The Effects of Antipsychotics on the Synaptic Plasticity Gene Homer1a Depend on a Combination of Their Receptor Profile, Dose, Duration of Treatment, and Brain Regions Targeted

Background: Antipsychotic agents modulate key molecules of the postsynaptic density (PSD), including the Homer1a gene, implicated in dendritic spine architecture. How the antipsychotic receptor profile, dose, and duration of administration may influence synaptic plasticity and the Homer1a pattern of expression is yet to be determined. Methods: In situ hybridization for Homer1a was performed on rat tissue sections from cortical and striatal regions of interest (ROI) after acute or chronic administration of three antipsychotics with divergent receptor profile: Haloperidol, asenapine, and olanzapine. Univariate and multivariate analyses of the effects of topography, treatment, dose, and duration of antipsychotic administration were performed. Results: All acute treatment regimens were found to induce a consistently higher expression of Homer1a compared to chronic ones. Haloperidol increased Homer1a expression compared to olanzapine in striatum at the acute time-point. A dose effect was also observed for acute administration of haloperidol. Conclusions: Biological effects of antipsychotics on Homer1a varied strongly depending on the combination of their receptor profile, dose, duration of administration, and throughout the different brain regions. These molecular data may have translational valence and may reflect behavioral sensitization/tolerance phenomena observed with prolonged antipsychotics.


Introduction
Antipsychotics are considered the gold-standard treatment for psychosis [1]. Despite the well-known involvement of dopamine D2 receptor (D2R) occupancy in the antipsychotic mechanism of action [2,3], and the demonstrated effects on synaptic plasticity and metaplasticity [4], there is still a dearth of information on some basic aspects regarding the antipsychotic impact on the synapse. How antipsychotic dose, receptor profile, duration of administration, and brain region targeted by each agent may influence synaptic plasticity is yet to be determined. A better understanding of these variables in antipsychotic effects may also influence our knowledge of the mechanism responsible for the loss or reduction of efficacy over time of antipsychotics treatment.
Notably, antipsychotic agents have been found to modulate the expression of glutamatergic genes of the postsynaptic density (PSD), an electron-dense thickening at the surface of the glutamatergic postsynaptic membrane [5][6][7], actively involved in behavioral disorders' molecular pathophysiology [8,9]. Among the most replicated findings, antipsychotics have been found to modulate the immediate-early gene Homer1a [10][11][12], whose expression has been demonstrated to sculpt the dendritic spine and overall synaptic plasticity [5,13].
Moreover, Homer has been per se considered a candidate gene in schizophrenia and other psychiatric diseases [14][15][16][17][18] and has been demonstrated to be involved in animal models of behavioral disorders, including schizophrenia [19][20][21][22][23][24][25]. The Homer1a gene codes for the truncated isoform of the long Homer1 protein. Once expressed, Homer1a competes with the long isoforms and disassembles protein clusters formed by the long Homer isoforms and multiple PSD targets, including PSD-95, Shank, and mGluR5 [26]. As a result of Homer1a induction, the synaptic microdomain in which Homer proteins operate undergoes multiple functional and architectural rearrangements, including changes in intracellular Ca ++ dynamics [27], shifts in surface channel functioning [28], and modifications in synaptic ultrastructure [29]. It modulates the interaction between metabotropic glutamatergic receptors (mGluRs) and intracellular effectors [30], thereby modulating the strength and direction of intracellular signaling [31]. Changes in Homer1a levels within the PSD may, in turn, affect the reciprocal interaction between NMDA and mGluRs, and regulate surface channel activity [28,32]. Given the complexity of Homer1a biological action, which is intrinsic to synaptic plasticity, the antipsychotic liability of modulating the expression of this gene appears relevant to these agents' molecular actions.
In previous reports, indirect and only partially informative evaluations of outcomes of the divergent duration of treatments (i.e., acute vs. chronic effects) on Homer1a expression were provided [33][34][35]. A direct and systematic evaluation of the combined effects of the dose and receptor profile has not been carried out yet. Based on all these considerations, we wanted to test the hypothesis that the putative effects of antipsychotics on the Homer1a gene expression dose (or antipsychotic receptor profile) may be differentially mediated by the dose, receptor profile, duration of treatment, and regions of interest at which the gene expression was evaluated.
With the scope of tackling these issues, univariate analyses were run to assess independent time, dose, or receptor profile effects within each region of interest (ROI). Furthermore, as a secondary goal, multivariate analyses were run to study the patterns of Homer1a mRNA expression throughout multiple cortical and striatal regions of interest after acute (i.e., single administration) or chronic (21-day administration period) treatment regimens.
The dose effect was investigated by incremental doses of the high multi-receptor second-generation antipsychotic asenapine, characterized by the breadth of activities at multiple serotonergic receptors [36,37]. To investigate the dose effect by a first-generation narrower receptor profile agent, in a separate set of analyses, we compared gene expression by three different doses of the relatively selective D2R blocker haloperidol. In the third set of analysis, we investigated the receptor profile effect on gene expression by comparing asenapine, haloperidol, and olanzapine at behaviorally and neurochemically comparable doses. Olanzapine was chosen as a reference second-generation antipsychotic, with a receptor profile and biological effects that may be considered sharply divergent from asenapine [38].
Interestingly, these three compounds exhibit significantly different D2/D1 ratios (25:1 for haloperidol, 3:1 for olanzapine, and approximately 1:1 for asenapine) [38][39][40] and this feature may be relevant in order to understand the contribution of the dopaminergic antagonism in triggering the transcription of key molecules of synaptic architecture, such as Homer1a. In fact, as well as D2R-blockers, even selective D1R antagonists have been found to influence striatal Homer1a expression considerably [41]. Moreover, asenapine acts as a more potent antagonist in particular at 5-HT1A and 5-HT2 receptors (pKB = 7.4 for 5-HT1A; pKB = 9 for 5-HT2), in comparison to olanzapine (pKB < 5 for both 5-HT1A and 5-HT2) [38]. Testing an antipsychotic agent with such remarkable serotonergic action was even more appealing for their potential to induce cortical Homer1a expression [35]. Therefore, although olanzapine and asenapine both belong to the class of atypical agents and show a complex, partially overlapping, multireceptor binding profile, they may widely differ in terms of glutamatergic activity and capability to affect synaptic plasticity.

Results
A three-way mixed ANOVA was run to understand the effects of ROI, treatment, and time on Homer1a mRNA expression. Data are the mean fold changes of the relative disintegration per minutes (DPM) compared to the vehicle ± standard error mean. Shapiro-Wilk's test confirmed that the relative DPM values were normally distributed (Tables S1-S6). Levene's test for equality of variances showed homogeneity of variances for the relative DPM throughout all ROIs. Outputs using raw (i.e., non-relativized) data are reported in Tables S7-S10.

Results
A three-way mixed ANOVA was run to understand the effects of ROI, treatment, and time on Homer1a mRNA expression. Data are the mean fold changes of the relative disintegration per minutes (DPM) compared to the vehicle ± standard error mean. Shapiro-Wilk's test confirmed that the relative DPM values were normally distributed (Tables S1-S6). Levene's test for equality of variances showed homogeneity of variances for the relative DPM throughout all ROIs. Outputs using raw (i.e., nonrelativized) data are reported in Tables S7-S10.

Striatum
No significant effect of asenapine doses was found in any striatal ROI at either acute or chronic time point (Table 1, Figure 2).
A statistically significant effect of time was found for all doses in distinct striatal ROIs (Table 1, Figure 2). More specifically, mean Homer1a expression was higher after acute compared to chronic administration of: ASE0.05 in DM, DL, and VM; ASE0.1 in all striatal ROIs with the exception of SAb; and ASE0.3 in DL, VM, VL, and CAb (Table 1, Figure 2).   A statistically significant effect of time was found for all doses in distinct striatal ROIs (Table 1, Figure 2). More specifically, mean Homer1a expression was higher after acute compared to chronic administration of: ASE0.05 in DM, DL, and VM; ASE0.1 in all striatal ROIs with the exception of SAb; and ASE0.3 in DL, VM, VL, and CAb (Table 1, Figure 2). Table 1. Dose and time-independent effects on asenapine-mediated Homer1a expression as assessed by univariate analysis in each region of interest. On the right side of the table, the "dose effect" was analyzed separately for acute and chronic administration regimens, in order to determine if asenapine-mediated Homer1a expression was different for groups exposed to different asenapine doses (i.e., ASE0.05, ASE0.1, and ASE0.3). On the left side of the table, the "time effect" was analyzed separately for each asenapine dose, in order to determine if asenapine-mediated Homer1a expression was different for groups with different durations of treatment (i.e., acute and chronic). Significant effects are given in bold. ASE0.05 = asenapine at 0.05 mg/kg; ASE0.1 = asenapine at 0.1 mg/kg; ASE0.3 = asenapine at 0.3 mg/kg. ns = not significant. At the multivariate analysis, we found a statistically significant interaction effect between time, asenapine dose, and ROI (Table 4). Significant ROI*time and ROI*asenapine dose interactions were subsequently found, while no asenapine dose*time interaction was observed (Table 4).

Cortex
At both the acute and chronic time-points, no significant asenapine dose effect was found in cortical ROIs (Table 1, Figure 2).
A statistically significant effect of time was found in multiple cortical ROIs (Table 1). Homer1a expression was significantly higher after acute compared to chronic administration of ASE0.1 in MC, SS, and IC and of ASE0.3 in IC (Table 1, Figure 2).
At the multivariate analysis, no significant three-way interaction between ROI, asenapine dose, and time was found (Table 4). A significant ROI*time two-way interaction was found, while no significant asenapine dose*ROI and asenapine dose*time two-way interactions were observed (Table 4).

Paradigm 2
In this paradigm, the three categorical independent variables were: Treatment with incremental haloperidol doses (thereafter haloperidol doses); time; and ROIs. The haloperidol doses variable comprised three categories: 0.25 mg/kg treatment group; 0.5 mg/kg treatment group; and 0.8 mg/kg treatment group. The time variable comprised two categories: Acute administration and chronic administration. The ROIs variable comprised five categories in the cortex (ACC; MAC; MC; SS; IC) and six categories in the striatum (DM; DL; VM; VL; CAb; SAb).

Striatum
A significant haloperidol dose effect was observed in the DM and the SAb after acute administration (Table 2, Figure 3). In the DM, Homer1a expression by HAL0.5 was significantly higher than expression by HAL0.8 (Bonferroni adjusted post-hoc test: p = 0.05; Figure 3). In the SAb, HAL0.5 administrations triggered significantly higher expression of Homer1a than HAL0.25 (Bonferroni adjusted post-hoc test: p = 0.032; Figure 3). At the chronic time-point, no significant dose effect was found in any ROI (Table 2, Figure 3).   A significant time effect was found in all striatal ROIs (Table 2). Specifically, in all ROIs, acute expression of Homer1a was significantly higher than chronic expression by HAL0.25, HAL0.5, and HAL0.8, with the only exception of the SAb for the HAL0.25 dose (Table 2, Figure 3).
At the multivariate analysis, we did not find a significant three-way interaction effect between ROI, haloperidol dose, and time (Table 4). A significant ROI*time two-way interaction was found, while no significant haloperidol dose*ROI and haloperidol dose*time two-way interactions were observed (Table 4).

Cortex
A significant haloperidol dose effect was found in the ACC at the acute time-point (Table 2). In this region, Homer1a expression by HAL0.5 was significantly higher than expression by HAL0.25 (Bonferroni adjusted post-hoc test: p = 0.027; Figure 3). At the chronic time-point, no significant haloperidol dose effect was found (Table 2, Figure 3).
A significant time effect was found in the MC, SS, and IC regions ( Table 2). In all these regions, Homer1a expression by HAL0.8 was significantly higher after the acute than the chronic administration regimen (Figure 3). In the MC, Homer1a expression by acute HAL0.5 was significantly higher than by chronic HAL0.5 administration (Table 2, Figure 3).
At the multivariate analysis, no significant three-way interaction between ROI, haloperidol dose, and time was observed (Table 4). A significant ROI*time two-way interaction was found, while no significant haloperidol dose*ROI and haloperidol dose*time two-way interactions were observed (Table 4).

Paradigm 3
In this paradigm, the three categorical independent variables were: Treatment with antipsychotics with different receptor profiles (thereafter treatment); time; and ROIs. The treatment variable comprised three categories: Olanzapine 2.5 mg/kg treatment group; haloperidol 0.5 mg/kg treatment group; and asenapine 0.1 mg/kg treatment group. The time variable comprised two categories: Acute administration and chronic administration. The ROIs variable comprised five categories in the cortex (ACC; MAC; MC; SS; IC) and six categories in the striatum (DM; DL; VM; VL; CAb; SAb).
The choice of antipsychotic doses to be compared was based on multiple considerations. These doses were reported to elicit a striatal D2 receptor occupancy ranging between 80% and 90% [42], and to exert comparable effects in both behavioral tasks relevant to antipsychotic action [43,44] and studies evaluating neurophysiological and neurochemical outcomes [45][46][47].
At the multivariate analysis, no significant three-way interaction between ROI, treatment, and time was observed (Table 4). A significant ROI*time two-way interaction was found, while no significant treatment*ROI and treatment*time two-way interactions were observed (Table 4). Table 4. Overall output of multivariate statistical analyses. In this table, we report the overall effects of multivariate statistical analyses carried out in the present study. Three-way interaction reported the effect of ROI*dose/treatment*time. Two-way effects were subdivided in within subjects (i.e., ROI*dose/treatment and ROI*time) and between subjects (i.e., dose/treatment*time). Bold highlight indicates significant effect. ROI = Regions of Interest; ASE = asenapine; HAL = haloperidol; DRPA = different receptor profile antipsychotics.

Cortex
No significant treatment effect was observed at either the acute or the chronic time-point (Table 3).
A significant time effect was found in the MC, SS, and IC regions (Table 3). In all these regions, ASE0.1 acute administration induced significantly higher Homer1a expression than ASE0.1 chronic administration (Figure 4). In the MC, acute HAL0.5 also induced significantly higher Homer1a expression than chronic HAL0.5 (Figure 4).
At the multivariate analysis, no significant three-way interaction between ROI, treatment, and time was observed (Table 4). A significant ROI*time two-way interaction was found, while no significant treatment*ROI and treatment*time two-way interactions were observed (Table 4).

Discussion
In this work, we aimed to evaluate whether the molecular effects exerted by antipsychotics on the Homer1a gene, which is a marker of glutamatergic synaptic plasticity and dopamine-glutamate interaction, were modulated by the antipsychotic dose/receptor profile, the duration of administration (acute vs. repeated), and brain regions evaluated. A graphical summarization of the major results of this study is given in Figures 5 and 6.

Discussion
In this work, we aimed to evaluate whether the molecular effects exerted by antipsychotics on the Homer1a gene, which is a marker of glutamatergic synaptic plasticity and dopamine-glutamate interaction, were modulated by the antipsychotic dose/receptor profile, the duration of administration (acute vs. repeated), and brain regions evaluated. A graphical summarization of the major results of this study is given in Figures 5 and 6.
Among the major findings of this work, we found: i) Significant treatment effects, since Homer1a expression was different in the selected areas and time-points among antipsychotics with different receptor profiles; ii) significant dose effects for haloperidol in selected striatal and cortical ROIs at the acute but not the chronic time-point; iii) significant time effects since Homer1a expression by all antipsychotics was significantly higher after acute administration compared to a prolonged treatment regimen; and iiii) a significant three-way interaction between topography (i.e., ROIs), antipsychotic dose, and time for asenapine in the striatum, and a significant two-way interaction between ROI and time in all groups investigated.  Among the major findings of this work, we found: (i) Significant treatment effects, since Homer1a expression was different in the selected areas and time-points among antipsychotics with different receptor profiles; (ii) significant dose effects for haloperidol in selected striatal and cortical ROIs at the acute but not the chronic time-point; (iii) significant time effects since Homer1a expression by all antipsychotics was significantly higher after acute administration compared to a prolonged treatment regimen; and (iv) a significant three-way interaction between topography (i.e., ROIs), antipsychotic dose, and time for asenapine in the striatum, and a significant two-way interaction between ROI and time in all groups investigated.
These data indicate that the effects of the duration of antipsychotic treatment on gene expression were affected by the brain regions where gene expression was measured. Moreover, the extent of gene expression, at least by the antipsychotics tested herein, was strongly affected by the duration of treatment, with acute (i.e., single administration) regimens inducing significantly higher Homer1a expression than chronic ones (i.e., 21-day-long administration period). These data indicate that the effects of the duration of antipsychotic treatment on gene expression were affected by the brain regions where gene expression was measured. Moreover, the extent of gene expression, at least by the antipsychotics tested herein, was strongly affected by the duration of treatment, with acute (i.e., single administration) regimens inducing significantly higher Homer1a expression than chronic ones (i.e., 21-day-long administration period).
The data reinforced the suggestion that the biological effects of antipsychotics on Homer1a depend on a complex combination of the receptor profile, topography, and duration of exposure to these agents. Since Homer is a family of molecules that are crucially implicated in glutamatergic signaling, PSD architecture, and whose expression is profoundly affected by dopaminergic modulation, the evaluation of the present results may take into consideration these biological elements. Antipsychotic-mediated modulation of dopaminergic and possibly non-dopaminergic (including glutamatergic) receptors should be accounted for in mechanistic explanations of changes in Homer1a expression. Homer effects on glutamatergic signaling and synaptic rearrangements should be considered when discussing the antipsychotic-mediated modulation of Homer. The data reinforced the suggestion that the biological effects of antipsychotics on Homer1a depend on a complex combination of the receptor profile, topography, and duration of exposure to these agents. Since Homer is a family of molecules that are crucially implicated in glutamatergic signaling, PSD architecture, and whose expression is profoundly affected by dopaminergic modulation, the evaluation of the present results may take into consideration these biological elements. Antipsychotic-mediated modulation of dopaminergic and possibly non-dopaminergic (including glutamatergic) receptors should be accounted for in mechanistic explanations of changes in Homer1a expression. Homer effects on glutamatergic signaling and synaptic rearrangements should be considered when discussing the antipsychotic-mediated modulation of Homer.
As expected, the first-generation antipsychotic haloperidol induced higher Homer1a expression than olanzapine at the acute time-point in striatal regions. This result is consistent with the hypothesis that antipsychotic-induced Homer1a expression in the striatum mainly depends on the degree of the dopamine D2R antagonism, whereby acute administration of potent D2R-blockers induces greater Homer1a expression in comparison to atypical agents in subcortical regions. Hence, acute variations in subcortical Homer1a mRNA have been proposed by our group as predictors of the typical or atypical character of the antipsychotic molecules [10,48]. These observations comply with previous reports [11,20,49,50]. From a clinical point of view, it may reflect the propensity of typical compounds to provoke extrapyramidal symptoms in humans, as discussed below in more detail.
It has been reported that Homer1a mRNA is under a prominent glutamatergic regulation since NMDAR activation upregulates gene expression [51]. Moreover, Homer1a expression is mediated by selective D2R antagonism [10], and possibly 5-HT2R antagonism and D1R agonism [52,53], either directly or indirectly, via glutamate-mediated pathways. Acute antipsychotic administration is known to increase, rather than decrease, extracellular levels of dopamine in the striatum, exerting a relevant blockade of presynaptic D2Rs [54,55]. Since post-synaptic D2Rs are blocked by the antipsychotic, it may be expected that dopamine may interact with post-synaptic D1Rs. In turn, activated D1Rs may cooperate with NMDARs to intensify NMDAR-mediated glutamatergic transmission [56], leading to increased Homer1a expression. According to these mechanistic considerations, the higher acute Homer1a expression by haloperidol compared to olanzapine appears to indicate a larger action on D2R by haloperidol than olanzapine, which in turn may be responsible for larger presynaptic dopamine release and larger NMDA receptor activation via post-synaptic D1Rs. Although no significant differences between haloperidol and asenapine treatments were found in this study, this may depend on the highly conservative type of post-hoc test we adopted herein, i.e., the Bonferroni correction. Notably, in a previous study [3], we found significant differences between these two agents, corroborating the view above.
A dose effect was found by haloperidol but not by asenapine. Moreover, the dose effect was demonstrated only in selected cortical and striatal areas and only after an acute treatment. Since increased Homer1a expression has been consistently associated to enhanced glutamatergic excitatory activity, possibly as a feedback mechanism to prevent NMDAR-mediated neuronal injury [57][58][59], our results may indicate that, in selected brain regions, acute haloperidol may trigger higher activation of glutamatergic neurons at the intermediate 0.5-mg compared to the high 0.8-mg dose.
At the 0.8-mg high dose, haloperidol may over-saturate synaptic D2Rs and interact with a fraction of D1Rs as well, competing with synaptic dopamine and therefore partially blocking intracellular transduction from these receptors. Antagonism at D1Rs would, in turn, trigger less NMDAR activation as compared to the activation given by the 0.5-mg dose. As a consequence of this reduced NMDAR activation, less Homer1a mRNA would be produced by neurons compared to the intermediate 0.5-mg dose. Indeed, it has been described that more than 90% striatal D2Rs are blocked by haloperidol at doses comparable to those used herein [60,61] and that haloperidol has some affinity to D1Rs, although lower than to D2Rs [62].
The loss of this pattern in chronic treatments suggests that these dynamics may be restricted to acute treatments, while more complex molecular rearrangements may occur during sustained antipsychotics, as it will be discussed below. Repeated administration may allow reaching plasma steady-state and targeting high-level D2Rs, thereby smoothing differences among doses. It has been described, by in vivo recordings of striatal D2R occupancy, that 7-day treatment with 0.25 or 0.5 mg/kg haloperidol resulted in non-significant differences in D2R occupancy, which was higher than 65% [60]. Additionally, behavioral outcomes, such as vacuous chewing movements, were found to be non-significantly different among haloperidol doses reminiscent of those used herein given for eight weeks [63]. Therefore, these studies support the view that the effects of divergent haloperidol doses mitigate during repeated administration and agree with the molecular observations in the present study.
The lack of dose dependence by asenapine may be due to the almost equivalent affinity to D1Rs and D2S by this agent or its strong serotonergic action [36,37], which is predicted to affect dopamine release and may prevent the saturation of D2Rs.
It could not be excluded that other receptor systems may intervene in these dynamics, e.g., the sigma1 receptor for haloperidol or the histaminergic, cholinergic, or serotonergic receptors for asenapine. More studies are needed to understand whether this different pattern is specific to relatively selective D2Rs blockers, such as haloperidol, and are avoided in antipsychotics with a broader receptor profile.
A consistent time effect was found for all antipsychotics tested. Accordingly, the biological effects exerted by haloperidol, asenapine, and olanzapine on Homer1a were significantly higher after a single acute administration than after repeated dosing. These results may indicate that antipsychotic effects on the Homer1a-mediated biological system diminish over repeated administrations, although these outcomes vary depending on the antipsychotic dose and brain region.
One possible explanation for these results is that prolonged antipsychotic treatment may be associated with a reduction of the depolarizing effects on neurons. Indeed, it has been demonstrated that prolonged antipsychotic treatments may cause depolarization blocks and reduce synaptic dopamine [64]. This experimental evidence may explain the loss of the dose effect observed in the present study at the chronic treatment.
Another possible explanation is an upregulation of post-synaptic D2Rs, which may cause the behavioral phenomena of sensitization/tolerance to prolonged antipsychotics [65]. The observed attenuation of Homer1a expression after chronic antipsychotics may represent an adaptive biological mechanism predisposing to antipsychotic sensitization/tolerance. Attenuated Homer1a expression may reflect (or cause) a reduction of glutamatergic neurons' activity in targeted areas as compared to acute administration. In turn, this may depend on D2R-related (and possibly 5-HT2A receptor-related) adaptive changes, i.e., upregulation [66,67], since it has been reported that increased D2R activation reduces glutamate activation via NMDA receptors [68].
Indeed, a time effect has been reported in a behavioral task (conditioned avoidance response, CAR) after repeated haloperidol, risperidone, and olanzapine treatments [69]. Specifically, antipsychotic efficacy in suppressing avoidance response had an early onset and increased progressively [69]. On the other hand, our data showed that antipsychotic effects on Homer1a had an early onset but significantly declined over time. Whether these molecular changes may underlie the reported behavioral sensitization effects of antipsychotics in rodents is still challenging.
Notably, Homer1a overexpression in striatal sites has been associated with reduced locomotor activity [70]. Remarkably, glutamatergic hyperactivity in the striatum has been considered a major pathophysiological mechanism in Parkinson's disease as well as in L-DOPA-induced dyskinesia [71][72][73]. It may be hypothesized that increased Homer1a expression by acute antipsychotics may be the consequence of glutamatergic activation in the striatum, which in turn may concur to cause acute extrapyramidal effects. The relative attenuation of Homer1a expression by persistent antipsychotic treatment may indicate a reduced striatal glutamatergic activation, possibly as a consequence of dopamine receptor sensitization, which may lead to an attenuation of the imbalance in motor-related circuits and reduced motor side effects. This hypothesis paves the way to the possibility of manipulating Homer1a expression as a tool to affect motor side effects of antipsychotics. Nonetheless, this hypothesis should be tested in specifically designed experiments.
Significant acute vs. chronic changes were found in the insular cortex with haloperidol and asenapine. From a neurobiological and neuropharmacological point of view, these results may implicate that the insular cortex represents a preferential site in which antipsychotics exert time-dependent effects, irrespective of the antipsychotic used. From a clinical point of view, the insular cortex has been considered a crucial region for the salience network, which is widely affected by dopaminergic and glutamatergic neurotransmission, and is involved in the ability to discriminate between self-generated and external information [74,75]. Alterations in the salience network and insula overactivations may underlie many of the clinical features of psychotic disorders [76,77]. Moreover, antipsychotic treatments have been found to restore insular abnormalities in patterns of connectivity [78]. In this context, decreased Homer1a expression after chronic antipsychotic exposure in this peculiar region may possibly indicate a reduced glutamatergic activation exerted by neuroleptics, which in turn may underpin potential mitigating effects on abnormal salience signaling.
One challenging result of our study is that despite evidence that topography, treatment, and duration of antipsychotic administration may combine to determine multiple differential patterns of Homer1a expression, multivariate analysis of the interactions among these factors was often negative. Despite the fact that the study was adequately powered to run univariate analyses, the power for multivariate analyses was lower, determining the possibility of false negative outcomes.
In conclusion, Homer1a gene expression is affected by antipsychotics with a specific dose, receptor profile, topography, and time-related dynamics. These data may deepen the knowledge of antipsychotics neurobiology and may have translational valence to infer their unique clinical action and side effects.

Animals and Drug Treatments
Male Sprague-Dawley rats, mean weight 250 g (Charles River Labs, Lecco, Italy), were obtained, and housed in a colony room at controlled temperature and humidity, under a 12 h light/dark cycle, with ad libitum access to laboratory chow and water. In the first days, animals were adapted to human handling. All procedures were approved by the local Animal Care and Use Committee and were in accordance to the NIH Guide for Care and Use of Laboratory Animals (NIH Publication No. 85-23, revised 1996). We used the least number of animals possible to obtain reliable results.

Study Design and Drug Treatment
The following agents were used: (1) Asenapine maleate (supplied as a powder by H. Lundbeck A/S, Copenhagen, Denmark), olanzapine, and haloperidol (both supplied as a powder by Sigma-Aldrich, Milan, Italy). The powder was dissolved in saline solution (NaCl 0.9%), adjusted to physiological pH, and injected i.p. (final volume: 1 mL/kg).

In Situ Hybridization
After sacrifice, animal brains were frozen on powdered dry ice and stored at −70 • C. Sectioning was made on a cryostat at −18 • C. Serial coronal sections of 12 µm were obtained at the level of the middle-rostral striatum (approximately from Bregma 1.20 to 1.00 mm). The rat brain atlas by Paxinos and Watson [85] was used as an anatomical reference. Sections were stored at −80 • C for subsequent analysis.
We initially designed a series of probes for radioactive in situ hybridization by Gene-Bank, among those oligodeoxyribonucleotide sequences that were found complementary to the mRNA sequence of the Homer1a gene. The initially designed sequences were double-checked with BLAST in order to avoid cross-hybridization. The selected probe was that ensuring the best specificity/sensitivity ratio and was manufactured by MWG Biotech, Milan, Italy.
On the day of the hybridization, the sections were prepared to receive the hybridization mix by sequential washes into 4% formaldehyde, PBS (pH 7.4), 0.25% acetic anhydride in 0.1 M triethanolamine/0.9% 25 NaCl, ethanol, chloroform, again in ethanol, and finally air-dried.
The labeling reaction mix consisted of 100 mCi 35 S-dATP, 125 units of terminal deoxynucleotidyl transferase (TdT), and 7.5 pmol/µL of oligodeoxyribonucleotide. The ProbeQuant G-50 Micro Columns (supplied by Amersham-GE Healthcare Biosciences, Italy) were used to separate unincorporated nucleotides from radiolabeled DNA.
Hybridized sections were incubated at 37 • C in a humidified chamber for 22-24 h and subsequently washed in 2× SSC/50% formamide at 43-44 • C in 1× SSC at room temperature. After drying, sections were exposed to an autoradiographic film (Kodak-10 Biomax MR). Sections were co-exposed with a 14 C standards' slide (ARC-146C, American Radiolabeled Chemical). Time of exposure was set to maximize the signal-to-noise ratio.
The ImageJ software (v. 1.46v, http://rsb.info.nih.gov/ij/) was used for quantitation of the optical density of the autoradiographic signal. Brain sections from both rats treated acutely and rats treated chronically were co-exposed to the same X-ray sheet to allow comparative statistical analysis. Optical density was measured within outlined regions of interest (ROIs), in correspondence of the cortex (ACC, MAC, MC, SS, and IC), and the striatum (DM, DL, VM, VL, Cab, and Sab) (Figure 1). The 14 C standard values from 4 through 12 were previously cross-calibrated to 35 S brain paste standards to obtain a calibration curve based on a "best fit" 3rd-degree polynomial. The calibration curve was used to convert the measurements of the optical density into "relative DPM".
For statistical analysis, data from each antipsychotic treatment were given as fold changes of the gene expression to the vehicle-induced expression. Specifically, acute antipsychotic signal intensity was given as fold changes of the acute vehicle signal intensity, and chronic antipsychotic signal intensity as fold changes of the chronic vehicle. The whole in situ hybridization procedure was performed blinded with coded frozen brains.

Statistical Analysis
The statistical analyses were performed using JMP9.0.1 and SPSS 24.0 software. The Shapiro-Wilk test was used to confirm normal distribution of the continuous variable (i.e., signal intensity of gene expression). ROIs, treatment, and time were the categorical variables that were used to analyze gene expression.
As a first goal, we wanted to capture significant effects of independent time or treatment in any topographic subdivision. Therefore, we carried out separate univariate analyses of time and treatment effect in each ROI. In all tests, significance was set at p < 0.05. The Bonferroni corrected post-hoc tests were used for pairwise comparisons in univariate analyses.
To contemporarily evaluate the effects of antipsychotic dose/receptor profile, topography (i.e., ROIs), and duration of treatment (i.e., time), we carried out a 3way mixed ANOVA, with ROIs as the within-subjects factor, and treatment and time as the between-subjects factors. To avoid multiple significant p-values of limited practical relevance, cortical and striatal ROIs were separately assessed since they represent distinct anatomical divisions within the forebrain, although functionally connected.
Follow-up evaluations of 2way interactions with the within-subjects factor (i.e., ROI*treatment; ROI*time) or the between-subjects factors only (treatment*time) were carried out.
According to the aims of this study, three separate analyses were carried out. Paradigm 1 and paradigm 2 examined whether Homer1a expression may be significantly different among increasing doses of a second-generation multireceptor antipsychotic (i.e., asenapine) and a first-generation antipsychotic with almost selective D2R antagonist action (i.e., haloperidol). Accordingly, Homer1a expression was separately compared throughout asenapine-treated animals (asenapine groups, paradigm 1) and throughout haloperidol-treated animals (Haloperidol Groups, Paradigm 2). Paradigm 3 was aimed at evaluating whether multireceptor profile and substantial action on serotonin receptors may affect the cortical or striatal expression of Homer1a differentially. Accordingly, Homer1a expression was compared among the intermediate asenapine and haloperidol doses and a behaviorally compatible olanzapine dose, a multireceptor antipsychotic with a broader receptor profile but narrower action on serotonin receptors compared to asenapine.