Selective mGlu1 Receptor Activation Restores Cortical Inhibitory Tone and Improves Cognitive Performance in a Subchronic Phencyclidine NMDA Hypofunction Model Authors: Deborah J Luessen, Isabel M Gallinger, Brenna Wolfe, Mac J Meadows, Michael Z Leonard, Geanne A Freitas, Craig W Lindsley, Erin S Calipari, Colleen M Niswender, P Jeffrey Conn Published: 2026 ABSTRACT AbstractBackgroundDeficits in prefrontal cortex (PFC) inhibitory transmission are thought to underlie the cognitive symptoms of schizophrenia that remain untreated by current pharmacotherapies. The metabotropic glutamate 1 (mGlu1) receptor, a subtype of the mGlu receptor family, is a promising therapeutic target because its activation enhances inhibitory transmission in the PFC through the excitation of somatostatin-expressing GABAergic (gamma-aminobutyric acidergic) interneurons (SST-INs).MethodsWe used a subchronic phencyclidine (PCP) NMDA receptor antagonist mouse model, mGlu1 positive allosteric modulators (PAMs), and mice with selective deletion of mGlu1 receptors from SST-INs to examine the cell type–specific role of mGlu1 signaling in cortical inhibition and cognition. We used ex vivo whole-cell patch-clamp electrophysiology to assess inhibitory transmission in the PFC, and cognitive performance was evaluated using tests of working memory and novel object recognition.ResultsSubchronic PCP treatment impaired inhibitory synaptic transmission in the PFC, particularly onto layer V pyramidal neurons projecting to the nucleus accumbens. Pharmacological activation of mGlu1 restored inhibitory drive and reversed PCP-induced deficits in working memory and novel object recognition. Importantly, selective deletion of mGlu1 receptors from SST-INs abolished the ability of mGlu1 activation to restore cortical inhibition and cognitive function, demonstrating that these receptors are required for the therapeutic effects of mGlu1 PAMs.ConclusionsThese findings identify SST-IN–expressed mGlu1 receptors as critical regulators of prefrontal cortical inhibition and cognition. Positive allosteric modulation of mGlu1 represents a promising therapeutic strategy for treating cognitive deficits associated with schizophrenia by restoring inhibitory circuit function.Keywords: Cognition, Electrophysiology, Excitation-inhibition balance, Neuropharmacology, Schizophrenia Cognitive deficits are a core feature of schizophrenia and are a major determinant of long-term functional disability; however, they remain largely untreated by current antipsychotic medications (1, 2, 3, 4, 5, 6). Converging clinical and preclinical evidence implicates deficient prefrontal cortex (PFC) GABAergic (gamma-aminobutyric acidergic) signaling in schizophrenia-associated cognitive dysfunction, suggesting that disrupted cortical inhibition contributes directly to impaired executive processes (7, 8, 9, 10, 11, 12, 13, 14, 15, 16). Accordingly, therapeutic strategies that restore inhibitory transmission within the PFC may improve cognitive function in schizophrenia (17, 18, 19). The metabotropic glutamate 1 (mGlu1) receptor, a subtype of the metabotropic glutamate receptor family, has emerged as an important regulator of cortical inhibitory signaling. Pharmacological activation of mGlu1 enhances inhibitory drive onto PFC pyramidal neurons, shifts the excitatory/inhibitory (E/I) balance toward inhibition, and improves cognitive performance in rodent models of schizophrenia (18, 19, 20, 21, 22). However, the cellular mechanisms underlying these procognitive effects remain poorly understood. Because mGlu1 activation selectively excites somatostatin-expressing interneurons (SST-INs) in the PFC (18), we hypothesized that mGlu1 receptors expressed by SST-INs mediate the restoration of cortical inhibition and cognition. NMDA receptor (NMDAR) hypofunction is a leading mechanistic framework for schizophrenia-associated cognitive impairment. Repeated administration of the NMDAR antagonist phencyclidine (PCP) produces persistent cortical inhibitory deficits and cognitive dysfunction, providing a well-established preclinical model for evaluating therapeutic strategies (23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34). Here, we combined a subchronic PCP (scPCP) model with a highly selective mGlu1 positive allosteric modulator (PAM) (VU6004909) and mice lacking mGlu1 receptors selectively in SST-INs to define the cell type–specific role of mGlu1 signaling in regulating cortical inhibition and cognitive function. Methods and Materials Animal Use Male and female C57BL/6J mice were acquired from Jackson Laboratories. Transgenic mice with deletion of GRM1 from SST-INs were generated as described in the Supplement (35). All experimental protocols were approved by the Vanderbilt Institutional Animal Care and Use Committee. For the scPCP exposure paradigm, 5-week-old C57BL/6J male and female mice received daily subcutaneous (s.c.) injections of 10 mg/kg PCP (0.9% saline, 10 μL/kg; s.c.) for 7 days, as described previously (31). RNAscope RNAscope was conducted using the RNAscope Multiplex Fluorescent v2 Assay with probes (see Key Resources Table and the Supplement) from Advanced Cell Diagnostics. The levels of Grm1 within Sst-positive cells in the prelimbic (PL)-PFC were measured as described in the Supplement. Surgery Mice underwent stereotaxic injections of retrograde AAV-GFP into either the nucleus accumbens (NAc) or the basolateral amygdala (BLA), as previously described (18,36) (Supplemental Methods). Electrophysiology Whole-cell recordings from visually identified layer V pyramidal neurons of the PL-PFC were performed as previously described (19) (Supplemental Methods). Multiple cells recorded from each animal were averaged to generate a single biological replicate for statistical analyses. Touchscreen Cognition Testing Spatial working memory was assessed using the touchscreen Trial-Unique Nonmatching-to-Location (TUNL) task, as previously described (37, 38, 39, 40) (Supplemental Methods). After reaching the acquisition criteria, mice received vehicle or VU6004909 (3–30 mg/kg, via intraperitoneal [i.p.] injections) 30 minutes before testing using a within-subject dosing design. Y-Maze Spontaneous alternation was assessed in a Y-maze, as previously described (18,19) (Supplemental Methods). Behavior was tracked using ANY-maze software (version 6). Novel Object Recognition Novel object recognition (NOR) testing was performed as previously described (18) (Supplemental Methods). Mice received vehicle or VU6004909 (30 mg/kg, i.p.) before testing. The recognition index was calculated from the time spent exploring the novel versus the familiar object. Compounds Commercially available drugs were purchased from Tocris and Sigma-Aldrich (Key Resources Table). VU6004909 was synthesized in-house, as described previously (41,42). Data Analysis Data are presented as mean ± SEM. Statistical analyses were performed using GraphPad Prism (version 11). Data were assessed for normality before analysis. Parametric tests, including Student t tests, one- or two-way analysis of variance (ANOVA), and repeated-measures ANOVA followed by Bonferroni post hoc comparisons where appropriate, were used as indicated in the Results and figure legends. Sex was initially included as a biological variable; because no significant sex effects or interactions were detected, the data were pooled for subsequent analyses. Results Selective mGlu1 Activation Reverses PCP-Induced Deficits in Spontaneous Inhibitory Transmission Onto Pyramidal Cells To determine whether scPCP disrupts inhibitory transmission in the PL-PFC, 5-week-old mice received PCP (10 mg/kg, s.c.) or saline once daily for 7 days, followed by whole-cell recordings from layer V pyramidal neurons after a 3-week washout. Isolating GABAergic spontaneous inhibitory postsynaptic currents (sIPSCs) (Figure 1A), we found that PCP treatment significantly reduced sIPSC frequency compared with saline-treated mice (t32 = 4.46, p < .0001) (Figure 1B), whereas sIPSC amplitude was unchanged (t32 = 0.06, p = .95) (Figure 1C). PCP treatment did not alter sIPSC rise time compared with saline-treated mice (Figure S1A), although sIPSC decay time was significantly reduced (Figure S1B). Likewise, PCP treatment did not significantly alter either the frequency (t17 = 1.81, p = .09) (Figure 1D) or amplitude (t17 = 0.74, p = .47) (Figure 1E) of spontaneous excitatory postsynaptic currents (sEPSCs). However, the inhibitory-to-excitatory (I/E) ratio was significantly reduced following PCP treatment (t13 = 3.42, p < .01) (Figure 1F), demonstrating that scPCP exposure shifts prefrontal synaptic balance toward excitation. Selective mGlu1 activation reverses PCP-induced deficits in spontaneous inhibitory transmission onto pyramidal cells. (A) Representative voltage-clamp recordings of sIPSCs from layer V prelimbic prefrontal cortex pyramidal neurons obtained 3 weeks following subchronic saline or PCP (10 mg/kg, s.c.) treatment and following bath application of DHPG (30 μM) and MTEP (3 μM). Scale bar, 2.5 seconds. (B) sIPSC frequency. (C) sIPSC amplitude. (D, E) sEPSC frequency and amplitude. (F) I/E ratio. (G) Effects of DHPG (30 μM), MTEP (3 μM), and the mGlu1 negative allosteric modulator VU0469650 (3 μM) on sIPSC frequency. (H) Bath application of the mGlu1 positive allosteric modulator VU6004909 (10 μM) alone or in the constant presence of DHPG (30 μM) and MTEP (3 μM) while recording sIPSCs in prelimbic-prefrontal cortex layer V pyramidal cells from mice treated with saline (gray bars) or PCP (blue bars). Data are presented as mean ± SEM. Statistical analyses were performed using Student t tests or two-way analysis of variance with Bonferroni post hoc comparisons, as appropriate (n = 2–5 cells/mouse; N = 3–4 mice/group). Open circle represents female mice; closed circle represents male mice. ∗∗p < .01, ∗∗∗∗p < .0001. aCSF, artificial cerebrospinal fluid; I/E, inhibitory/excitatory; mGlu, metabotropic glutamate; ns, not significant; PCP, phencyclidine; s.c., subcutaneous; sEPSC, spontaneous excitatory postsynaptic current; sIPSC, spontaneous inhibitory postsynaptic current; sPSC, spontaneous postsynaptic current. To determine whether mGlu1 activation restores inhibitory transmission following PCP exposure, we bath-applied the group I mGlu receptor agonist DHPG (30 μM) in the continuous presence of the selective mGlu5 negative allosteric modulator MTEP (3 μM) while recording sIPSCs from layer V pyramidal neurons in saline- and PCP-treated mice (Figure 1G). Two-way ANOVA revealed significant main effects of drug application (F2,56 = 22.37, p < .0001) and treatment group (F1,56 = 5.41, p = .024), but no significant drug × treatment interaction (F2,56 = 1.10, p = .34). Bonferroni post hoc comparisons demonstrated that DHPG/MTEP significantly increased sIPSC frequency in both saline- and PCP-treated mice compared with their respective artificial cerebrospinal fluid (aCSF) control animals (p < .0001 for both comparisons) (Figure 1G). Importantly, co-application of the selective mGlu1 negative allosteric modulator VU0469650 (VU'650; 3 μM) abolished the effects of DHPG/MTEP in both treatment groups, restoring sIPSC frequency to levels comparable to those observed in the presence of aCSF alone (Figure 1G). The selective mGlu1 PAM VU6004909 produced comparable physiological effects. Bath application of VU6004909 together with DHPG/MTEP increased sIPSC frequency in both saline- and PCP-treated mice (Figure 1H), indicating that positive allosteric modulation recapitulates the restoration of cortical inhibition produced by orthosteric mGlu1 activation. Selective mGlu1 Activation Reverses Persistent Inhibitory Deficits Onto PFC→NAc Pyramidal Cells Following Adolescent PCP Treatment To determine whether scPCP differentially affected projection-defined PL-PFC neurons, retrograde AAV-GFP was infused into either the NAc or BLA, followed by whole-cell recordings 4 weeks later (Figure 2A). PCP treatment differentially affected spontaneous inhibitory transmission across projection-defined pyramidal neuron populations (Figure 2B). A two-way ANOVA revealed a significant treatment × projection interaction (F1,34 = 8.06, p = .0076), with no significant main effects of treatment (F1,34 = 2.31, p = .14) or projection (F1,34 = 1.69, p = .20). Bonferroni post hoc comparisons demonstrated that PCP significantly reduced sIPSC frequency onto nucleus accumbens (NAc)–projecting pyramidal neurons compared with saline-treated controls, whereas the sIPSC frequency onto BLA-projecting neurons was unaffected. In contrast, PCP treatment did not significantly alter sIPSC amplitude in either projection-defined population (Figure 2C). Likewise, neither the frequency nor the amplitude of sEPSCs differed between saline- and PCP-treated mice in either projection population (Figure 2D, E). To determine whether these projection-specific changes altered the excitatory-inhibitory balance, we calculated I/E ratios for individual projection-defined pyramidal neurons (Figure 2F). A two-way ANOVA revealed a significant main effect of treatment (F1,21 = 6.13, p = .022), but no significant main effect of projection (F1,21 = 3.71, p = .068) or treatment × projection interaction (F1,21 = 2.08, p = .16). Bonferroni post hoc analysis demonstrated that PCP significantly reduced the I/E ratio of NAc-projecting neurons, whereas the I/E ratio of BLA-projecting neurons remained unchanged (Figure 2F). Selective mGlu1 activation reverses persistent inhibitory deficits onto PFC→NAc pyramidal cells following adolescent PCP treatment. (A) Schematic depicting retrograde viral labeling of NAc- or BLA-projecting layer V pyramidal neurons in the prelimbic-PFC. Green fluorescent protein-expressing neurons were identified for whole-cell electrophysiological recordings. Scale bar, 1 μm. (B) sIPSC frequency. (C) sIPSC amplitude. (D, E) sEPSC frequency and amplitude. (F) I/E ratio. (G) Effects of bath application of DHPG (30 μM) and MTEP (3 μM) on sIPSC frequency in PFC→NAc pyramidal neurons. Data are presented as mean ± SEM. Statistical analyses were performed using two-way analysis of variance with Bonferroni post hoc comparisons (n = 2–4 cells/mouse; N = 3–4 mice/group). Open circle represents female mice; closed circle represents male mice. ∗p < .05, ∗∗p < .01, ∗∗∗p < .001, ∗∗∗∗p < .0001. aCSF, artificial cerebrospinal fluid; BLA, basolateral amygdala; I/E, inhibitory/excitatory; mGlu, metabotropic glutamate; NAc, nucleus accumbens; ns, not significant; PCP, phencyclidine; PFC, prefrontal cortex; sEPSC, spontaneous excitatory postsynaptic current; sIPSC, spontaneous inhibitory postsynaptic current; sPSC, spontaneous postsynaptic current. Bath application of DHPG (30 μM) in the continuous presence of the selective mGlu5 negative allosteric modulator MTEP (3 μM) significantly increased sIPSC frequency in both saline- and PCP-treated mice (Figure 2G). Two-way ANOVA revealed significant main effects of drug application (F1,33 = 45.03, p < .0001) and treatment group (F1,33 = 8.66, p = .0059), but no significant drug × treatment interaction (F1,33 = 0.001, p = .97). Bonferroni post hoc comparisons demonstrated that DHPG/MTEP significantly increased sIPSC frequency in both saline- and PCP-treated mice compared with their respective aCSF controls. In summary, scPCP exposure produces projection-specific deficits in inhibitory transmission selectively within the PFC→NAc pathway that are reversible by mGlu1 receptor activation. mGlu1 PAM VU6004909 Improves Cognitive Performance Following scPCP Treatment Persistent effects of scPCP exposure on spatial working memory were assessed using the touchscreen TUNL task (37, 38, 39,43) (Figure 3A). Saline- and PCP-treated mice acquired the touchscreen task at comparable rates, requiring a similar number of sessions to reach the acquisition criteria (t14 = 0.47, p = .65) (Figure 3A). Mice were then tested under nondelay (0-second) and delay (6-second) conditions. Two-way ANOVA revealed significant main effects of delay (F1,28 = 8.18, p < .01) and treatment group (F1,28 = 7.33, p < .01), as well as a significant delay × treatment interaction (F1,28 = 16.65, p < .001). Bonferroni post hoc comparisons demonstrated that saline-treated mice exhibited lower response accuracy during 6-second delay trials than nondelay trials, whereas PCP-treated mice displayed significantly reduced response accuracy during nondelay trials compared with saline-treated control mice (Figure 3B). mGlu1 positive allosteric modulation improves touchscreen cognitive performance following subchronic PCP treatment. (A) Acquisition of touchscreen responding. (B) Response accuracy during nondelay and 6-second delay Trial-Unique Nonmatching-to-Location trials. (C) Response accuracy during nondelay trials following vehicle or VU6004909 (30 mg/kg, intraperitoneal) treatment. (D) Response accuracy during 6-second delay trials following vehicle or VU6004909 treatment. Data are presented as mean ± SEM. Statistical analyses were performed using Student t tests or two-way analysis of variance with Bonferroni post hoc comparisons, as appropriate (N = 8 mice/group). Open circle represents female mice; closed circle represents male mice. ∗p < .05, ∗∗p < .01, ∗∗∗p < .001. mGlu, metabotropic glutamate; ns, not significant; PCP, phencyclidine; veh, vehicle. We next determined whether the selective mGlu1 positive PAM VU6004909 (30 mg/kg, i.p.) could improve TUNL performance following saline or PCP treatment. During nondelay trials, two-way ANOVA revealed significant main effects of VU6004909 treatment (F1,28 = 12.77, p = .0013) and PCP exposure (F1,28 = 9.69, p = .0042), but no treatment × PCP interaction (F1,28 = 2.79, p = .11). Bonferroni post hoc comparisons demonstrated that VU6004909 significantly increased response accuracy in PCP-treated mice compared with vehicle-treated PCP mice (Figure 3C). During delay trials, two-way ANOVA revealed a significant main effect of VU6004909 treatment (F1,28 = 16.85, p = .0003), but no significant main effect of PCP exposure (F1,28 = 0.34, p = .56) or treatment × PCP interaction (F1,28 = 0.00, p > .9999). Bonferroni post hoc comparisons demonstrated that VU6004909 significantly increased response accuracy in both saline- and PCP-treated mice compared with their respective vehicle-treated mice (Figure 3D). Together, these findings show that mGlu1 PAMs improve touchscreen cognitive performance following adolescent PCP exposure. scPCP-Induced Impairments in NOR Are Reversed by mGlu1 PAM VU6004909 Recognition memory was assessed using NOR 3 weeks after PCP or saline treatment following pretreatment with VU6004909 (30 mg/kg, i.p.) or vehicle (Figure 4A). Recognition memory was quantified as the novelty recognition index (Figure 4B). Two-way ANOVA revealed significant main effects of VU6004909 treatment (F1,44 = 11.20, p = .0017) and PCP exposure (F1,44 = 7.44, p = .0092), as well as a significant treatment × exposure interaction (F1,44 = 12.43, p = .0010). Bonferroni post hoc comparisons demonstrated that PCP-treated mice exhibited a significantly lower recognition index than saline-treated mice, indicating impaired recognition memory (Figure 4B). Pretreatment with VU6004909 significantly increased the recognition index in PCP-treated mice, restoring performance to levels comparable with those of saline-treated control mice, whereas VU6004909 did not significantly alter recognition memory in saline-treated mice (Figure 4B). mGlu1 positive allosteric modulation reverses subchronic PCP-induced impairments in novel object recognition. (A) Schematic of the NOR task. (B) Novelty recognition index following saline or PCP treatment and vehicle or VU6004909 (30 mg/kg, intraperitoneal) administration. (C) Raw interaction times with familiar and novel objects during the NOR task. Total distance traveled is shown in Figure S2. Data are presented as mean ± SEM. Statistical analyses were performed using two-way analysis of variance with Bonferroni post hoc comparisons (N = 11–13 mice/group). Open circle represents female mice; closed circle represents male mice. ∗p < .05, ∗∗p < .01, ∗∗∗p < .001, ∗∗∗∗p < .0001. mGlu, metabotropic glutamate; NOR, novel object recognition; PCP, phencyclidine; s.c., subcutaneous; veh, vehicle. To further characterize object exploration behavior, we analyzed raw interaction times with the familiar and novel objects (Figure 4C). Two-way ANOVA revealed a significant main effect of object (F1,88 = 4.09, p = .046), no significant main effect of treatment group (F3,88 = 0.70, p = .55), and a significant object × treatment interaction (F3,88 = 11.10, p < .0001). Bonferroni post hoc comparisons demonstrated that saline-treated mice spent significantly more time exploring the novel object than the familiar object, whereas PCP-treated mice failed to exhibit a preference for the novel object. In contrast, PCP-treated mice pretreated with VU6004909 displayed significantly greater exploration of the novel object than the familiar object, consistent with restoration of recognition memory (Figure 4C). Total object exploration during the NOR training phase did not differ among treatment groups (Figure S2), indicating comparable initial object engagement and familiarity acquisition prior to testing. Likewise, total distance traveled during the NOR test phase did not differ among treatment groups (Figure S2), indicating that the observed differences in recognition memory were unlikely to reflect alterations in exploratory behavior or locomotor activity. Together, these findings demonstrate that scPCP exposure produces persistent impairments in recognition memory that are reversed by positive allosteric modulation of mGlu1. Selective Knockdown of mGlu1 Receptors From SST-INs Occludes the Effects of mGlu1 Activation on Cortical Inhibition To determine whether SST-IN mGlu1 receptors mediate cortical inhibition, we generated mice with selective knockdown of Grm1 from SST-INs (Grm1-SST-KD). RNAscope fluorescent in situ hybridization confirmed efficient recombination, with wild-type (WT) mice displaying robust Grm1 transcript co-expression within Sst-positive interneurons in the PFC (Figure 5A), whereas Grm1-SST-KD mice exhibited a marked reduction in Grm1 puncta per SST-IN (t13 = 4.46, p = .004) (Figure 5B). Next, we examined whether knockdown of mGlu1 receptors from SST-INs altered basal synaptic transmission onto layer V pyramidal neurons. Whole-cell recordings revealed no significant differences between Grm1-SST-KD and WT littermate controls in sIPSC frequency (t22 = 0.09, p = .93) (Figure 5C) or amplitude (t22 = 0.66, p = .52) (Figure 5D). Likewise, neither the frequency (t20 = 1.43, p = .17) (Figure 5E) nor the amplitude (t20 = 0.46, p = .65) (Figure 5F) of sEPSCs differed between genotypes. The I/E ratio was comparable between WT and Grm1-SST-KD mice (t12 = 0.91, p = .38) (Figure 5G), indicating that mGlu1 knockdown in SST-INs does not alter basal excitatory or inhibitory synaptic transmission in the PL-PFC. Selective knockdown of mGlu1 receptors from SST interneurons occludes mGlu1-mediated enhancement of cortical inhibition. (A) Representative RNAscope fluorescent in situ hybridization images showing Grm1 (green) and Sst (red) expression in WT and Grm1-SST-KD mice. Scale bars, 20 μm. (B) Quantification of Grm1 puncta per SST interneuron. (C, D) sIPSC frequency and amplitude. (E, F) sEPSC frequency and amplitude. (G) I/E ratio. (H) Effects of bath application of DHPG (30 μM) and MTEP (3 μM) on sIPSC frequency. Data are presented as mean ± SEM. Statistical analyses were performed using Student t tests or two-way analysis of variance with Bonferroni post hoc comparisons, as appropriate (n = 2–4 cells/mouse; N = 3–4 mice/group). Open circle represents female mice; closed circle represents male mice. ∗∗p < .01, ∗∗∗p < .001. I/E, inhibitory/excitatory; mGlu, metabotropic glutamate; sEPSC, spontaneous excitatory postsynaptic current; sIPSC, spontaneous inhibitory postsynaptic current; SST, somatostatin; WT, wild-type. Knockdown of Grm1 from SST-INs did not alter basal synaptic transmission onto layer V pyramidal neurons. Bath application of DHPG (30 μM) in the continuous presence of the selective mGlu5 negative allosteric modulator MTEP (3 μM) significantly increased sIPSC frequency in WT mice but not in Grm1-SST-KD mice (Figure 5H). Two-way ANOVA revealed a significant main effect of drug application (F1,26 = 17.21, p = .0003), no significant main effect of genotype (F1,26 = 1.66, p = .21), and a significant drug × genotype interaction (F1,26 = 6.49, p = .017). Bonferroni post hoc comparisons demonstrated that DHPG/MTEP significantly increased sIPSC frequency in WT mice compared with aCSF control mice, whereas no effect was observed in Grm1-SST-KD mice. Together, these findings demonstrate that mGlu1 receptors expressed by SST-INs are required for the mGlu1-mediated enhancement of cortical inhibitory transmission. Cognitive Performance Is Impaired in Mice Lacking mGlu1 Receptors in SST-INs To determine whether SST-IN mGlu1 receptors contribute to cognitive function under basal conditions, Grm1-SST-KD mice and WT littermates were evaluated in NOR and Y-maze spontaneous alternation tasks. Grm1-SST-KD mice exhibited a significantly lower novelty recognition index than WT littermate controls (t27 = 4.03, p < .001) (Figure 6A), indicating impaired recognition memory. To further characterize object exploration behavior, we analyzed raw interaction times with the familiar and novel objects (Figure 6B). Two-way ANOVA revealed a significant object × genotype interaction (F1,54 = 5.33, p = .025), with no significant main effects of object (F1,54 = 1.28, p = .26) or genotype (F1,54 = 2.40, p = .13). Bonferroni post hoc comparisons demonstrated that WT mice spent significantly more time exploring the novel object than the familiar object, whereas Grm1-SST-KD mice failed to exhibit a preference for the novel object, consistent with their reduced recognition index. Total object exploration during the NOR training phase did not differ between WT and Grm1-SST-KD mice (Figure S3B), indicating comparable initial object engagement and familiarity acquisition prior to testing. Likewise, the total distance traveled during the NOR test phase did not differ between genotypes (Figure S3A). We further assessed spontaneous alternation behavior in the Y-maze. Grm1-SST-KD mice exhibited significantly lower spontaneous alternation than WT littermate controls (t34 = 2.50, p = .02) (Figure 6C), whereas total locomotor activity during the task did not differ between genotypes (Figure S3C). Together, these findings demonstrate that the knockdown of mGlu1 receptors from SST-INs results in persistent impairments in cognitive performance. Cognitive performance is impaired in mice lacking mGlu1 receptors in SST interneurons. (A) Novelty recognition index in WT and Grm1-SST-KD mice. (B) Raw interaction times with familiar and novel objects during the novel object recognition task. Total distance traveled is shown in Figure S3A. (C) Percentage spontaneous alternation during the Y-maze task. Total locomotor activity is shown in Figure S3B. Data are presented as mean ± SEM. Statistical analyses were performed using Student t tests or two-way analysis of variance with Bonferroni post hoc comparisons, as appropriate (N = 14–21 mice/group). Open circle represents female mice; closed circle represents male mice. ∗p < .05, ∗∗∗p < .001. mGlu, metabotropic glutamate; SST, somatostatin; WT, wild-type. Discussion Using complementary electrophysiological, behavioral, and genetic approaches, the present study identifies mGlu1 receptor signaling as a key regulator of cortical inhibitory circuit function and cognition following developmental NMDAR hypofunction. Selective potentiation of mGlu1 restored inhibitory synaptic transmission onto layer V pyramidal neurons of the PL-PFC, including the PFC→NAc projection that exhibited pronounced disinhibition following adolescent scPCP exposure. Consistent with these physiological effects, treatment with the mGlu1 PAM VU6004909 improved multiple domains of cognition, reversing deficits in novel object recognition and spatial working memory. By combining pharmacological activation with selective deletion of mGlu1 receptors from SST-INs, we further demonstrate that SST-IN–expressed mGlu1 receptors are required for the ability of mGlu1 activation to enhance cortical inhibition and support cognitive function. Together, these findings establish a mechanistic link between mGlu1-dependent regulation of SST-IN function, restoration of prefrontal inhibitory circuitry, and improved cognitive performance, supporting mGlu1 positive allosteric modulation as a promising therapeutic strategy for cognitive deficits associated with schizophrenia-specific mechanisms underlying physiological and behavioral processes relevant to schizophrenia. NMDAR antagonist models serve as a powerful approach to disrupt cortical I/E balance and recapitulate cognitive deficits associated with schizophrenia, allowing for testing of potential treatment approaches (28,30,44, 45, 46, 47, 48, 49). Many studies have focused on the effects of a single dose of NMDAR antagonists; however, subchronic administration of NMDAR antagonists like PCP induces persistent behavioral and physiological deficits reminiscent of those observed in individuals with schizophrenia (25,50, 51, 52). For example, repeated PCP administration significantly decreased levels of GAD67 in the PFC of rats (51). Another study showed that scPCP administration reduced theta oscillations in the medial PFC and induced abnormal cortical synchronization in putative cortical pyramidal cells (52,53). However, there is currently a lack of information on the cell type-specific effects of scPCP administration within the PFC and its subsequent consequences on cognitive function. We found that scPCP treatment decreased cortical inhibition onto layer V pyramidal cells, which is reversed by mGlu1 activation. Previous studies have demonstrated that mGlu1 activation increases the activity of SST-INs in layer V of the PL-PFC (18), leading us to hypothesize that the ability of mGlu1 activation to reverse deficits in cortical inhibition following scPCP treatment is via actions on mGlu1 receptors expressed in SST-INs. As such, we show that mGlu1-mediated effects on spontaneous inhibitory transmission in the PFC are not observed in mice bearing selective deletion of mGlu1 receptors in SST-INs compared with WT littermates. The present study focused on the effects of mGlu1 deletion from SST-INs on PFC physiology and cognitive functions known to be largely regulated by this brain region; however, mGlu1 deletion in the mouse line used in these studies is not restricted to the PFC. Moreover, because Grm1 deletion occurs throughout development in this constitutive SST-specific knockdown line, we cannot exclude the possibility that developmental adaptations contribute to the physiological and behavioral phenotypes observed in adulthood. Likewise, SST-INs in other brain regions have been implicated in the pathophysiology of schizophrenia (54,55) and may also contribute to the observed behavioral phenotypes. Therefore, although our findings support a critical role for SST-IN–expressed mGlu1 receptors in regulating cortical inhibitory transmission and cognitive function, the relative contributions of developmental adaptations and extra-prefrontal SST populations remain to be determined. Future studies examining the effects of VU6004909 in SST-Grm1 knockdown mice will be valuable for further defining the contribution of these receptors to the procognitive actions of mGlu1 PAMs in vivo and for determining whether additional mGlu1-expressing neuronal populations also contribute to these behavioral effects. Although the present studies were performed using the scPCP model of NMDAR hypofunction, disrupted cortical inhibitory signaling and altered E/I balance are convergent features observed across multiple preclinical models of schizophrenia as well as in patients with the disorder (56, 57, 58). Thus, the restoration of cortical inhibition by mGlu1 receptor activation may represent a broader therapeutic mechanism beyond the scPCP model. Nevertheless, future studies examining mGlu1 PAMs in complementary models of schizophrenia, including genetic and developmental models, will be important for determining the generalizability of these findings. Although both male and female mice were included throughout these studies, and sex was initially incorporated as a biological variable in all statistical analyses, no significant main effects of sex or interactions involving sex were detected; therefore, the data were pooled across sexes. However, the present study was not specifically powered to detect sex-dependent differences. Given the established sex differences in schizophrenia risk, symptom presentation, and cortical circuit function, future studies specifically powered to examine sex as an independent biological variable will be important for determining whether mGlu1 receptor signaling and the procognitive efficacy of mGlu1 PAMs differ between males and females. Substantial evidence suggests that cortical inhibition regulates various forms of cognition (59, 60, 61), and cortical interneuron dysfunction and associated reduced overall strength of connections from excitatory cells to interneurons may underlie impaired cognitive function in individuals with schizophrenia (59,60,62). These findings are consistent with the human postmortem literature demonstrating selective abnormalities in SST-INs in schizophrenia. For example, reduced SST messenger RNA expression has been reported in the dorsolateral PFC of individuals with schizophrenia and schizoaffective disorder, supporting the hypothesis that impaired SST-IN function contributes to cortical circuit dysfunction and cognitive impairment in the disorder (63). The present findings extend these observations by identifying mGlu1 receptor signaling as a potential mechanism through which SST-IN function can be modulated to restore cortical inhibitory transmission and improve cognition. Different subtypes of prefrontal interneurons form specialized networks to orchestrate cognitive performance in learning and memory tasks (64, 65, 66). Because mGlu1 activation increases SST-IN–mediated cortical inhibition (18), we hypothesize that it is through an SST-IN–dependent mechanism that mGlu1 PAMs improve cognitive performance. We found that mice lacking mGlu1 receptors on SST-INs displayed impaired NOR compared with littermate control mice. Interestingly, we did not observe baseline changes in spontaneous inhibitory or excitatory transmission on layer V pyramidal neurons in mice lacking mGlu1 on SST-INs compared with littermate control mice. The absence of detectable differences in basal inhibitory transmission suggests that mGlu1 receptor signaling is not required for maintaining resting inhibitory tone but instead may be recruited dynamically during periods of elevated glutamatergic activity associated with cognitive processing. Because mGlu1 receptors are activated by synaptically released glutamate, their contribution to SST-IN function may become particularly important during memory encoding, retrieval, and other behaviorally relevant network states rather than under basal recording conditions (67,68). Thus, normal baseline synaptic transmission does not necessarily predict intact circuit function during cognitively demanding tasks, providing a potential explanation for the cognitive impairments observed in Grm1-SST knockdown mice despite preserved basal inhibitory transmission. Because mGlu1 receptor activity increases persistent action potential firing of SST-INs in mice (18), loss of mGlu1 receptor–mediated regulation of SST-INs may decrease SST-IN excitation and lead to hyperexcitability of pyramidal cells following the release of glutamate during memory tasks. SST-INs influence pyramidal cell activity directly or through inhibition of parvalbumin (PV)-INs, resulting in perisomatic inhibition of pyramidal neurons (69). Thus, the ability of mGlu1 PAMs to reverse cognitive deficits could be driven by the impaired ability of SST-INs to inhibit PV-IN activity. Disruptions in E/I balance within the PFC have been tightly linked to deficits in executive functions, including working memory, attention, and cognitive flexibility. However, considerably less is known about how cortical disinhibition influences downstream projection-defined circuits that mediate these behaviors. Excitatory projections from the medial PFC to regions such as the NAc and BLA regulate distinct aspects of cognition, motivation, and associative learning (70, 71, 72, 73). Among these pathways, the PFC→NAc circuit has emerged as a critical regulator of behaviors requiring sustained attention, integration of contextual information, and cost-benefit evaluation (70, 71, 72,74, 75, 76). Therefore, we hypothesized that disruption of inhibitory control over this projection following adolescent scPCP exposure contributes to persistent cognitive dysfunction. Consistent with this hypothesis, scPCP selectively reduced inhibitory synaptic transmission onto PFC→NAc projection neurons, whereas inhibitory signaling onto PFC→BLA neurons remained intact. Importantly, these projection-specific deficits were restored by mGlu1 activation, paralleling the ability of the mGlu1 PAM VU6004909 to improve performance in both the TUNL and novel object recognition tasks. Together, these findings suggest that restoration of inhibitory control within the PFC→NAc circuit represents one mechanism through which mGlu1 positive allosteric modulation improves cognition following developmental NMDAR hypofunction. Although the TUNL task is widely used to assess spatial working memory, the behavioral profile observed here suggests that mGlu1 potentiation may improve broader executive processes in addition to mnemonic function. Specifically, PCP impaired response accuracy during both nondelay and delay trials, whereas VU6004909 improved performance across both conditions. These findings suggest that mGlu1 receptor activation may enhance attentional processes, task engagement, and executive control, in addition to working memory, all of which likely contribute to improved performance during touchscreen-based cognitive testing. Consistent with this interpretation, hyperactivity within the NAc has been associated with impaired attentional processing (77), whereas SST-IN–mediated inhibition is thought to improve signal-to-noise ratios during cognitive tasks by suppressing inappropriate pyramidal cell activity (64). Thus, restoration of inhibitory control over PFC→NAc neurons may enhance executive performance through multiple complementary mechanisms. Although corticostriatal projection neurons are frequently classified as intratelencephalic (IT) neurons, projection target alone does not definitively distinguish IT and pyramidal tract (PT) neuronal subclasses, as considerable heterogeneity exists within these populations. Future studies incorporating additional electrophysiological features, such as hyperpolarization-activated voltage sag, together with molecular and anatomical markers, will allow a more precise classification of these projection-defined neuronal populations and determine whether mGlu1 receptor signaling differentially regulates IT and PT circuits. Likewise, although retrograde viral injections were carefully targeted to the BLA, we cannot completely exclude the possibility of limited off-target labeling of neurons projecting to adjacent structures. Nevertheless, the primary mechanistic conclusions of this study are supported by the selective and robust physiological deficits observed in PFC→NAc projection neurons, which were the focus of our subsequent mechanistic investigations. Importantly, layer V pyramidal neurons of the medial PFC project broadly to numerous additional downstream targets, including the mediodorsal thalamus, hippocampus, and other cortical and subcortical structures that also contribute to cognitive processing (78). It is therefore likely that mGlu1 receptor activation influences additional projection-defined neuronal populations beyond those examined here. Defining how mGlu1 signaling regulates these parallel corticostriatal and corticofugal circuits will be an important direction for future studies and may provide a more comprehensive understanding of the circuit mechanisms underlying the procognitive efficacy of mGlu1 positive allosteric modulators. Conclusions Collectively, these findings identify mGlu1 receptor signaling as a critical regulator of SST-IN–mediated cortical inhibition and projection-defined prefrontal circuit function following developmental NMDAR hypofunction. By restoring inhibitory transmission within cortical microcircuits and the PFC→NAc pathway, mGlu1 positive allosteric modulators improve cognitive performance across complementary behavioral assays. These findings provide mechanistic support for mGlu1 receptor potentiation as a promising therapeutic strategy for treating cognitive deficits associated with schizophrenia. Acknowledgments and Disclosures This research was supported by the National Institutes of Health (Grant Nos. MH119673, NS031373, and MH062646 [to PJC] and MH065215 [to DJL]). DJL was responsible for conceptualization. DJL, IMG, MJM, BW, MZL, and GAF were responsible for investigation. CWL contributed to resources. DJL, ESC, CMN, and PJC were responsible for supervision. DJL contributed to writing–original draft. All authors were responsible for writing–review and editing of the final manuscript. We thank the William K. Warren Foundation for endowing the Warren Center for Neuroscience Drug Discovery (WCNDD). Behavioral experiments were performed using the Vanderbilt University Medical Center Murine Neurobehavior Core and Cell Imaging Shared Resource. The illustrative graphics were created with BioRender (https://www.biorender.com). PJC, CWL, and CMN receive research support from Boehringer Ingelheim and Acadia Pharmaceuticals. CWL also receives support from Ono Pharmaceutical. PJC, CWL, and CMN are inventors on multiple patents for allosteric modulators of metabotropic glutamate receptors. All other authors report no biomedical financial interests or potential conflicts of interest. Footnotes Supplementary material cited in this article is available online at https://doi.org/10.1016/j.bpsgos.2026.100819. Contributor Information Deborah J. Luessen, Email: deborah.joyce.luessen@emory.edu. P. Jeffrey Conn, Email: jeffrey.conn@vanderbilt.edu. Supplementary Material Supplemental Methods and Figures S1–S3 Key Resources Table References Associated Data Supplementary Materials Supplemental Methods and Figures S1–S3 Key Resources Table