Neural substrates of cognitive impairment in a NMDAR hypofunction mouse model of schizophrenia and partial rescue by risperidone

Abstract

N-methyl D-aspartate receptor (NMDAR) hypofunction is a pathophysiological mechanism relevant for schizophrenia. Acute administration of the NMDAR antagonist phencyclidine (PCP) induces psychosis in patients and animals while subchronic PCP (sPCP) produces cognitive dysfunction for weeks. We investigated the neural correlates of memory and auditory impairments in mice treated with sPCP and the rescuing abilities of the atypical antipsychotic drug risperidone administered daily for two weeks. We recorded neural activities in the medial prefrontal cortex (mPFC) and the dorsal hippocampus (dHPC) during memory acquisition, short-term, and long-term memory in the novel object recognition test and during auditory processing and mismatch negativity (MMN) and examined the effects of sPCP and sPCP followed by risperidone. We found that the information about the familiar object and its short-term storage were associated with mPFC→dHPC high gamma connectivity (phase slope index) whereas long-term memory retrieval depended on dHPC→mPFC theta connectivity. sPCP impaired short-term and long-term memories, which were associated with increased theta power in the mPFC, decreased gamma power and theta-gamma coupling in the dHPC, and disrupted mPFC-dHPC connectivity. Risperidone rescued the memory deficits and partly restored hippocampal desynchronization but did not ameliorate mPFC and circuit connectivity alterations. sPCP also impaired auditory processing and its neural correlates (evoked potentials and MMN) in the mPFC, which were also partly rescued by risperidone. Our study suggests that the mPFC and the dHPC disconnect during NMDAR hypofunction, possibly underlying cognitive impairment in schizophrenia, and that risperidone targets this circuit to ameliorate cognitive abilities in patients.

Keywords: atypical antipsychotic drugs; auditory evoked potentials; hippocampus; neural synchrony; novel object recognition; phencyclidine; prefrontal cortex; theta and gamma oscillations.

FIGURE 1

Experimental design and main behavioral results. (A) Experimental protocol and behavioral tests used. S indicates the date of surgery to implant recording electrodes. (B) The novel object recognition test (NOR) in detail. Right and left buttons on a joystick were pressed for the duration of each visit to timestamp the information into the recording files. Representative examples of histological verification of electrode locations within the prelimbic PFC and the CA1 area of the HPC. We recorded neural activities during the surgical implantations to increase the chances of implanting the electrodes within the target areas. At the end of the experiment, before euthanasia, a gentle current was passed through the electrodes to mark the recording areas (arrows). Also shown is a summary table that illustrates the different pharmacological groups of the study and the number of mice used in each group. (C) Discrimination indices (DIs) during the 3-min (STM) and the 24 h (LTM) tests. DIs decreased significantly after sPCP and recovered after chronic risperidone (STM, LTM: F_(2,6) = 16.64, 15.15, p = 0.004, <0.0005; one-way repeated measures ANOVA, n = 9 mice; the results including all the mice are shown in Supplementary Figure 1). (D) Number and mean duration of visits to familiar and novel objects during the STM and LTM tests. After sPCP, mice visited both objects evenly by increasing the number of visits to familiar objects, reflecting poor recognition memory (F_(1,13) = 15.12, 5.2, p = 0.002, 0.037). (E) The duration of the visits decreased within a session as the mice became less interested in the objects ([baseline STM, LTM]: F_(2,26) = 3.52, 3.56, p = 0.005, 0.0006; one-way repeated measures ANOVA with novel and familiar objects combined). A sharp decrease in duration occurred during the first 5 visits to both objects in the two tests, and this was more pronounced for novel objects (first 5 visits to novel objects vs. first 5 visits to familiar objects; [STM, LTM]: F_(2,26) = 36.03, 11.57, p < 0.0005, 0.002). These behaviors were not overtly disrupted by sPCP or risperidone.

FIGURE 2

sPCP disrupted theta and gamma synchronization in mPFC-dHPC circuits during quiet wakefulness. Some of these alterations were ameliorated by risperidone. (A) sPCP increased high gamma power in the mPFC (baseline vs. sPCP: p = 0.047, paired t-test; n = 21 mice) and decreased it in the dHPC (p = 0.004). Risperidone rescued aberrant gamma power in the mPFC (F_(2,16) = 0.11, p = 0.9; one-way repeated measures ANOVA; no post-hoc differences between BAS and sPCP-RIS), but not in the dHPC (F (_2,14) = 6.73, p = 0.01). (B) sPCP weakened local and inter-regional theta-gamma coupling (6–10 Hz with 40–60 Hz; l-PAC; p = 0.008; ir-PAC PFC_phase-HPC_amp; p = 0.001; one-way repeated measures ANOVA) that were partially restored by risperidone (F_(2,14) = 0.23, p = 0.125; no post-hoc differences between BAS and sPCP-RIS). (C) sPCP promoted the spiking activity of neuron populations (multi-unit activity or MUA) in the mPFC that was reduced by risperidone (F_(2,6) = 4.58, p = 0.047). (D) sPCP disrupted the coupling of spikes to theta oscillations in the dHPC that was restored by risperidone (F_(2,6) = 19.4, p = 0.002). Risperidone also boosted spike-delta coupling in the dHPC (3–6 Hz, F_(2,6) = 5.39, p = 0.046). Spike-LFP coupling was estimated via the pairwise phase consistency method (PPC). (E) sPCP did not significantly affect the directionality of theta and high gamma signals within the circuit, but some tendencies were observed in the high gamma band. Circuit directionality was estimated via the phase slope index (PSI). (F) Proposed neural mechanism for the effects of sPCP and risperidone on mPFC-dHPC circuits during quiet alertness. In red, changes produced by sPCP, in orange sPCP-induced deviations ameliorated by risperidone. See also Supplementary Figure 3.

FIGURE 3

Neural substrates of memory acquisition and effects of sPCP and risperidone. (A) Power spectra showing the effects of sPCP and risperidone in the mPFC and the dHPC during the familiarization phase. Vertical dashed lines mark the first second upon initiation of the visits. (B) Prefrontal theta power increased in healthy animals during the late visits to the objects. sPCP increased non-specifically theta power in the mPFC (baseline vs. sPCP: F_(1,10) = 7.46, p = 0.021; mixed ANOVA with time of visits (early vs. late) and treatment (baseline vs. sPCP) as factors) and reduced theta and high gamma power in the HPC (baseline vs. sPCP: F_(1,12) = 4.81, 9.59, p = 0.053, 0.011), disrupting the normal neural dynamics of memory acquisition. Risperidone did not rescue these power changes. (C,D) Intrinsic dHPC theta-gamma coupling (7–10 Hz with 60–80 Hz) did not change with memory acquisition. sPCP disrupted theta-gamma coordination that was partially rescued by risperidone ([l-PAC] F_(1,12) = 37.68, p < 0.0005; mixed ANOVAs as above; differences with sPCP-SAL controls: F_(1,22) = 6.31, p = 0.02; mixed ANOVA with time of visits (early vs. late) and treatment (SAL vs. sPCP) as factors). Similar results were obtained for inter-regional theta-gamma coupling ([ir-PAC] F_(1,12) = 41.09, p < 0.0005; data not shown; differences with sPCP-SAL controls: F_(1,22) = 4.62, p = 0.024; mixed ANOVAs as above). (E) In healthy mice, mPFC→dHPC high gamma signals tended to occur during the late visits to the objects (PSI vs. shuffle, p = 0.069; paired t-student; marked with an arrow). sPCP disrupted these flows of information (baseline vs. sPCP, p = 0.096; PSI vs. shuffle, p = 0.49) that were not restored by risperidone (PSI vs. shuffle, p = 0.38). (F) Proposed neural mechanism for memory acquisition and effects of sPCP and risperidone. In red, changes produced by sPCP, in orange sPCP-induced deviations ameliorated by risperidone.

FIGURE 4

Neural substrates of short-term memory and effects of sPCP and risperidone. (A) Power spectra in the mPFC and the dHPC during the 3-minute memory test and effects of sPCP and risperidone. Vertical dashed lines mark the first second upon initiation of the visits. (B) Theta and high gamma oscillations were similar during the visits to familiar and novel objects in both regions. sPCP increased theta power in the mPFC non-specifically interfering with the normal theta dynamics (F_(1,7) = 8.26 p = 0.024; mixed ANOVA with object (familiar vs. novel) and treatment (baseline vs. sPCP) as factors). sPCP also reduced dHPC high gamma power (F_(1,12) = 8.37, p = 0.023; differences with sPCP-SAL controls: F_(1,20) = 6.44, p = 0.022; mixed ANOVA with object (familiar vs. novel) and treatment (SAL vs. sPCP) as factors) that were partially rescued by risperidone. (C,D) Intrinsic hippocampal theta-gamma coupling (7–10 Hz with 60–80 Hz) was similar during visits to familiar and novel objects. sPCP disrupted theta-gamma coordination that was partially rescued by risperidone ([l-PAC] F_(1,12) = 10.36, p < 0.009; differences with sPCP-SAL controls; F_(1,22) = 19.13, p = < 0.0005; mixed ANOVAs as above). Similar results were obtained for inter-regional theta-gamma coupling ([ir-PAC] F_(1,12) = 16.38, p = 0.002, data not shown; differences with sPCP-SAL controls: F_(1,22) = 17.5, p < 0.0005; mixed ANOVAs as above). (E) In healthy mice, mPFC→dHPC high gamma signals were detected during the visits to the familiar objects (PSI vs. shuffle, p = 0.004; paired t-student). sPCP disrupted this flow of information (baseline vs. sPCP, p = 0.008; PSI vs. shuffle, p = 0.19) that was not restored by risperidone (PSI vs. shuffle, p = 0.24). (F) mPFC→dHPC PSI at high gamma frequencies during the visits to familiar objects correlated strongly with discrimination indices during baseline (R = 0.7, p = 0.026), but not after sPCP (R = –0.28, p = 0.43) or risperidone (R = –0.39, p = 0.38). (G) Proposed neural mechanism for STM and effects of sPCP and risperidone. In red, changes produced by sPCP, in orange sPCP-induced deviations ameliorated by risperidone.

FIGURE 5

Neural substrates of long-term memory and effects of sPCP and risperidone. (A) Power spectra in the mPFC and the dHPC during the 24h memory test and effects of sPCP and risperidone. Vertical dashed lines mark the first second upon initiation of the visits. (B) Power in the mPFC and the dHPC was similar during visits to familiar and novel objects. sPCP reduced high gamma power in the dHPC non-specifically (F_(1,12) = 15.33, p = 0.006; mixed ANOVA with object (familiar vs. novel) and treatment (baseline vs. sPCP) as factors; differences with SAL controls, F_(1,20) = 5.28, p = 0.021; mixed ANOVA with object (familiar vs. novel) and treatment (SAL vs. sPCP) as factors). As above, risperidone partially rescued hippocampal gamma power. (C,D) Intrinsic hippocampal theta-gamma coupling (7–10 Hz with 60–80 Hz) was similar during visits to familiar and novel objects. sPCP disrupted theta-gamma coordination that was partially rescued by risperidone ([l-PAC] F_(1,12) = 15.96, p < 0.003; differences with sPCP-SAL controls: F_(1,22) = 6.31, p = 0.021; mixed ANOVAs as above). Similar results were obtained for inter-regional theta-gamma coupling ([ir-PAC] F_(1,12) = 33.19, p < 0.0005; differences with sPCP-SAL controls: F_(1,22) = 5.97, p = 0.02; data not shown). (E) In healthy mice, dHPC→mPFC theta signals were detected during the visits to the familiar objects (PSI vs. shuffle, p = 0.031; paired t-student). sPCP disrupted this flow of information (baseline vs. sPCP, p = 0.024; PSI vs. shuffle, p = 0.94) that was not rescued by risperidone (PSI vs shuffle, p = 0.64). (F) dHPC→mPFC PSI at theta frequencies during the visits to familiar objects correlated strongly with discrimination indices during baseline (R = –0.72, p = 0.01), but not after sPCP (R = –0.42, p = 0.17) or risperidone (R = –0.47, p = 0.24). (G) Proposed neural mechanism for LTM and effects of sPCP and risperidone. In red, changes produced by sPCP, in orange sPCP-induced deviations ameliorated by risperidone.

FIGURE 6

Neural substrates of auditory processing in the prelimbic cortex and effects of sPCP and risperidone. (A) AEP protocol. (B) Ratio of AEP responses detected in the PFC after the auditory stimuli. The ratio decreased in sPCP-treated animals ([baseline vs. sPCP, sPCP vs. saline]: p = 0.024, 0.024, paired and unpaired t-test, respectively) and was partially recovered by risperidone. (C) Mean AEP and corresponding spiking activity (multi-unit firing rates) at three different timescales. P2 (40-70 ms) was associated with less spiking activity after sPCP ([baseline vs. sPCP, sPCP vs. saline as above]: p = 0.013, 0.068; paired and unpaired t-tests, respectively), whereas P3 (200–300 ms) was reduced in amplitude and spiking activity ([baseline vs. sPCP, sPCP vs. saline as above]: amplitude: p = 0.015, 0.055; firing rate: p = 0.024, 0.06). Risperidone increased the spiking activity associated within P2 with respect to sPCP (F_(2,10) = 2.49, p = 0.076; one-way repeated measures ANOVA) and augmented the amplitude and spiking activity associated with P3 component ([amplitude, firing rate]: p = 0.065, 0.097; paired and unpaired t-tests, respectively). (D) Protocol used for the oddball paradigm to assess mismatch negativity (MMN). (E) MMN was detected during the presentation of the 6-8 KHz target-frequent combination (left) but not the 8–6 kHz combination (right). Shown is the subtraction of target-frequent responses. MMN was absent in the sPCP-treated group (baseline vs. sPCP area under the curve; p = 0.021, paired t-test) and not restored by risperidone. Vertical dashed lines mark the start and end of tone presentation.