COMPARTMENTALIZED D2-DOPAMINE RECEPTORS: ROLES IN SIGNALING & POTENTIAL MECHANISM OF ANTIPSYCHOTIC DRUG ACTION

Dopamine is the most widely distributed catecholamine neurotransmitter in the brain. D2-dopamine receptors (D2R) are one member of a receptor class, known as G-protein coupled receptors, which transduce cellular signals upon an interaction with dopamine. Historically, it was believed that in schizophrenic patients, an excess of dopaminergic signaling at D2R was the cause of psychotic symptoms. Thus, antipsychotics reversed psychosis by blocking excessive dopaminergic signaling from D2R. Sixty years after the introduction of antipsychotic drugs, the connection between the anti-dopaminergic activity of these drugs at D2R and the suppression of psychotic symptoms remains unknown. Understanding the ways that different molecules mediate changes in the ability of D2R to signal and internalize, and thereby affect its cellular compartmentalization is critical to our development of potent antipsychotic agents with fewer side effects. Upon dopamine binding to the receptor, D2R is thought to generate an intracellular signal or signals by activating a heterotrimeric α-β-γ Gprotein that is a component of the receptor complex. In contrast to other G protein beta subunits, Gβ5 is a unique beta subunit that has not been shown to interact with other heterotrimeric G proteins in vivo. Previously, it was assumed that regulators of G-protein signaling (RGSs, specifically the R7 RGS family of proteins) and Gβ5 are required for the physiological activity of Gβ5. Here we show that no R7 RGS proteins are required for Gβ5 to interact with both compartmentalized and non-compartmentalized forms of D2R. Additionally, we have identified one way that Gβ5 may modulate D2R signaling, by specifically blocking the internalization of the receptor in response to dopamine without disrupting the G protein

For example, R7 RGS family proteins contain a Gγ-like (GGL) domain that has been shown to specifically bind Gβ5 subunits and enhance GAP function (32,(38)(39)(40)(41). In fact, it is thought that in vivo, Gβ5 does not form G protein Gβγ dimers and that complex formation between Gβ5 and the GGL domaincontaining R7 RGS proteins is necessary for stabilizing both Gβ5 and R7 RGS proteins (32,(38)(39)(40)(41). The genetic ablation of Gβ5 resulted in the loss of all R7 RGS proteins (39), and conversely, Gβ5 protein was not detected in the retina of a triple knockout mouse line lacking the R7 RGS proteins, RGS6, RGS7, and RGS11 (42). Furthermore, the Gβ5 long isoform (Gβ5l) that forms a complex with the R7 RGS protein, RGS9-1, was absent from the photoreceptors of RGS9 knockout mice (Chen et al. 2000).
However, it has not been demonstrated that Gβ5 exists solely as a heterodimer with R7 RGS proteins in all tissues where Gβ5 may be expressed. Alternative proteins, not abundantly expressed in retinal cells, could contribute to stabilizing Gβ5 expression in other regions.
Previously, it has been shown that the complexes of Gβ5 and R7 RGS proteins can target to D2R and other GPCRs but these interactions are thought to occur through protein domains, such as the DEP domain, that are present within R7 RGS proteins (44-48).
Here we report that Gβ5 can functionally interact with D2R in HEK293 cells and that D2R coexpression stabilizes Gβ5 to enhance Gβ5 expression. Moreover, the D2R-Gβ5 interaction likely occurs independently of R7 RGS proteins suggesting that Gβ5 may have additional cellular functions in addition to its established role as a component of the R7-RGS/Gβ5 complex.

Chemicals
All chemicals and reagents were purchased from Sigma-Aldrich, Fisher Scientific or from suppliers that have been specifically identified below.

Triton X-100 Biochemical Fractionation of Proteins
The method for the Triton X-100 (TX100) biochemical fractionation of proteins has been adapted from our previous publication

Biotinylation of D2R-AP by Biotin Ligase Fusion Proteins
We utilized an in cell biotin transfer assay to detect whether or not Gβ5 interacted with detergent soluble or insoluble forms of D2R. The plasmids containing cDNAs for D2R-AP, Gβ5-BL or KRAS-BL as described in Figure 5A were transfected into HEK293 cells in biotin-free media. 48 hr post-transfection, 10 µM biotin was added to the media. After 2 minutes of incubation at 25 °C, the cells were washed 3 × using ice-cold (4 °C) 1 × PBS. The cells were then vigorously resuspended in ice-cold 2% v/v TX100 lysis buffer and incubated for 1 hr at 4 °C, with vortexing every 15 minutes. Soluble and insoluble proteins were then harvested using the method as described in the "Triton X-100 Biochemical Fractionation of Proteins" section.
We also used the biotinylation assay to detect recruitment of β-arrestin 2 to D2R after dopamine stimulation. The plasmids for D2R-AP and Arr-BL with or without

Receptor Internalization Assay
To determine the effect of overexpression of Gβ subunits (Gβ1 or Gβ5) on receptor internalization we used an ELISA-based assay to determine the amount of receptor present at the plasma membrane after the application of dopamine.

Data Analysis
Signals from the target protein bands were quantified using ImageJ image processing and analysis software (National Institutes of Health, Bethesda, MD, http://rsbweb.nih.gov/ij/). Statistical analyses were performed using Microsoft Excel or GraphPad Prism 4 software (GraphPad Software, Inc.). Images were collected using exposure settings that did not saturate any of the pixels acquired by the camera. The signals resulting from detergent-soluble and insoluble preparations of a protein, respectively, were expressed as a fraction of the total signal and Student's t-test for independent means of unequal variance was used to determine if the amounts of signal from the target protein bands in each experimental group were significantly different. When testing the significance of means for more than 2 experimental groups, one-way ANOVA was used to first determine group statistical significance and only followed by Tukey's post-hoc analysis if the initial comparison was found to be significant.

Coexpression of D2R in HEK293 cells enhances the detergent-resistance of Gβ5
even in the absence of exogenous coexpression of R7 RGS proteins.
We had previously shown that the vast majority of D2-dopamine receptors (D2R) expressed endogenously in the brain, or exogenously in the plasma membrane To provide further support for the idea that targeting to D2R can contribute to enhanced detergent resistance of D2R-interacting proteins the striatum, we compared the detergent-solubility of Gβ5 endogenously expressed in mouse striatum and the cortex. We found that the percent of striatal Gβ5 that was extracted into cold solutions (4 °C) of the non-ionic detergent Triton X-100 was almost halved (from ~40% to 20%), relative to Gβ5 extracted from the cortex (Celver et al. 2012).
One explanation for the increased detergent-resistance of striatal Gβ5 is that D2R, which we have shown is highly resistant to detergent solubilization, is expressed at high concentrations in the striatum compared to the cortex and Gβ5 is then targeted to the detergent-resistant striatal D2R through an interaction with RGS9-2 or other R7 RGS proteins . Therefore, in a control experiment using HEK293 cells, we tested if D2R could enhance the detergentresistance of Gβ5 independently of exogenously expressed R7 RGS proteins.
We found that coexpression of D2R with Gβ5 in HEK293 cells significantly increased the percent of Gβ5 that segregated into the TX100-insoluble cellular fraction (from ~40% to 70%), even in the absence of exogenously coexpressed R7 RGS protein constructs ( Fig. 1A  Dopamine pretreatment (10 µM for 30 min) had no effect the TX100-solubilitiy of Gβ5.
We report that the closely related D2-like dopamine receptor, D4R, also segregates into the TX100-insoluble cellular fraction and that Gβ5 is similarly retargeted to the TX100-insoluble cellular fraction after D4R coexpression ( Fig. 1A, B, C and D).
The above phenomenon is specific to dopamine receptors as we have previously The other, more canonical G protein Gβ subunits are intrinsically resistant to detergent solubilization (Rehm and Ploegh 1997), and thus a similar D2Rmediated retargeting of other Gβ subunits, such as Gβ1, to the TX100-insoluble cellular fraction was not observed ( Fig. 1E and F).

D2R coexpression specifically enhances the expression and stability of Gβ5.
In addition to translocating Gβ5 to the TX100-insoluble fraction we observed that the coexpression of D2R simultaneously and dramatically increased the cellular expression of Gβ5 protein (Fig. 1A, 2A and B, and 3A).
The actions of D2R in increasing Gβ5 expression levels were specific. First, coexpression of D2R increased expression levels of Gβ5 by more than 400%, but, in contrast, coexpression of another GPCR, the mu opioid receptor (MOR) did not significantly alter expression levels of Gβ5 ( Fig. 2A and B).
Second, the expression level of the G protein Gβ subunit, Gβ1, was instead, significantly decreased after D2R coexpression ( Fig. 2C and D).
To explore if D2R-mediated stabilization of Gβ5 contributed to the enhanced Gβ5 expression observed after D2R expression, we treated HEK293 cells expressing Gβ5 alone, or coexpressing D2R and Gβ5, with cycloheximide, a protein translation/synthesis inhibitor, and the decay of the cellular Gβ5 protein signal after cycloheximide treatment for 3 and 6 hr was monitored by Western blotting.
We found that coexpression of D2R significantly decreased the decay of the Gβ5 signal observed at both 3 and 6 hr (Fig. 3). For example, after 6 hr of cycloheximide treatment, the levels of Gβ5 protein in cells expressing Gβ5 alone had decayed to less than 30%, but in cells coexpressing D2R greater than 60% of the original Gβ5 signal remained (Fig. 3). Thus, D2R coexpression significantly inhibited the cellular degradation of Gβ5.

An "in-cell biotin proximity biotinylation assay" indicates physical interactions in living cells between D2R and Gβ5 molecules that segregate into the detergentinsoluble cellular fraction.
Traditional coimmunoprecipitation techniques (Berggard et al. 2007) for probing for either direct or indirect physical interactions between D2R and Gβ5 first require solubilizing the proteins in non-ionic detergents that preserve proteinprotein interactions. Since the vast majority of D2R segregates into a cellular fraction that is insoluble in non-ionic detergents (e.g. TX100) it was not feasible for us to probe for DR interactions using coimmunoprecipitation. Furthermore, we were unable to coimmunoprecipitate D2R and Gβ5 molecules segregating into the TX100-soluble fraction, possibly due to the relatively low concentration of D2R molecules that segregate into this fraction.
Thus, to assess D2R and Gβ5 interactions, we utilized a novel in-cell proximity In accordance with the above results, we show that the majority (~70%) of D2R-AP that was biotinylated by KRAS-BL segregates into the TX100-soluble fraction ( Fig. 4B and C). However, we found that the segregation of D2R-AP biotinylated by Gβ5-BL, more closely matched the segregation of the parent protein with ~70% of biotinylated D2R-AP segregating into the TX100-insoluble fraction ( Fig. 4B and C). These results may be interpreted to suggest that 1) D2R segregating into the TX100-resistant cellular fraction (i.e. the majority of the plasma membrane-expressed D2R) is not compartmentalized from Gβ5 as it was from KRAS and many other cellular proteins and, 2) that Gβ5, unlike other cellular proteins, efficiently interacts in living cells with D2R molecules that segregate into both TX100-soluble and insoluble cellular fractions (Fig. 4D).

Effect of coexpression of Gβ5 on cellular coupling between D2R and Gαo G proteins.
We then tested if the coexpression of Gβ5 could alter the cellular functions of D2R. To test the effects of Gβ5 coexpression on D2R-mediated G protein activation we utilized a bioluminescence resonance energy transfer (BRET) based assay, recently developed by Hollins and colleagues (Hollins et al. 2009), which measures the release of free Gβγ subunits from the activated G protein.
The BRET pair that is utilized is the Gβγ dimer tagged with Venus (Gβγ-Venus) and masGRK3ct-NanoLuc, a Gβγ binding protein construct fused to a newly engineered luciferase variant (64). The use of this system to monitor coupling between D2R and associated G proteins has been described in detail in a Using this assay system and a concentration of dopamine of 1 nM we found that coexpression of either of two different Gβ5 concentrations had no effect on the amplitude of the dopamine and D2R-elicited BRET response ( Fig. 5A and B).
The inability to detect an effect of Gβ5 on the D2R-elicted response amplitude was not due to saturation of the BRET signal as we confirmed, in the same cells, that the dopamine concentration (10 nM) was sub-saturating. Thus, Gβ5 has no effect on the efficacy of coupling of D2R to associated Gαo G proteins.
We then examined the effects of Gβ5 coexpression on the deactivation kinetics of D2R-Gαo G proteins signaling where the dopamine signal was reversed by the application of 100µM haloperidol. At the lower level of Gβ5 expression, obtained using Gβ5 cDNA transfection concentrations that were similar to those utilized the biochemistry experiments described above, no significant effect of Gβ5 was observed on the deactivation kinetics ( Fig. 5A and C). However, with much higher Gβ5 protein concentrations (> 3X), a small but significant acceleration of the deactivation kinetics was detected.

Coexpression of Gβ5, but not Gβ1, inhibits agonist-induced internalization of D2R, and Gβ5 coexpression does not affect agonist-induced internalization of MOR
To quantify receptor internalization we measured the amount of receptor at the surface of HEK293 cells both before and after agonist treatment through a modification of a previously described enzyme-linked immunosorbent assay However, coexpression of Gβ5 had no effect on D2R-AP biotinylation suggesting that Gβ5 did not inhibit recruitment of β-arrestin to D2R.

D2R-Gβ5 interactions can occur independently of R7 RGS proteins.
Several lines of investigations have led to the supposition that Gβ5 is found expressed only as a heterodimer with R7 RGS family proteins (32,38,39,42,43  We have previously reported that when R7 RGS proteins, such as RGS9-2, and Gβ5 are transiently expressed in HEK293 cells, D2R co-expression does not significantly alter protein expression levels of either the R7 RGS protein or Gβ5. In other words when Gβ5 is present in a complex with R7 RGS proteins, D2R coexpression does not enhance or stabilize Gβ5 protein expression. However, here we have reported that D2R coexpression can dramatically enhance levels of transiently coexpressed Gβ5 protein (Fig. 1A, 2A and B, and 3A), indicating that Gβ5 is not in a complex with endogenously expressed R7 RGS proteins.
Thus, our data suggest that, in HEK293 cells, D2R interacts either directly or indirectly with Gβ5, but in a manner that is independent of R7 RGS proteins. It is not clear from our data however if D2R is interacting with the Gβ5 monomer or with a complex of Gβ5 with other cellular proteins such a G protein Gγ subunits.
The D2R-Gβ5 interaction has functional consequences and can bias D2R signaling.
We found that 1) the interaction stabilized and enhanced Gβ5 expression and 2) the interaction inhibited dopamine-induced D2R internalization but did not affect the coupling of D2R to G proteins or the dopamine-mediated recruitment of β-arrestin to D2R. non-canonical G protein-independent cellular signals but do not promote D2Relicited G protein signals (72). However, we believe that this is the first report of a GPCR-interacting cellular protein that modulates the receptor to abolish agonist-induced internalization but does not affect D2R-G protein coupling.
The abolition of dopamine-induced D2R internalization by Gβ5 was not through suppression of interactions with β-arrestin, as Gβ5 did not alter baseline interactions of D2R with β-arrestin or dopamine-induced recruitment of β-arrestin to D2R (7B and C). Gβ5 had no effect on MOR internalization indicating that the prevention of D2R-internalization by Gβ5 likely occurs through targeting of Gβ5 to D2R and is not a consequence of non-specific disruption of the cellular internalization machinery. One model that may be suggested is that internalization of D2R requires one or more bridges between D2R and the cellular internalization, that are in addition to that made through β-arrestin, and Gβ5 expression disrupts such additional connections.
The expression of D2R in detergent-insoluble plasma membrane microcompartments (Sharma et al. 2013) and the targeting of Gβ5 to these microcompartments did not require dopamine pretreatment, indicating that Gβ5 is preassembled in a manner that allows Gβ5 to edit the actions of dopamine at D2R.

D2R-Gβ5 interactions specific and are not caused by non-specific aggregation of the two proteins.
Coexpression of Gβ5 did not alter either the cell surface levels of D2R, the fraction of D2R expressed at the cell surface or the amplitude of D2R-G protein coupling, but clearly inhibited dopamine-induced D2R internalization. These observations indicate that the interaction with D2R and stabilization of Gβ5 was not caused by non-specific aggregation of the two proteins.
The majority of the D4-dopamine receptor, which is a member of the D2-like dopamine receptor family, also segregates into detergent-resistant cellular fractions and recruits Gβ5 to the same biochemical fraction. However, these interactions are unique and do not extend to other cell-expressed GPCRs such as mu opioid receptors (MOR), the vast majority of which are readily solubilized in non-ionic detergents ). In addition, D2R coexpression does not significantly alter the detergent-solubility of Gβ1 ( Fig. 1E and F) or enhance cellular Gβ1 expression levels ( Fig. 2C and D).
Here we have provided evidence for a novel and specific interaction of Gβ5 that is significant because it suggests that Gβ5 can directly modulate D2R, an important GPCR, to bias D2R to signal canonically through G proteins but can prevent dopamine-induced receptor internalization. In addition our data suggests that Gβ5 may be stabilized by protein partners other than R7 RGS proteins. Nevertheless, these experiments were performed in HEK293 cells where concentrations of both D2R and Gβ5 are likely to be higher than that found in native tissue. Hence, definitive in vivo evidence for the above supposition will        A common property of all available antipsychotic drugs is that they specifically block the D2 dopamine receptor (D2R) at therapeutic concentrations (3). Also, it has been demonstrated that the clinical potency of the first generation or "typical" antipsychotics is directly correlated with their relative affinities for D2-like dopamine receptors (Hartman, 1996). From these data it is evident that antipsychotic drugs necessarily block D2R to produce reduction of psychotic symptoms, however it is still unclear how antipsychotic induced blockade of D2R is sufficient to produce this effect.
In this report we demonstrate that all antipsychotics tested in this study can decrease the segregation of D2R into detergent resistant membrane structures (DRMs) and enhance in the accessibility of DRM localized D2R to other cellular proteins, as measured by a novel biotinylation assay. It is this consistent alteration between antipsychotics that represents a model for antipsychotic drug actions and a potential target for the development of antipsychotic drugs that are more directly effective in accomplishing a complete reduction of positive schizophrenic symptoms.
In several large, multi-site randomized controlled-trial studies of the effectiveness of antipsychotic drugs it was determined that there existed little or no differences between the newer atypical and conventional typical antipsychotic drugs (16,17).
Unique among these antipsychotic agents, was the drug clozapine that was found to be efficacious for applications in treatment resistant schizophrenia.
Fundamental to our understanding of the classical mechanism of antipsychotic drug action is that antipsychotics block of the native ligand dopamine from binding to D2R that leads to the prevention of downstream signaling. However, with notable few exceptions, complete occupancy of D2R, by antipsychotics in the brain is sufficient to produce serious extrapyramidal side (EPS) effects and therefore demonstrates the need to titrate doses accordingly to produce a consistent reduction in schizophrenic symptoms (7). Interestingly even when sufficiently high doses of the atypical agent quetiapine were used to produce clinically significant reduction of schizophrenic symptoms, receptor occupancy was significantly lower than the "therapeutic window" (8). It therefore seems unlikely that simple measures of receptor occupancy are conclusive in determining antipsychotic efficacy.
The difference in effectiveness of suppression of psychotic symptoms in schizophrenia seen with the newer atypical antipsychotics might be due to their decreased binding affinity of D2R (9,10). Differences between the binding affinities of atypical and typical antipsychotics are almost entirely (~99%) due to the relative differences in the dissociation constant. The enhanced dissociation of antipsychotic from receptor is proposed to enable endogenous dopamine to continue to signal via D2Rs, despite receptor blockade, and may be involved in the reduction of extrapyramidal symptoms in the atypical antipsychotics. Yet, the difference between the relative binding affinities of typical and atypical drugs does not explain the unique effectiveness of clozapine in producing a reduction in treatment-resistant schizophrenic symptoms.
Previous studies on the antipsychotic interaction with D2R have measured ligand-binding properties using receptor radioactive assay or PET imaging (7,8,9). Although these types of studies can describe quantitative changes in  (13). Together these discoveries suggest that not only does the majority of cellular exist in a membrane fraction that is resistant to detergent solubilization, but also that this fraction specifically allows for the compartmentalization of D2R and its associated signal transducers. Therefore, drugs that are able to alter the relative compartmentalization of D2R into or out of DRMs could potentially alter its signaling characteristics. Here

TX100 Biochemical Fractionation of Proteins
The method for the Triton X-100 (TX100) biochemical fractionation of proteins has been adapted from previous publications (11,13).

In-Cell Biotin Transfer Assay
We utilized an in cell biotin transfer assay to detect whether or not KRAS-BL
For HRP conjugated antibodies, blots were developed with Supersignal West Femto substrate (Pierce), and images were taken on the Chemiluminescence setting on Bio Rad Gel Doc XRS. The intensity of each band was quantified using Image J software. Images were taken using exposure settings that did not saturate any of the charge-coupled device camera pixels.

Data and Statistical Analysis
Signals from the target protein bands were quantified using the free image processing and analysis software ImageJ (National Institutes of Health, Bethesda, MD, http://rsbweb.nih.gov/ij/). Statistical analyses were performed using Microsoft Excel or GraphPad Prism 4 software (GraphPad Software, Inc.).
The signals resulting from detergent-soluble and insoluble preparations of a protein, respectively, were expressed as a fraction of the total signal per sample of cells or in cases specifically indicated as a fraction of vehicle in the corresponding (T100 soluble or insoluble) sample. When testing the significance of means for more than 2 experimental groups, one-way ANOVA was used to first determine group statistical significance and only followed by either Dunnett's post-hoc test, for determination of mean difference from vehicle treated controls or Tukey's post-hoc analysis if comparing between multiple different treatment conditions. Post-hoc analyses were performed only if the results of the initial ANOVA were determined to be significant (p<0.05).

All antipsychotic drugs tested enhanced the detergent solubility of total cellular
D2R-AP without enhancing expression of D2R-AP, with the exception of clozapine.
Previously we have shown that D2R expressed in the brain or exogenously in HEK293 cells exists predominantly within a fraction of the plasma membrane that is insoluble in nonionic detergent and D2R retargets proteins to these biochemical fractions (11,13). Furthermore we have demonstrated that these insoluble biochemical fractions respond to dopamine treatment by reducing available cell surface D2R within this fraction, likely through internalization (13).
Therefore we asked whether or not alternative D2R binding ligands produce changes in the accessibility of either (soluble or insoluble) pool of D2R.
24 hour, saturating concentrations of antipsychotics appeared to specifically enhance the pool of soluble D2R-AP (Fig. 1B). The exception to this statement was the antipsychotic drug, clozapine, which showed no difference from vehicle treated cells, in terms of solubility of all cellular D2R-AP. Furthermore, antipsychotics did not uniformly enhance the expression of total D2R-AP under these treatment conditions (Fig. 1C). Rather, it appeared that two drugs, resperidone and aripiprazole, decreased the overall expression of the D2R-AP construct.
We have previously demonstrated that the insoluble pool of receptor is functionally segregated at the plasma membrane (13). Therefore it may be that antipsychotics that enhance the soluble fraction of cellular D2R thereby disrupt targeting to insoluble biochemical fraction. It is likely that through an alteration in the targeting of D2R to different compartments in the plasma membrane alters the ability of D2R to signal once it has reached this compartment.

Antipsychotic drugs enhanced surface translocation of D2R-AP as measured by modified ELISA assay
To compare the relative abilities of these antipsychotics to enhance the surface receptor concentrations we transiently expressed only D2R-AP in 96-well plates and treated these cells with a 10µM concentration of antipsychotic for 24 hours.
We found that after this treatment, total surface receptors were significantly increased from 1.7 to approximately 5 fold after the drug treatment (Fig. 2).
Haloperidol and clozapine enhance surface translocation of D2R-AP, but do not increase detergent solubility of surface D2R-AP.
In order to examine whether or not the effect of antipsychotics on the solubility of total cellular D2R-AP was also conferred to surface receptors we examined the solubility of surface receptors after antipsychotic treatment using an indirect detection method summarized in figure 3A. Although haloperidol and clozapine both robustly enhanced the total amounts of cell surface receptors (Fig. 3D), we found that the relative solubility of the surface receptors was unchanged compared to vehicle treated cells (Fig. 3C). A summary of this effect is provided in figure 9A.

Certain antipsychotic drugs enhance the accessibility of D2R-AP to interactions
occurring at the plasma membrane.
To assess the potential for changes in the cell signaling of D2R upon antipsychotic drug treatment we used an in-cell, proximity-dependent, biotin transfer assay that involves the E. coli biotin ligase (BL). Biotinylation of D2R-AP, which occurs within 5 minutes prior to cell lysis, is further evidence of an interaction that has occurred in living cells (14,15).
To assess if antipsychotics could disrupt the compartmentalization of D2R-AP at the plasma membrane, we used an in-cell proximity dependent biotinylation, a diagram of the proximity dependent biotinylation assay is provided in figure 4.
The effect of antipsychotics on the accessibility of cell surface D2R is likely due to the increased receptors levels that were observed in figure 2.
Several antipsychotics enhance the accessibility of cell surface D2R-AP as measured by in-cell proximity dependent biotinylation assay, however haloperidol most robustly enhances accessibility of insoluble pools of receptor ( fig. 5B).
Haloperidol treatment alters the biochemical fractionation of biotinylated D2R-AP, by enhancing soluble fraction of biotinylated D2R-AP (Fig. 5C). Additionally, haloperidol most robustly enhanced insoluble biotinylated D2R-AP ( Fig 5D) and robustly enhanced the biotinylation of D2R-AP only after 4 hours of treatment ( Fig 7B). However, there was no effect observed in the biotinylation of soluble D2R-AP at 4 hours (Fig. 7C).
Biotinylation in the insoluble pool of receptor increased more dramatically with haloperidol treatment, approximately 8 fold, than those of soluble receptors, which differed from vehicle by only ~3 fold ( Fig. 6C and D).
Haloperidol's effect on insoluble D2R-AP compartmentalization is blocked by the drug clozapine.
To determine if the effect of haloperidol was specific to the interaction of haloperidol with the receptor and not through a non-specific interaction, we treated D2R-AP expressing cells with a 1µM concentration of haloperidol and found an increase in biotinylation of D2R-AP in both soluble and insoluble fractions (Fig. 8B) which was consistent with our previously observed results in figures 6C and D. A similar strategy was originally used in the discover of D2dopamine receptors, which used other antipsychotic drugs could block tritiated haloperidol from interaction with brain tissue (16,17). A diagrammatic representation of the results obtained of decreased insoluble D2R-AP compartmentalization is provided in figure 9B.

Unique responses of plasma membrane and intracellular D2R to antipsychotic treatment
Understanding the way D2R functions under antipsychotic blockade is crucial to the future development of safer and more effective of antipsychotic drugs.
Generally, greater than 50% of receptors are bound by antipsychotic drugs for therapeutically effective concentration of these agents (18). Because of a wide difference in the relative affinity of antipsychotics for D2R (e.g. approximately 225 fold difference in binding constant of droperidol relative to quetiapine) we used saturating (10µM) concentrations of all antipsychotics to study their function on the receptor (30).
Previously we demonstrated that both detergent insoluble form and soluble forms of D2R exist within the plasma membrane (13). Insoluble D2R has recently been shown to exhibit the property of plasma membrane compartmentalization in HEK293 cells (13). Whereas, the detergent soluble form of D2R is likely to originate from a fluid region of the cell membrane that does not functionally restrict the access of D2R to other signaling molecules (19). We found that 1) all antipsychotics enhanced the accessibility of cell surface receptors and 2) the majority of antipsychotics enhanced the solubility of total cellular receptors.
Importantly, the total solubility of D2R is not representative of surface D2R solubility, as evidenced by the observation that when we examined the solubility of cell surface receptors, we found that haloperidol did not enhance the fraction of soluble receptors compared to vehicle. It is possible that D2R exhibits undiscovered intracellular roles in addition to the signaling, as there are significant pools of intracellular D2R in striatal neurons and in transfected cell lines (20).
Previously it has been shown that alternative signaling via arrestin may occur in endosomal targeted pools of adrenoreceptors (21). Furthermore, internalization processes mediated through arrestin may activate a number of non-G-protein mediated signal transduction mechanisms that may yet be undescribed (22).
Therefore it is unlikely that the only physiologically important forms of the receptor are those that are targeted to the cellular plasma membrane.

Changes in receptor compartmentalization mediated by antipsychotics is unlikely
to be mediated through to lipid-drug interactions.
It is a common assumption that the therapeutic There are chemical agents that have been used to disrupt theoretical plasma membrane compartments, such as β-methylcyclodextrin (βMCD). βMCD is a widely used to chelate cholesterol that are thought to disrupts plasma membrane lipid raft compartments, through sequestration of cholesterol (24). However haloperidol's disruption of insoluble D2R compartments, is unlikely to occur through a non-specific disruption of detergent insoluble pools because 1) surface D2R is not made more soluble upon administration of the antipsychotic drug haloperidol and the 2) effect of increasing biotinylation of soluble D2R-AP is blocked through the co-administration of the drug clozapine.

Altered D2R compartmentalization likely produces changes in receptor signaling
In general, the effect of antipsychotic drugs in increasing the total biotinylation of D2R-AP was likely mediated through a similar process as the total enhancement of cell surface D2R-AP because these two results tended to correlate well.
However, we found haloperidol uniquely disrupts the plasma membrane compartmentalization of insoluble D2R-AP, at the cell surface.
Signaling mediated by the D2R-dopamine complex involves the G protein activation that suppresses the formation of cAMP, though gα i -mediated inhibition of adenylate cyclase. As an enzymatic reaction, the rate of activation of G proteins is dependent on both those molecules' concentration and proximity.
Spatial restriction limits the local concentration of signaling molecules and can enhance the local cellular concentration of signaling molecules that could be diminished in their global cellular concentration. Therefore, by enhancing the accessibility of insoluble D2R to plasma membrane molecules, the unique capacity of restricted-access, plasma membrane targeted D2R to limit signaling to certain molecules has been disrupted.
On the other hand, most antipsychotics do not change the accessible fraction of cell surface receptor and this change does not appear to alter the ability of receptors at the cell surface to interact with membrane-targeted molecules.
Rather, antipsychotic induced plasma membrane targeted receptors exhibit similar potential signaling properties to receptors that were integrated in the plasma membrane prior to antipsychotic treatment; i.e. there was minimal difference between total receptor interactions normalized per unit of receptor between drugs. Therefore, by virtue of enhanced surface expression and altered solubility, antipsychotic binding may produce actions of D2R on yet undefined pathways and not on classically defined interactions with G proteins and ion channels (25).
Although a number of authors have suggested moieties within D2R that reduce the relative affinity of haloperidol or other antipsychotic drugs for D2R, at the time of publication, no mutation construct has been shown to demonstrate selectivity of drug-receptor interaction (26)(27)(28)(29). Furthermore, by significantly disrupting the largely hydrophobic binding pocket of D2R, this process may produce receptors that lack proper folding or membrane targeting.

Antipsychotic actions of clozapine may be related an ability to maintain a large intracellular pool of insoluble D2R.
Although binding of the dopamine D2 receptor has been shown to be directly correlated to the therapeutic action of antipsychotic drugs, it is unclear how antagonistic blockade of D2R relates to the therapeutic effectiveness (16,30).
We have shown here that antipsychotic drugs all seem to enhance the surface translocation of D2R-AP, through a mechanism independent of total expression of receptor. This result fits with the well-documented ability of antipsychotic agents to antagonize D2R arrestin recruitment and internalization (31)(32)(33).
However, it appears that most of the antipsychotics we tested also demonstrate an ability to enhance the soluble fraction of D2R, through an undocumented process. The exception to this finding was the drug clozapine, an agent that is uniquely efficacious at reducing the psychotic symptoms of treatment resistant schizophrenia (34).
Previously Seeman et al. demonstrated that antipsychotic drug potencies directly correlated with their relative binding affinities to D2R. Interestingly, clozapine, a drug with lowered incidence of movement disorder side effects also fits this correlation. Therefore it is likely that clozapine mediates its antipsychotic therapeutic action through binding to D2R (16,17). The reason why certain the atypical antipsychotics possess this lower liability might be due to a rapid dissociation of the drug from the D2R. However recently it has been shown that, at least in electrophysiology experiments with the Xenopus laevis oocyte system, that most antipsychotic do not fully dissociate from their target receptor, even after extended washing (35,36).
The data we present here implicates the solubility of total cellular receptor as a potential therapeutic target for the development of antipsychotics that exhibit similar properties to clozapine.

Conclusions
D2R is a receptor with numerous physiological functions that is implicated in the pharmacotherapy of a number of diseases, notably Parkinson's disease and schizophrenia. Although it is more than 60 years since the discovery of the first antipsychotic agent, their exact mechanism of action remains to be elucidated.
Here we show that antipsychotics can produce changes in the biochemical characteristics of D2R and that these changes are likely mediated through direct binding to D2R. It may be that antipsychotics produce changes in cell signaling through directly blocking the interaction of D2R and dopamine, but the effects of antipsychotics on receptor compartmentalization remain an intriguing possibility for the future development of more efficacious agents. Critical to our understanding of D2R function is the further elucidation of compartmentalized, non-compartmentalized, and intracellular forms of D2R-AP.