carl Zeiss Microscopy). Pixel-by-pixel colocalization analysis was computed using custom-made Matlab (The MathWorks, Natick, MA, USA) scripts as previously described 21.
HEK293 cells were seeded on 18 mm coverslips (Marienfeld, Lauda-Königshofen, Germany) and cultured in DMEM (Invitrogen, Carlsbad, CA, USA) containing 10% FBS (Invitrogen). After 24 hours, cells were co-transfected with WT mGluR5 and either pIRESneo-Flag/HA-EGFP expressing p11-Flag/HA (WT p11) or empty vector (as control) using Lipofectamine 2000 (Invitrogen) according to the manufacturer’s recommendations. At 48 hr after transfection, cells were rinsed briefly in Krebs Ringer Buffer, consisting of, in mM, 110 NaCl, 4 KCl, 1 NaH2PO4·H2O, 25 NaHCO3, 1.5 CaCl2·H2O, 1.2 MgCl2·6H2O, 10 D-glucose and 20 HEPES, and subsequently incubated with 5 µM Fura-2 AM (Invitrogen) for 30 min at 37 °C. Cells were rinsed in Krebs Ringer Buffer and left for deesterification of the dye for 15 min at 37 °C. The coverslips were then mounted in an open heated Chamlide chamber (Live Cell Instrument, Seoul, Korea) and perfused with Krebs Ringer Buffer at ~37 °C.
Calcium imaging was performed on a Zeiss Axiovert 200 inverted microscope with a 40×/1.3 NA oil immersion objective (Carl Zeiss, Oberkochen, Germany). At the beginning of every experiment, cells expressing GFP were identified by exciting at 488/10 nm using a Polychrome IV monochromator (FEI, Gräfelfing, Germany) and collecting the emission at 525/50 nm using a cooled CCD camera (ORCA-ER, Hamamatsu, Hamamatsu city, Japan). Fura-2 was excited at 340 nm and 380 nm and its emission was detected at 510/80 nm at an image acquisition rate of 1 Hz using MetaFluor software (Molecular Devices, Sunnyvale, CA, USA). After baseline recording of intracellular calcium, cells were continuously perfused with 10 µM DHPG (Sigma-Aldrich, St Louis, MO, USA) in Krebs Ringer Buffer. Average intensities of responding cells were extracted in the 340/380 ratiometric images in MetaFluor.
To test the mutant mGluR5 deficient in p11 binding, HEK293 cells were transfected with either WT or MT mGluR5. Cells were loaded with 4 µM Oregon Green Bapta 488-1 AM (Invitrogen) and imaged on a Zeiss Observer.D1 inverted microscope (Carl Zeiss) with a 40×/1.3 NA oil immersion objective and equipped with a iXon +897 EMCCD camera (Andor Technology, Belfast, UK), an X-cite exacte lamp (Lumen Dynamics, Ontario, Canada) and a GFP excitation-emission filter set. Average intensities of responding cells were extracted in Zen software (Carl Zeiss).
Frequency analysis of the calcium peaks was performed in a custom written Matlab (The Mathworks, Natick, MA, USA) script. Peaks were defined as increases in fluorescence intensity of more than 10%. An oscillating cell was defined as a cell which displayed at least two peaks. As the frequency of mGluR5-mediated calcium peaks decreased over time and calcium activity occassionally ceased, analysis was performed in two time windows; 5 and 10 minutes after activation, when the majority of responding cells were signaling. The significance of differences in frequencies was tested using Wilcoxons rank sum test in Matlab. The distribution of mean frequencies of calcium peaks was analyzed in control cells (n=179) and p11-overexpressing cells (n=225); in WT mGluR5-expressing cells (n=223) or MT mGluR5-expressing cells (n=87).
All the behavioral tests were performed by experimenters blinded to the genotype of the animals and/or drug treatment, and were done in light cycle. A naïve cohort of animals was used only for a single behavioral test after administration of drugs. For the tests of baseline behaviors, a cohort of animals was used for multiple tests with the following order: sucrose preference test, forced swim test or tail suspension test, and then open field test. In the multiple tests, at least 24 hrs of interval was made between tests. We observed that EMX-p11 KO mice were vulnerable to mild stress. Wild-type littermate mice and EMX-p11 KO mice were subjected to a mild stressor (consecutive I.P. injection without any substance 22) before behavioral tests.
Tail suspension test and forced swim test were carried out as described previously 3, 23. We used an automated TST/FST device (Clever Sys, Reston, VA, USA) for measuring the duration of behavioral immobility. For the tail suspension test, the mice were suspended by the tail and videotaped for 6 min. Immobility during the last 4 min was analyzed. For forced swim test, the mice were placed in a glass cylinder (16 cm diameter, 25cm height), which was half-filled with water (22–24 °C), and videotaped for 6 min. Immobility time during the last 4 min was analyzed.
Sucrose preference test was performed as described previously 6 with some modifications. The mice were given a choice of two water bottles for a 1 day-habituation period and then one water bottle was replaced with a bottle containing 1.5% sucrose solution. The consumption of water and sucrose solution was measured after 24 hr. The sucrose preference was calculated as the ratio of consumed sucrose solution to consumed water.
Novelty suppressed feeding test was performed based on the protocol reported previously 17, in which the treatment with 3mg/kg MPEP for 1 hr and 24 hr provided comparable antidepressant-like effects in NSF17. The mice (2 mice per cage) were IP injected (3mg/kg MPEP) and then starved for 24 hr in the home cage. The mice were habituated in the test room for 30 min. After placing the mice in the corner of the test box (50×50×20cm), the latency to eat a food pellet located in the center of the test box was measured. After the test, the mice were returned to their home cage and allowed to eat food pellets for 10 min and food consumption was measured for home cage feeding.
Open-field test was performed as described 6. Briefly, the mice were habituated for 30 min in the test room in their home cage. Locomotor activity in open field box (50x50x22.5 cm) was measured for 60 min. Automated Accuscan software (Gerber Technology, Tolland, CT, USA) was used to calculate total distance traveled.
For implanting head-plate, C57BL/6J mice were anesthetized with avertin (20 mg/ml of tribromoethanol, 20 µl/g, i.p., Sigma) and placed in a stereotaxic device (David Kopf instruments, Tujunga, California, USA). A custom-designed head-plate with a circular window (1 cm diameter) was implanted by cementing to the skull with Super-Bond (Sun Medical Co., Moriyama City, Shiga, Japan) and dental cement (Stoelting, Wood Dale, Illinois, USA). After 4 days of recovery, mice were habituated to the head-restrained condition for 15 min, and recordings were performed next day. On the date of recording, mice were anesthetized with isoflurane (1.5% in oxygen, Hana Pharm Co., Gangnam-Gu, Seoul), after which their head-plate was fixed to a holder device. Holes were drilled in the skull above the right mPFC for multi-unit activity (2.0 mm AP, 0.4 ML) and the temporal cortex for the ground electrode (2.0 mm AP, 2.0 ML); the dura was cleanly removed to allow insertion of the electrode. A silicon probe electrode (Neuronexus, Ann Arbor, Michigan, USA) was fixed to a micromanipulator (Stoelting) and lowered into mPFC region 24. The holes were sealed with 1.5% liquid agar, and the mice were allowed to recover from the anesthesia. The electrode was connected to a Digital Lynx acquisition system (Neuralynx, Bozeman, Montana, USA) via a 32-channel preamplifier head-stage (HS-32; Neuralynx). Neural signals from mPFC were recorded with a Digital Lynx acquisition system (Neuralynx). Data were digitized at 32 kHz and band-pass filtered at 300–5 kHz for multi-unit activity and at 0.5–50 Hz for local field potentials (LFPs). To elucidate the effect of MPEP treatment on the neuronal firing of mPFC, basal multi-unit activity was recorded for 30 min. Multi-unit activity and LFP were acquired after treatment with MPEP (10mg/kg, Tocris, Avon, Bristol, UK) for 1 hr.
All analyses were performed using Matlab (version R2013a; Mathworks, Natick, Massachusetts, USA) and Neuroexplorer (Nex Technologies, Madison, Alabama, USA) software. Multi-unit activity was extracted from the spikes in four detection sites using threshold-based ‘Spike Extractor software’ (Neuralynx). For single-unit analyses, extracted spikes were clustered using AutoKlustaKwik in SpikeSort 3D software (Neuralynx). Auto clustering was performed with energy and/or PCA1 features. This was followed by manual adjustment of the clusters when the auto-clustering function was unable to define a cluster due to the low quality of recordings. As described in Polina Anikeeva, et al. 25, only high-quality isolated units (L-ratio < 0.2; isolation distance > 15) were used for data analysis. Based on previous work 26, 27, we used the criteria of baseline firing rate >10Hz combined with narrow average spike width (under 250 µm) to classify putative inhibitory interneurons28. Putative inhibitory interneurons were required to have a high-frequency component in inter-spike interval histogram.
Mice were deeply anesthetized with avertin (20 mg/ml of tribromoethanol, 20 µl/g, i.p.) and perfused first with heparin and then with 4% formaldehyde in phosphate-buffered saline (PBS). The brains were removed and post-fixed overnight at 4°C. Coronal sections (40 µm thick) were acquired using a vibratome (Leica VT1000S; Leica, Ernst-Leitz-Straße, Wetzlar, Germany) and collected in PBS. Slices were mounted onto glass slides with Vectashield mounting medium (Vector Labs, Burlingame, California, USA). Brain slices were visualized by fluorescence microscopy (Olympus, Center Valley, Pennsylvania, USA) using a rhodamine filter to detect the position of 1,1' - Dioctadecyl - 3,3,3',3' - tetramethylindocarbocyanine iodide (DII, Sigma Aldrich, St Louis, MO, USA) coated electrodes.
For all graphic data, n indicates number of biological replicates. All quantitative biochemical data are representative of three independent experiments, and all behavioral data are representative of at least two experiments. The sample size for cell-based biochemical experiments was determined by empirical evidence accumulated in our laboratory. To determine the numbers of animal to use per group (for biochemical and behavioral experiments), we based our calculations on empirical data accumulated and also on previous literature. The number of animals for each experiment was appropriate to detect biochemical and behavioral differences. Animal exclusion was not made for data analyses. Based on the number of comparisons and the pattern of data distribution, appropriate statistical tests were used to analyze the data. Unpaired two-tailed t-test was used to evaluate the difference between two groups, in the analysis of the quantification of western blots and baseline behaviors of animals. The variances in two groups were similar in most data sets. When the variances in two groups were significantly different (in the case of Figure 3c), we performed unequal variance t- test (Welch’s correction). Two-way ANOVA followed by post hoc test was used to check the effect of drug treatments (MPEP & LY341495) on different genotypes. For calcium imaging experiments, non parametric Wilcoxon rank sum test was used because the data were not normally distributed. For in vivo recording, paired t-test was used to compare the effect of MPEP before and after treatment. In the case of excitatory neurons, non–parametric Wilcoxon rank test was used because data were not normally distributed. All values are presented as mean ± SEM.
Previously, TASK1 potassium channel and TRPV5/TRPV6 calcium channels were identified as binding partners for p11, and the Ser/Thr-Ser/Thr-Val motif was mapped as a binding motif for p11 29, 30. We found the Ser-Thr-Val sequence in the cytoplasmic tail of mGluR5. Thus, we tested a potential interaction between p11 and mGluR5 by immunoprecipitation experiments using HEK293 cells. We observed that mGluR5 was co-precipitated with p11 (Figure 1a). To address the possible involvement of annexin A2 (AnxA2), which exists as a heterotetramer with p11, in the interaction between p11 and mGluR5, we compared WT p11 and a C83Q p11 mutant which prevents interaction of p11 with AnxA2 6. We found that mGluR5 preferentially interacts with WT p11, indicating that AnxA2 is required for the interaction of p11 with mGluR5 (Figure 1a). In addition, we were able to co-precipitate mGluR5 in a detergent-soluble membrane fraction derived from mouse cortical or hippocampal lysates with GST-p11 fused to an N-terminal peptide of AnxA2 (GST-p11-AnxA2 fusion protein) which mimics p11/AnxA2 heterotetramers 6, 31 (Figure 1b). To evaluate the Ser-Thr-Val sequence in the cytoplasmic tail of mGluR5 as a binding site for p11, a series of overlapping synthetic peptides (15-mer) covering the Ser-Thr-Val sequence were used for the interaction assay with GST-p11-AnxA2 fusion protein (Supplementary Figure S1). We were able to identify a binding motif (amino acid sequence from 836 to 844) for the interaction with p11. Replacement by alanine (MT) of five critical amino acids (WT) in the motif greatly reduced the binding of the cytoplasmic tail of mGluR5 to GST-p11-AnxA2 fusion protein (Figures 1c and d). These results suggest that mGluR5 directly interacts with p11 through its motif containing the Ser-Thr-Val sequence in the cytoplasmic tail.
To test a possible role for p11 in mGluR5 signaling, the effect of an mGluR1/5 agonist, (S)-3,5-dihydroxyphenylglycine (DHPG), on calcium response was analyzed in HEK293 cells. The frequency of DHPG-induced calcium oscillations seen in cells expressing WT mGluR5 was increased by p11 overexpression (Figures 1e and f for DHPG treatment for 10 min; Supplementary Figures S2a and b for DHPG treatment for 5 min). Furthermore, the frequency of DHPG-induced calcium oscillations was decreased in cells expressing MT mGluR5 compared to cells expressing WT mGluR5 (Figures 1g and h for 10 min; Supplementary Figures S2c and d for 5 min treatment). These results indicate an involvement of p11 in the function of mGluR5.
Previously, p11 was shown to increase the surface expression of serotonin receptors and ion channels 3, 29, 30, 32, 33. Here we examined the possibility that p11 might regulate the surface localization of mGluR5. First, we examined the subcellular distribution of p11 and mGluR5 in HEK293 cells using immunofluorescence microscopy. A dramatic redistribution of endogenous p11 from cytosol into plasma membrane was observed in the cells transfected with WT mGluR5 (Figures 2a and b; Supplementary Figures S3a and b). In contrast, when cells were transfected with MT mGluR5 plasmid, p11 was partially present in the membrane but a significant portion remained in the cytoplasm (Figure 2c; Supplementary Figure S3c). A detailed pixel-by-pixel quantitative analysis indicated that WT mGluR5 and p11 predominantly colocalized at the plasma membrane, while a relatively low degree of colocalization for MT mGluR5 and p11 was observed throughout the plasma membrane and cytoplasm (colocalization panels in Figures 2b and c, and Supplementary Figures S3b and c).
Next, we analyzed the cell surface expression of mGluR5 by biotinylation of cell surface proteins after knockdown or overexpression of p11. Knockdown of p11 decreased (Figures 2d and e), whereas overexpression of p11 increased (Figures 2f and g) the surface (biotinylated) level of mGluR5 in COS-7 cells. MT mGluR5 showed a decreased surface expression compared to WT mGluR5 (Figures 2h and i). A reduced level of mGluR5 protein was detected in the synaptosomal fractions from p11 null mice (25% reduction compared to the level from WT mice) (Figures 2j–l). Taken together, these results indicate that p11 and mGluR5 mutually facilitate their accumulation at the plasma membrane, and that p11 increases the surface availability of mGluR5.
In the brain, mGluR5 and p11 are expressed in glutamatergic neurons as well as in GABAergic neurons 8, 34–37. Imbalances between glutamatergic and GABAergic transmission have been implicated in psychiatric disorders 38. To investigate a possible role for mGluR5 and p11 in glutamatergic neurons and in GABAergic neurons in depression-like behaviors, we generated floxed mGluR5 or floxed p11 mice bred with EMX-Cre or GAD-Cre mice to delete mGluR5 or p11 in forebrain glutamatergic or in pan-GABAergic neurons, respectively (Figure 3). We employed the forced swim test (FST) and tail suspension test (TST) to assess immobility and sucrose preference test (SPT) to measure anhedonia of the mice. EMX-mGluR5 KO mice showed depression-like behaviors of increased immobility in FST (Figure 3a) and in TST (Figure 3b), and reduced sucrose consumption (Figure 3c). In contrast, GAD-mGluR5 KO mice displayed an antidepressant-like behavioral phenotype in the FST, TST and SPT (Figures 3d–f). Naïve EMX-p11 KO mice did not show a baseline difference in depression-like behaviors in our previous study7. However, EMX-p11 KO mice were found to be vulnerable to a mild stress (Supplementary Figure S4a). Upon exposure to the mild stress, EMX-p11 KO mice, consistent with the phenotype of EMX-mGluR5 KO mice, displayed depression-like phenotypes in the FST, TST and SPT (Figure 3g–i). In contrast to the results obtained with EMX-p11 KO mice and EMX-mGluR5 KO mice, GAD-p11 KO mice, consistent with the phenotype of GAD-mGluR5 KO mice, displayed antidepressant-like phenotypes in the FST, TST and SPT (Figure 3j–l). Locomotor activity was not altered in any of these KO mice except for the EMX-mGluR5 KO mice, which showed hyperlocomotor activity in open field (Supplementary Figure S4b–e). Together, these results indicate that both mGluR5 and p11 play opposite roles in excitatory versus inhibitory neurons to regulate depression-like behaviors.
There is substantial evidence indicating that mGluR5 antagonists exert antidepressant-like activity 10, 39. Since p11 regulates the surface availability of mGluR5, p11 might regulate the antidepressant-like activity of mGluR5 antagonists. To test a possible role for p11, we examined the antidepressant-like effects of a widely studied mGluR5 antagonist, 2-methyl-6-(phenylethynyl)pyridine (MPEP), in wild-type (WT) and p11 null mice. We employed a novelty suppressed feeding (NSF) test, which represents depression-like and anxiety-like behaviors. The NSF test has been widely used to characterize the behavioral effects of antidepressants including SSRIs6, 7, 40, 41, and has been used to measure antidepressant-like effects of antagonists for mGluRs17, 42. The advantage of the test is that a stress-induced animal model of depression is not required to obtain the behavioral effects of antidepressants. Furthermore, WT mice and p11 null mice were known not to show a baseline difference in the NSF test 8. Thus the behavioral effect of drugs can be compared clearly in both genotypes. We observed that the antidepressant-like effect of MPEP seen in WT mice was abolished in p11 null mice in the NSF test (Figure 4a). We also observed an antidepressant-like effect of MPEP in WT mice in FST (Supplementary Figure S5a). p11 null mice showed an increase of baseline immobility as reported previously4. The antidepressant-like effect of MPEP seen in WT mice was abolished in p11 null mice in the FST (Supplementary Figure S5a). However, the behavioral effect of MPEP was not seen in WT mice in the TST and SPT (Supplementary Figures S5b and c), in which a stress-induced animal model of depression might be required to achieve an antidepressant-like effect of MPEP.
An antagonist acting on mGluR2/3, LY341495, exerted antidepressant-like activities in WT mice in NSF test and FST, but the behavioral effect was not blocked in p11 null mice (Figure 4b and Supplementary Figure S5d). In contrast to mGluR5, mGluR2/3 does not have the p11-binding (Ser/Thr-Ser/Thr-Val) motif in the cytoplasmic tail. Consistently, the surface levels of mGluR2 and 3 are not altered by overexpression of p11 (Supplementary Figures S6a–d). The levels of mGluR2/3 in synaptosomal fractions from p11 null mice are not altered compared to the level from WT mice (Supplementary Figures S6e–g). These results indicate a specific involvement of p11 in the antidepressant-like behavioral effect of the mGluR5 antagonist.
To further investigate the antidepressant-like action of mGluR5 antagonism, we examined the behavioral effect of MPEP in cell type-specific KO mice of p11 or mGluR5 by using the NSF test. The antidepressant-like effect of MPEP shown in the wild type littermates was well preserved in EMX-p11 and EMX-mGluR5 KO mice (Figures 4c and f), whereas GAD-p11 KO and GAD-mGluR5 mice showed a baseline antidepressant-like phenotype in the NSF test and no further antidepressant-like effect by MPEP (Figures 4d and g). Since PV-positive interneurons account for the largest population of GABAergic interneurons 43, we generated floxed p11 or floxed mGluR5 mice bred with PV-Cre mice to delete p11 or mGluR5 in PV-positive GABAergic interneurons. PV-p11 KO mice did not show a baseline difference in NSF and the antidepressant-like effect of MPEP seen in WT mice was abolished in PV-p11 KO mice in the NSF test (Figure 4e). We also observed that PV-p11 KO mice did not show a baseline difference in FST and locomotor activity compared to their wild-type littermates (Supplementary Figures S7a and b). In contrast to PV-p11 KO mice, PV-mGluR5 KO mice mimicked the phenotype of GAD-mGluR5 KO mice in the NSF test (Figures 4g and h). The antidepressant-like phenotypes of PV-mGluR5 KO mice were also observed in FST and these mice showed hyperlocomotor activity in open field (Supplementary Figures 7c and d). Neither genetic KO nor MPEP treatment caused any significant change in home cage feeding (Supplementary Figure S8). In summary, the antidepressant-like effect of MPEP seen in WT mice was abolished in PV-p11 KO mice, and the antidepressant-like behavior induced by MPEP seen in WT mice was mimicked by genetic deletion of mGluR5 in PV-positive GABAergic interneurons. These results suggest a primary role for PV-positive GABAergic interneurons in the antidepressant-like action of the mGluR5 antagonist.
Since mGluR5 is known to positively regulate GABAergic neurons 44–46, treatment with MPEP would be expected to inhibit GABAergic interneurons and thereby reduce inhibition of glutamatergic neurons. To elucidate the possible mechanism of MPEP action, we recorded in vivo neural activities. We analyzed medial prefrontal cortex (mPFC) (Supplementary Figure S9) because alterations of mPFC have been implicated in depression and in antidepressant-like behavioral effects 47–50. Neural activities of mPFC in awake mice were acquired before (basal firing rate) and after MPEP treatment, and different responses of firing rate (increase, decrease, or no change) were measured among populations of inhibitory and excitatory neurons (Figure 5a). After MPEP treatment, 75 % of GABAergic (inhibitory) neurons showed highly decreased firing rates (Figures 5a and b). In contrast, 57 % of glutamatergic (excitatory) neurons showed significant enhancement of neuronal firing rates compared to the basal firing rates (Figures 5a and c). This electrophysiological result, consistent with our behavioral data, indicates that GABAergic interneurons are a primary target of the mGluR5 antagonist. In conclusion, our results suggest that the antidepressant-like action of MPEP is driven by inhibition of PV-positive GABAergic interneurons with a resultant increase in glutamatergic transmission.
In this study, we have identified p11 as a regulator of the surface expression of mGluR5. We have shown that both p11 and mGluR5 play opposite roles in excitatory versus inhibitory neurons to regulate depression-like states. Deletion of p11 or mGluR5 in forebrain glutamatergic neurons caused depression-like behaviors but deletion of p11 or mGluR5 in GABAergic neurons caused antidepressant-like behaviors (Figure 3). The most parsimonious explanation for our data is that mGluR5 acts as an amplifier in both excitatory and inhibitory neurons. In excitatory neurons, a decrease of glutamatergic activity causes depression-like behavioral phenotypes (Figure 5d, upper middle). Conversely, in inhibitory neurons, the decrease of mGluR5 activity decreases GABA release, and increases glutamatergic activity, thereby causing an antidepressant-like behavioral phenotype (Figure 5d, upper right). Importantly, the inhibition of mGluR5 function (either by genetic KO or pharmacological inhibition by MPEP) in glutamatergic neurons is overcome by decrease of inhibitory input from GABAergic neurons as a result of inhibition of mGluR5 function in GABAergic neurons (Figure 5d, lower right), explaining how MPEP induces an antidepressant-like activity in WT mice (Figure 5d, lower left).
Previous studies showed that p11 null mice showed a depression-like phenotype3, whereas mGluR5 null mice are known to show an antidepressant-like phenotype16. In addition, p11 KO and mGluR5 KO in some cell types also resulted in different phenotypes (Figures 4e and h; Supplementary Figure S4b and d; Supplementary Figure S7). The difference in the behavioral phenotypes of the KO mice is likely due to their different expression patterns in the brain. While mGluR5 is widely expressed through cortex, hippocampus and striatum51, high level of p11 expression was observed only in selective cell types in the brain2, 6, 7, 52. Thus, p11 is likely a cell type-specific regulator of mGluR5 rather than a general regulator. Previously, various p11 binding proteins were identified including serotonin receptors, channels and nuclear proteins3, 4, 6, 53. Multiple effectors of p11 might also affect the behavioral phenotypes of p11 KO mice, resulting in somewhat different phenotypes compared to mGluR5 KO mice. As in the case of mGluR5, p11 causes increase of the levels of 5-HTR1b3, 5-HTR44, ASIC1a33, TASK129, TRPV5/630 and Nav1.8.32 at the cell surface. So far, only a few studies have examined depression-like behavioral phenotypes in knockout animal models for these proteins. Genetically disrupting the 5-HTR4 receptor in mice caused a decrease in stress-induced hypophasia and novelty-induced exploratory activity, which are related to depressive behavioral paradigms54. Deletion of ASIC1a in mice resulted in antidepressant-like behaviors observed as a decreased immobility time in FST and TST55. It is possible for these receptors and channels to cooperate or compete with mGluR5 in p11-dependent behavioral responses. It would be interesting to know whether selective knockout of these p11 effectors in glutamatergic versus GABAergic neurons causes opposing behavioral phenotypes as in the case of mGluR5 knockout. Although the interpretation of behavioral phenotypes of KO mice is complicated with multiple factors being involved, we were able to show a role for p11 in the regulation of antidepressant-like activity of the mGluR5 antagonist, MPEP. The antidepressant-like effect of MPEP seen in WT mice was abolished in p11 null mice as well as PV-p11 KO mice (Figures 4a and e; Supplementary Figure S5a). Our results suggest that MPEP dominantly acts on PV-positive GABAergic neurons and indirectly modulates glutamatergic neurotransmission (Figures 4f–h, and Figure 5). The antidepressant-like phenotype observed in mGluR5 null mice16 is consistent with pharmacological inhibition of mGluR5, and is explained well by the dominant role for the inhibition of PV-positive GABAergic neurons in the antidepressant-like effect.
Behavioral effects of both mGluR5 antagonism and SSRIs are regulated by p11. However, the neurobiological mechanisms of the actions of the two agents are different. EMX-p11 KO mice blocked the behavioral effects of fluoxetine7 but not the behavioral responses to MPEP (Figure 4c). The therapeutic effect of SSRIs is delayed and can be seen after chronic exposure to the drug for several weeks. During chronic exposure to SSRIs, the levels of p11 and annexin A2 are elevated in the cortex and hippocampus6, 56. Since p11 plays a role in the surface expression of serotonin receptors, the elevated p11 amplifies serotonin signaling2–4. Importantly, the increased p11/annexin A2 complex binds to a chromatin-remodeling factor named SMARCA3 to regulate gene transcription6. The p11/annexin A2 heterotetramer provides two hydrophobic pockets for the binding of a conserved binding motif, which is represented by -P-#-F-X-F (: hydrophobic amino acid, P: proline, #: basic amino acid, F: phenylalanine, X: any amino acid) and is located in the N-terminus of SMARCA36. Notably, this motif does not exist in mGluR5. Since SMARCA3 KO mice blocked neurogenic and behavioral responses to the SSRIs, SMARCA3-regulated downstream genes are likely to mediate the actions of SSRIs6. In contrast, the behavioral effect of mGluR5 antagonists is seen upon acute treatment. In this case, p11 directly binds to mGluR5 through a different binding motif containing a conserved Ser-Thr-Val sequence (Figure 1c and Supplementary Figure S1), and regulates the surface availability of mGluR5 (Figures 2d–l). We did not observe any altered expression of p11 and annexin A2 after treatment with an agonist or an antagonist of mGluR5 (data not shown). However, it is possible that the function of p11 and annexin A2 is upregulated by other mechanisms such as phosphorylation. The fast-acting feature of mGluR5 antagonists may well be explained by altered synaptic transmission as a result of mGluR5 inhibition rather than gene regulation.
It is still obscure whether p11 increases membrane insertion of the mGluR5 or decreases endocytosis of the receptor to increase the surface expression of mGluR5. In addition to p11, Homer57, Norbin18, Calmodulin58 and Tamalin59 are known to bind to the cytoplasmic tail of mGluR5, and regulate the cell surface localization and signaling of mGluR5 (Supplementary Figure S10). Recently, the linkage between Homer1 and depression was revealed by a genome-wide association study60, supporting a possible role for mGluR5 trafficking in the pathophysiology of depression. Future studies should address cell biological mechanisms by which these binding proteins regulate the trafficking of mGluR5. In addition, cooperative or competitive relationship between these binding proteins to regulate trafficking of mGluR5 also remains to be addressed.
Our study elucidates a molecular and cellular mechanism by which mGluR5 antagonism achieves its antidepressant-like activity. In addition to the treatment of depression10, 39, mGluR5 has been considered as a potential therapeutic target for a variety of neurological disorders 61–64. mGluR5 antagonists are attractive candidates for the treatment of Fragile X Syndrome65 and of levodopa-induced dyskinesias (LID) in Parkinson’s disease66. In contrast, positive allosteric modulators (PAM) of mGluR5 represent an attractive strategy to alleviate cognitive deficits associated with schizophrenia by facilitation of NMDA receptor function64, 67, 68. Our study suggests a possible involvement of specific cell types in the therapeutic actions of such mGluR5 agents, and can guide future investigations to identify target cells for the treatment of neurological diseases.
We thank Dr. H. Wang for discussion and sharing materials. We thank Dr. A. Contractor for providing us floxed mGluR5 mice. We thank C. Hsaio for assistance with animal maintenance. We acknowledge R. Norinsky and the Rockefeller University Transgenics Services Laboratory for their excellent IVF services, and H. Zebroski III and the Rockefeller University Proteomics Resource Center for peptide synthesis. We thank Drs. K. Thomas and A. North at the Rockefeller University Bio-imaging Resource Center for their help with fluorescent microscopy. We acknowledge E. Griggs for graphics. This work was supported by DOD/USAMRMC Grants W81XWH-09-1-0392 (Y.K.), W81XWH-10-1-0691 (M.F.) and W81XWH-09-1-0402 (P.G.); NIH grants MH074866 (P.G.) and DA010044 (P.G.); the Fisher Center for Alzheimer’s Research Foundation (P.G.); The JPB Foundation (P.G.); Swedish Research Council and the Erling-Persson Family Foundation (A.A.); the National Leading Research Laboratory Program, 2011-0028772 (D.K.).
CONFLICT OF INTEREST
Authors declare no conflict of interest.