The novel antipsychotic cariprazine stabilizes gamma oscillations in rat hippocampal slices

Background and Purpose Gamma oscillations are fast rhythmic fluctuations of neuronal network activity ranging from 30 to 90 Hz that establish a precise temporal background for cognitive processes such as perception, sensory processing, learning, and memory. Alterations of gamma oscillations have been observed in schizophrenia and are suggested to play crucial roles in the generation of positive, negative, and cognitive symptoms of the disease. Experimental Approach In this study, we investigated the effects of the novel antipsychotic cariprazine, a D3‐preferring dopamine D3/D2 receptor partial agonist, on cholinergically induced gamma oscillations in rat hippocampal slices from treatment‐naïve and MK‐801‐treated rats, a model of acute first‐episode schizophrenia. Key Results The D3 receptor‐preferring agonist pramipexole effectively decreased the power of gamma oscillations, while the D3 receptor antagonist SB‐277011 had no effect. In treatment‐naïve animals, cariprazine did not modulate strong gamma oscillations but slightly improved the periodicity of non‐saturated gamma activity. Cariprazine showed a clear partial agonistic profile at D3 receptors at the network level by potentiating the inhibitory effects when the D3 receptor tone was low and antagonizing the effects when the tone was high. In hippocampal slices of MK‐801‐treated rats, cariprazine allowed stabilization of the aberrant increase in gamma oscillation power and potentiated resynchronization of the oscillations. Conclusion and Implications Data from this study indicate that cariprazine stabilizes pathological hippocampal gamma oscillations, presumably by its partial agonistic profile. The results demonstrate in vitro gamma oscillations as predictive biomarkers to study the effects of antipsychotics preclinically at the network level.


| INTRODUCTION
To date, schizophrenia is understood as a neurodevelopmental disease arising from complex interactions between genetic and environmental factors, inducing changes in neuronal networks and alterations in multiple neurotransmitter systems (Birnbaum & Weinberger, 2017). The pathophysiology underlying the symptoms of schizophrenia remains poorly understood. Some of the molecular changes ultimately leading to the symptoms of the disease have been partially uncovered by the beneficial effects of antipsychotics. These drugs are used primarily to decrease the positive symptoms of the disease, which are mainly Abbreviations: ACSF, artificial CSF; CI, confidence interval; Physo, physostigmine.
Maria A. Meier and Clement E. Lemercier contributed equally to this work. attributed to an excessive dopaminergic tone leading to overactivation of dopamine D 2 -like receptors (Miyamoto, Miyake, Jarskog, Fleischhacker, & Lieberman, 2012). However, much less is known regarding the molecular and network backgrounds of negative and cognitive symptoms, as very few drugs with multiple receptor profiles have been associated with alleviation of these symptoms in the clinic.
Cariprazine (United States: Vraylar ® ; Europe: Reagila ® ) is a novel antipsychotic drug that was recently approved for the treatment of schizophrenia in adults (United States and Europe) and has shown treatment benefits against risperidone in patients with predominantly negative symptoms (Németh et al., 2017). Cariprazine is a high-affinity dopamine D 3 -preferring D 3 /D 2 receptor partial agonist with additional partial agonistic action at 5-HT 1A receptors. It also exhibits high affinity for the 5-HT 2B and moderate affinity for the 5-HT 2A receptors, with an antagonistic profile at both of these receptors (Kiss et al., 2010). Cariprazine has been shown to occupy both D 2 and D 3 receptors in the brain of patients with schizophrenia, albeit with a preference for D 3 receptors, with ED 50 ratios in the threefold to sixfold range (Girgis et al., 2016). This is in contrast to currently available antipsychotic medications, such as clozapine, risperidone and olanzapine, which at antipsychotic-effective doses show high occupancy of D 2 receptors with no significant occupancy of D 3 receptors (Graff-Guerrero et al., 2009;Mizrahi et al., 2011). Therefore, studies have suggested that the superior efficacy of cariprazine against the negative symptoms of schizophrenia (Németh et al., 2017) may arise from its unique pharmacological profile, in particular, its high affinity for D 3 receptors compared to other antipsychotics (Gyertyán et al., 2011;Kiss et al., 2010). However, the link between the unique pharmacological profile and clinical effectiveness is not yet fully understood because of a lack of pharmacological data at the system level. Investigation of the modulation of gamma oscillations has been considered a useful method to identify molecular mechanisms by which cariprazine influences the symptoms of schizophrenia (Ahnaou, Huysmans, Van de Casteele, & Drinkenburg, 2017;Schulz, Heidmann, et al., 2012).
Gamma oscillations are a measure of synchronous population activity at frequencies between 30 and 90 Hz (Fries, 2009) and are known to be generated by rhythmic firing of fast-spiking parvalbumin-containing perisomatic interneurons (Cardin et al., 2009). The power of gamma oscillations increases during sensory processing, perception, attention, working memory or long-term memory representation in both animals (Lu, Jefferys, Toescu, & Vreugdenhil, 2011) and humans (Herrmann, Munk, & Engel, 2004;Rodriguez et al., 1999), indicating that gamma oscillations are a useful biomarker to investigate cognitive processes at the network level.
NMDA receptor hypofunction has been implicated in the pathophysiology of schizophrenia (Jentsch & Roth, 1999). A single dose of a non-competitive NMDA receptor antagonist, such as ketamine or phencyclidine, induces psychotic, negative and cognitive symptoms in healthy humans resembling those observed in schizophrenia (Newcomer et al., 1999). In addition, patients receiving NMDA antagonists have been reported to experience exacerbations of schizophrenia symptoms (Lahti, Weiler, Tamara Michaelidis, Parwani, & Tamminga, 2001). In rodents, NMDA receptor antagonists induce a schizophrenia-like phenotype, including hyperlocomotion, stereotypies, disruptions in pre-pulse inhibition, impairments in attention, social behaviour and cognitive deficits (Cadinu et al., 2017;Wiescholleck & Manahan-Vaughan, 2013). In these animals, a single acute dose of an NMDA receptor antagonist induced a strong increase in gamma oscillations both in the hippocampus and neocortex (Kehrer et al., 2007;Lemercier, Holman, & Gerevich, 2017). Furthermore, a previous report suggested that a single acute dose of NMDA receptor antagonists mimics acute first-episode schizophrenia, whereas chronic administration simulates the chronic symptoms of the disease with predominating negative symptoms (Jentsch & Roth, 1999).
Previous studies indicated the pivotal modulatory role of D 3 receptors in hippocampal gamma oscillations in vitro (Lemercier, Schulz, Heidmann, Kovács, & Gerevich, 2016;Schulz, Heidmann, et al., 2012). With regard to the high preference of cariprazine for D 3 receptors, we hypothesized that cariprazine modulates hippocampal gamma oscillatory activity and this effect may underlie its beneficial effects observed in schizophrenia. Therefore, to gain insight on its working mechanism at the network level, we investigated the effects of cariprazine on pharmacologically induced oscillatory activity in hippocampal slices from treatment-naïve and MK-801-treated rats.
2 | METHODS 2.1 | Animals and slice preparation Acute brain slices were prepared from 6-to 9-week-old (180-230 g) Wistar rats (n = 85) of either sex and own breed in accordance with What is already known • Gamma oscillations are disturbed in patients with schizophrenia and in animal models of the disease.

What does this study add
• The novel antipsychotic cariprazine stabilized hippocampal gamma oscillations in rats pretreated with MK-801.

| Extracellular recordings
Local field potentials were recorded from the stratum pyramidale in the CA3b area of the hippocampus with glass pipettes filled with ACSF (resistance <3 MΩ) as described earlier (Çalışkan et al., 2015;. Recordings were amplified by a custommade amplifier, low-pass filtered at 1 kHz and sampled at 5 kHz by a CED 1401 interface (Cambridge Electronic Design, Cambridge, UK).

| Data analysis and statistics
The data and statistical analysis comply with the recommendations of the British Journal of Pharmacology on experimental design and analysis in pharmacology (Curtis et al., 2018). Experiments were designed to generate groups of equal size. Slices were randomly allocated to recording chambers and randomized into drug-treated or control group before treatment. For blinding, operators and analysts were different persons. Group size selection was planned by the program G*Power3 (Faul, Erdfelder, Lang, & Buchner, 2007). Power spectra were calculated every 2 min with a 120-s window throughout the recording. Based on these power spectra, peak power, peak frequency, half bandwidth, or quality factor (Q factor) of the oscillations were determined off-line by using a custom-made script for the Spike2 software (version 7.10, Cambridge Electronic Design, Cambridge, UK). The Q factor of the oscillation (Lemercier et al., 2017) was calculated by means of the following equation: Q = f 0 /B, where f 0 is the peak frequency and B is the bandwidth at 50% of maximum peak power and was used instead of half bandwidth in experiments where oscillations desynchronized and altered their peak frequency.
The Q factor describes the relative distribution of the frequencies around the peak frequency and is therefore independent of changes in peak frequency. Oscillation parameters show a high variability and were therefore normalized in every slice to a 10-min baseline period before drug application, ACh/physostigmine (Physo) wash-out, or corresponding time in control experiments. Statistical analysis was performed with GraphPad Prism (GraphPad Software Inc., San Diego, USA). The D'Agostino-Pearson normality test was used to test the Gaussian distribution of the data. Absolute power values displayed a log-normal distribution and are therefore presented as geometric mean with its confidence interval (CI). All other data are presented as mean ± SEM. At least four animals were used in each experimental group. The group data subjected to statistical analysis have a minimum of n = 5 independent samples per group. Several slices of each brain were used for experiments but not more than one slice per hemisphere for each treatment group. No outliers were removed from the data. Statistical comparisons between drug-induced effects and time-matched control changes were performed using Student's unpaired t-test. Absolute power values were compared using the non-parametric Mann-Whitney U test. In all analyses, the significance level was set at P < .05.

| Materials
Physostigmine (Physo), MK-801, pramipexole and SB-277011 were obtained from Tocris Bioscience (Bristol, UK). ACh was purchased from Sigma-Aldrich (Taufkirchen, Germany) and cariprazine was provided by Gedeon Richter (Budapest, Hungary). Stock solutions were prepared in water, and the experimental compounds were further diluted in ACSF to reach the final concentrations. Cariprazine was applied at a concentration of 10 μM. Pharmacologically effective cariprazine plasma concentrations were 213 and 115 nM in rat and humans, respectively (Gyertyán et al., 2011;Mauri et al., 2018). Since brain concentrations were found to be min. 7.6-fold higher than the corresponding plasma concentrations in rats (Gyertyán et al., 2011), we estimated a concentration range of 1-2 μM in brain during active behaviour. To reach this steady-state concentration by diffusion into the slice held in the interface-type chamber, a factor of 5 was used.

| Nomenclature of targets and ligands
Key protein targets and ligands in this article are hyperlinked to corresponding entries in http://www.guidetopharmacology.org, the common portal for data from the IUPHAR/BPS Guide to PHARMA-COLOGY (Harding et al., 2018), and are permanently archived in the Concise Guide to PHARMACOLOGY 2019/20 (Alexander, Christopoulos et al., 2019;Alexander, Kelly et al., 2019).

| D 3 receptor activation inhibits hippocampal gamma oscillations
Bath application of ACh (10 μM) and the acetylcholinesterase inhibitor physostigmine (2 μM) induced gamma oscillations in the CA3 area of the hippocampus which reached a plateau after~90 min with a peak power of 463.6 μV 2 (95% CI [183.5, 1171]), peak frequency of 39.1 ± 0.9 Hz and a half bandwidth of 5.6 ± 1.8 Hz (n = 19). The D 3 receptor agonist pramipexole (30 μM; Mierau et al., 1995) significantly decreased the peak power of these oscillations to 38.1 ± 8.2% (P = .002; Figure 1a,b,c) and broadened the half bandwidth to 198.1 ± 19.1% (P = .002), whereas it did not significantly affect the peak frequency of the gamma oscillations.
The dopamine D 3 receptor antagonist SB-277011 (30 μM) did not affect the power, peak frequency or half bandwidth compared to control gamma oscillations in the hippocampus (Figure 1d,e).
To test whether non-saturated gamma oscillations were differentially F I G U R E 1 Activation of dopamine D 3 receptors inhibits hippocampal gamma oscillations. (a) Spectrogram shows the inhibitory effect of pramipexole (30 μM), a D 3 receptor-preferring agonist, on gamma oscillations in the CA3 area induced by ACh (10 μM) and Physo (2 μM) in hippocampal slices. Brighter colours mean higher power values. (b) Original traces (left) and power spectra (right) of gamma oscillations before and after pramipexole application. (c) Bar diagrams show the effect of pramipexole (n = 7) on peak power, peak frequency and peak bandwidth compared to time-matched parameters from control slices where only ACh and Physo were applied (n = 9). (d-e) The D 3 receptor antagonist SB-277011 (30 μM) did not modulate hippocampal gamma oscillations. (d) Original traces (left) and power spectra (right) before and after SB-277011 application. (e) Bar diagrams show the effect of SB-277011 (n = 9) on peak power, peak frequency and peak bandwidth compared to time-matched parameters from control slices (n = 9). Cont., control; Freq., frequency; Norm, normalized; Physo, physostigmine; Pram., pramipexole; PSD, power spectral density; SB, SB-277011. Data are presented as mean ± SEM. * P < .05 affected by cariprazine, we induced oscillations with lower concentrations of ACh (5 μM) and physostigmine (1 μM, Figure 2). These oscillations developed with a significantly lower peak power (43.7 μV 2 , 95% CI: 19.2 to 99.1, n = 19, P = .015) and a higher peak frequency (45.9 ± 1.1 Hz, P < .001), while the half bandwidth remained unchanged (5.9 ± 0.6 Hz), consistent with previous observations with carbachol-induced oscillations (Fisahn, Pike, Buhl, & Paulsen, 1998).
Similar to the saturated gamma oscillations, cariprazine (10 μM) did not significantly modulate the power and peak frequency of these oscillations (Figure 2d,e,f) compared to the changes with control during the same time interval. However the half bandwidth of the nonsaturated oscillations was significantly reduced to 90.7 ± 5.9% (P = .044 compared to the control change: 113.4 ± 8.2%, Figure 2d,g).

| Cariprazine shows a partial agonistic profile when modulating gamma oscillations
To evaluate how cariprazine interacts with the D 3 receptor to modulate gamma oscillations depending on the current level of D 3 receptor activation, we applied cariprazine prior to pramipexole treatment that was administered at two different concentrations. The low concentration of pramipexole (10 μM) did not significantly decrease the oscillation power, although a slight and non-significant reduction was observed (Figure 3a,c). Cariprazine, however, potentiated the effect of 10-μM pramipexole (Figure 3a,c).
On the contrary, the higher concentration of pramipexole (30 μM) decreased the power (P = .011) and increased the half bandwidth (P = .003), as well as the peak frequency (P < .001; Figure 3b-e). Cariprazine (10 μM) inhibited the effect of pramipexole on the peak power (P = .048) and peak frequency (P = .016) whereas the effect on the half bandwidth was not significant (Figure 3b-e). Thus, while cariprazine augmented the effect of the agonist at low D 3 receptor stimulation, it had an antagonistic effect when the agonist tone was high.

| Cariprazine does not affect the temporal dynamics of hippocampal gamma oscillations in nontreated animals
Neural network activity is dynamic and characterized by the appearance and disappearance of synchronization in time. Therefore, we F I G U R E 2 Cariprazine increases the periodicity of non-saturated hippocampal gamma oscillations. (a-b) Spectrograms show gamma oscillations in the CA3 area induced by non-saturating concentrations of ACh (5 μM) and Physo (1 μM) without (a) and with cariprazine (10 μM) application (b). Brighter colours mean higher power values. (c) Original traces (top) and power spectra (below) of induced gamma oscillations in control slices after gamma oscillation induction (baseline) and 60 min later (control). (d) Original traces (top) and power spectra (below) of induced gamma oscillations before (baseline) and after cariprazine (10 μM) application. The inset shows the same power spectra rescaled to demonstrate the effect of cariprazine on the half bandwidth. (e-g) Bar diagrams show the effect of cariprazine (n = 7) on peak power (e), peak frequency (f) and peak bandwidth (g) compared to time-matched parameters from control slices (n = 7). CAR, cariprazine; Cont., control; Freq., frequency; Norm, normalized; Physo, physostigmine; PSD, power spectral density. Data are presented as mean ± SEM. * P < .05 investigated whether cariprazine was capable of modulating gamma activity in the temporal domain. We induced gamma oscillations in the hippocampus, followed by interruption of the ACh/physostigmine application for 40 min after 100 min of perfusion. Following washout of ACh/physostigmine, the gamma oscillations desynchronized, as observed by a significant decrease in the power to 6.6 ± 2.1% ( Figure 4a,d), while the peak frequency increased by 15.9 ± 4.4% ( Figure 4b). Because of this shift in frequency, we expressed the spectral bandwidth as the Q factor expressing the relative distribution of the frequencies around the particular peak frequency, as described in Section 2. The Q factor significantly decreased during the wash-out of ACh/physostigmine to 27.1 ± 2.4% (Figure 4c). After repeated wash-in, gamma oscillations resynchronized: The power increased again to 69.2 ± 6.9% of the baseline, 30 min after repeated wash-in, and the Q factor increased to 61.5 ± 3.9% during the same time interval. The peak frequency of the oscillations also decreased back to 107.5 ± 1.3%. Cariprazine (10 μM) did not significantly influence the temporal dynamics of the oscillations (Figure 4a,e). With cariprazine treatment, the power reached during the 30 min resynchronization period 62.0 ± 10.1% of the baseline (Figure 4a,e), while the Q factor increased to 74.3 ± 11.8% (Figure 4c) and the peak frequency decreased to 103.8 ± 1.4% (Figure 4b).

| Cariprazine stabilizes hippocampal gamma oscillations in MK-801-treated rats
In the next series of experiments, we investigated whether cariprazine affects disturbed gamma oscillations. To accomplish this, the effects of cariprazine were studied in hippocampal slices of rats that received a single dose of MK-801, a model of first-episode schizophrenia (Manahan-Vaughan et al., 2008;Wiescholleck & Manahan-Vaughan, 2013). In line with previous results (Lemercier et al., 2017), the hippocampal gamma oscillations induced by bath application of ACh (10 μM) and physostigmine (2 μM) had a significantly higher peak power after 90 min of induction compared to saline-treated animals (P < .001, Figure 5a,b,d) and a broader half bandwidth (MK-801: 5.19 ± 0.64 Hz, P = .033, compared to saline-treated animals: 3.55 ± 0.37 Hz; not shown); however, the Q factor of the oscillations did not change (MK-801: 9.9 ± 1.0, P = .103, compared to saline F I G U R E 3 Cariprazine demonstrates a partial agonistic profile when modulating gamma oscillations.  (Figure 5f).

| Cariprazine accelerates the temporal dynamics of gamma oscillations in the hippocampus of MK-801-treated animals
Wash-out of ACh/physostigmine gradually desynchronized gamma oscillations in MK-801-treated rats (Figure a-d). The power of the oscillations decreased to 8.5 ± 1.1% after 40 min. During the same time, peak frequency showed a non-significant trend to increase and the Q factor significantly decreased to 34.5 ± 2.7% (Figure 6c).
Repeated wash-in of ACh and physostigmine resynchronized gamma oscillations by increasing the power to 65.8 ± 11.9% and the Q factor to 66.5 ± 6.1% (Figure 6a,c). The peak frequency further increased to 109.1 ± 1.9% of the baseline (Figure 6b,g). Cariprazine did not influence the dynamics of the peak power (Figure 6a,f). The oscillations achieved in the presence of cariprazine 30 min after re-application of ACh/physostigmine 69.9 ± 15.8% of their initial power. Similarly, the peak frequency further increased and was comparable to the changes with control (Figure 6b,g). However, cariprazine clearly and significantly accelerated the increase in Q factor after repeated administration of ACh/physostigmine reaching 97.9 ± 4.8% of the original baseline (P = .006, Figure 6c (Dickinson, Bruyns-Haylett, Jones, & Milne, 2015;Herrmann & Demiralp, 2005;Uhlhaas & Singer, 2015). Our understanding of the function and role of dopamine D 3 receptors in schizophrenia pathophysiology is rather limited. D 3 receptors have a high affinity for dopamine (Sokoloff, Giros, Martres, Bouthenet, & Schwartz, 1990) and are suggested to be activated by volume transmission during the slower tonic dopamine release from dopaminergic fibres (Grace, 2016), and control the amplitude of the fast, phasic dopamine release during behaviourally salient stimuli (Sokoloff and Le Foll, 2017). In schizophrenia, this fine control of the phasic dopamine release amplitude is disturbed, resulting in a state of aberrant salience (Kapur, The normalizing effect of cariprazine at the level of neural networks, which was apparent at stable oscillations, could also be observed in pathological hippocampal networks. MK-801 is considered as one of the most reliable models to induce a disease-like phenotype in animals, including the negative and cognitive symptoms of schizophrenia (Cadinu et al., 2017;Meltzer et al., 2013). Hippocampal gamma oscillations in these animal models were found to be abnormally increased both in vivo (Kittelberger, Hur, Sazegar, Keshavan, & Kocsis, 2012) and in vitro (Kehrer et al., 2007;Lemercier et al., 2017).
These electrophysiological abnormalities of gamma oscillations were suggested to underlie the symptoms of the disease (Gonzalez-Burgos et al., 2015;Uhlhaas & Singer, 2010) and have been shown to be modulated by at least some of the available antipsychotics (Hudson, Rind, O'Brien, & Jones, 2016;Schulz, Heidmann, et al., 2012). In our in vitro experiments, following MK-801 pretreatment, the oscillations did not reach a plateau and continuously increased until the end of the measurement. Application of cariprazine, acting as a partial agonist at the D 3 receptors, effectively stabilized the unbridled oscillations and prevented the network from the deleterious effects of MK-801 pretreatment (Figure 5c). This balancing activity could be one important mechanism by which cariprazine reaches its beneficial behavioural effects during positive and negative symptoms.
The temporal dynamics of gamma oscillatory activity in slices can be considered a plastic capability of the network. The second induction of gamma oscillations in the same network occurs significantly faster indicating that plastic changes had happened during the first induction and the circuitry remembers earlier oscillations (Zarnadze et al., 2016). This paradigm might be a physiological model of synaptic plasticity at the network level rather than selective electrical stimulation of one of the inputs at a chosen frequency. The speed of the resynchronization of gamma oscillations thus indicates the extent of activity-dependent network modification in the hippocampus within a given time. Acceleration of this resynchronization may therefore indicate a dynamic network reorganization, which may be of benefit to cognitive functions (Reinhart & Woodman, 2014). In earlier observations, antagonism at D 3 receptors was found to speed up the second synchronization (Lemercier et al., 2016), suggesting that D 3 receptors have negative effects on neuronal plasticity at the network level and antagonism at these receptors may support plasticity within the network. These data may provide one possible explanation as to why D 3 receptor knockout animals perform better in cognitive tests and why F I G U R E 6 Cariprazine accelerates the temporal dynamics of gamma oscillations in the hippocampus of MK-801-treated animals. (a-c) Cariprazine (n = 8) did not affect the dynamics of peak power (a) and peak frequency (b) but speeded up the resynchronization of the Q factor (c) compared to control (n = 21). Grey area indicates the presence of ACh/Physo in the slice. (d) Original traces (left) and power spectra (right) of gamma oscillations 40 min after wash-out of ACh/Physo (140 min after the first application) and 30 min after the repeated wash-in (170 min after the first application). (e) Original traces (left) and power spectra (right) of gamma oscillations in the presence of cariprazine 40 min after wash-out of ACh/Physo (140 min after the first application) and 30 min after the repeated wash-in (170 min after the first application). (f-h) Bar diagrams show the effect of cariprazine on the peak power (f), peak frequency (g) and peak bandwidth (h) of resynchronized gamma oscillations 30 min after re-application of ACh/Physo (170 min) compared to time-matched parameters from control slices. CAR, cariprazine; Physo; physostigmine, Norm, normalized; Q factor, quality factor. Data are presented as mean ± SEM. * P < .05 D 3 receptor antagonists have pro-cognitive effects in animal models (Nakajima et al., 2013). The partial agonist cariprazine did not alter the dynamics in control animals but significantly accelerated the resynchronization in MK-801-treated animals as measured by means of the Q factor. The Q factor (as described in Section 2) describes the distribution of the frequencies around the peak and herewith the periodicity of the oscillations (Lemercier et al., 2017). After wash-out of ACh, the gamma oscillations desynchronized, as evident from the decreasing power and Q factor values. The resynchronization occurred significantly faster in the presence of cariprazine but only in MK-801-treated rats. Thus, these results suggest that cariprazine may have no effect on cognitive functions in healthy animals but could improve cognitive symptoms in schizophrenia models. In line with this, cariprazine has been shown to be effective against these domains in animal models (Barnes et al., 2018;Neill et al., 2016;Watson et al., 2016;Zimnisky et al., 2013) and in patients with schizophrenia (Daniel et al., 2017). D 3 receptors are expressed in the rat and human cortex (Bouthenet et al., 1991;Khan et al., 1998;Richtand, Kelsoe, Segal, & Kuczenski, 1995). In the hippocampus, the expression of the D 3 receptor mRNA was found in the pyramidal cell layer of the cornu ammonis (Bouthenet et al., 1991) and the receptor protein on dendrites in both the stratum oriens and radiatum (Khan et al., 1998), suggesting the presence of D 3 receptors on pyramidal cell dendrites.
Although the expression of the dopamine D 3 receptor is low in these regions, it seems comparable to that of D 2 receptors (Khan et al., 1998). Additionally, the D 3 receptor has the highest affinity for dopamine amongst all dopamine receptor subtypes, particularly when it is uncoupled from G proteins (Sokoloff et al., 1990;Van Tol et al., 1991). There are several possible cellular mechanisms by which D 3 receptors on pyramidal cells can modulate local neuronal oscillations.
In prefrontal pyramidal cells and auditory brainstem interneurons, D 3 receptor activation was shown to decrease the excitability by inhibition of calcium influx via low-voltage-activated Cav3.2 calcium channels localized to the axon initial segment (Bender, Ford, & Trussell, 2010;Clarkson, Liptak, Gee, Sohal, & Bender, 2017). D 3 receptors were also found to reduce inhibitory synaptic inputs in CA1 pyramidal cells by inducing endocytosis of GABA A receptors (Swant, Stramiello, & Wagner, 2008). Since synchronous perisomatic inhibition by fast-spiking interneurons is thought to be responsible for the generation of the gamma oscillations, reduction of the synchronizing currents by D 3 receptors could explain the gammainhibiting effects of these receptors.
Although our experiments with the D 3 receptor agonist pramipexole and cariprazine indicated that cariprazine modulates gamma oscillations by acting on D 3 receptors, antipsychotics often display a wide pharmacological profile with effective interactions on multiple receptors. In case of cariprazine, it has been reported that it possesses also partial agonistic and antagonistic actions at 5-HT 1A , 5-HT 2A and 5-HT 2B receptors, respectively (Kiss et al., 2010). Among them, the 5-HT 1A receptors (Krause & Jia, 2005) were found to modulate gamma oscillations in the hippocampus suggesting that besides the D 3 receptors, these receptors might also be involved in the mechanisms of action of cariprazine. Further investigations are needed to fully characterize the contribution of other receptors targeted by cariprazine in the modulation of gamma oscillations.
In summary, our results showed a dopamine D 3 receptor partial agonistic profile for cariprazine at the network level by studying oscillatory activities. Additionally, cariprazine had a clear stabilizing effect on pathological hippocampal gamma oscillations in an NMDA hypofunction model of schizophrenia, which can be explained by its partial agonistic profile. Considering that perturba-

CONFLICT OF INTEREST
The study was supported by a grant from Allergan Plc and Gedeon Richter Plc. Allergan Plc and Gedeon Richter PLc were involved in the study design, analysis, interpretation of data, the decision to present results, and writing of manuscript. Charité -Universitätsmedizin Berlin was involved in the study design, all experimental parts, report and discussion of the results, and writing the manuscript. B.L. and B.K. are

DATA AVAILABILITY STATEMENT
Data reported in this manuscript are available within the article. Additional data may be requested at http://www.allerganclinicaltrials.com/ PatientDataRequest.htm.