The role of the serotonin transporter in prefrontal cortex glutamatergic signaling following short‐ and long‐access cocaine self‐administration

Abstract Vulnerability to drug addiction relies on substantial individual differences. We previously demonstrated that serotonin transporter knockout (SERT−/−) rats show increased cocaine intake and develop signs of compulsivity. However, the underlying neural mechanisms are not fully understood. Given the pivotal role of glutamate and prefrontal cortex in cocaine‐seeking behavior, we sought to investigate the expression of proteins implicated in glutamate neurotransmission in the prefrontal cortex of naïve and cocaine‐exposed rats lacking SERT. We focused on the infralimbic (ILc) and prelimbic (PLc) cortices, which are theorized to exert opposing effects on the control over subcortical brain areas. SERT−/− rats, which compared to wild‐type (SERT+/+) rats show increased ShA and LgA intake short‐access (ShA) and long‐access (LgA) cocaine intake, were sacrificed 24 h into withdrawal for ex vivo molecular analyses. In the ILc homogenate of SERT−/− rats, we observed a sharp increase in glial glutamate transporter 1 (GLT‐1) after ShA, but not LgA, cocaine intake. This was paralleled by ShA‐induced increases in GluN1, GluN2A, and GluN2B NMDA receptor subunits and their scaffolding protein SAP102 in the ILc homogenate, but not postsynaptic density, of these knockout animals. In the PLc, we found no major changes in the homogenate; conversely, the expression of GluN1 and GluN2A NMDA receptor subunits was increased in the postsynaptic density under ShA conditions and reduced under LgA conditions. These results point to SERT as a critical regulator of glutamate homeostasis in a way that differs between the subregions investigated, the duration of cocaine exposure as well as the cellular compartment analyzed.


| INTRODUCTION
Drug addiction is classified as a compulsive and relapsing psychiatric disease, characterized by the transition from limited to compulsive use. Such transition may result from a negative emotional state that becomes manifest during abstinence. 1 One molecule that plays a key role in this transition to compulsivity is serotonin. 2 The plasmalemmal serotonin transporter (SERT), whose main function is to uptake serotonin back into the presynaptic terminal, plays a critical role in negative emotionality as well as drug intake. In fact, the lack of SERT in animals has been closely associated with increased anxiety 3 and with a higher intake of cocaine. 4 We have recently focused our attention on cocaine self-administration behaviors using different conditions of psychostimulant intake that mimic both limited or compulsive drug use, the so-called short-access (ShA) or long-access (LgA) cocaine, respectively. 2,[4][5][6] We found that cocaine intake was increased in SERT knockout (SERT −/− ) rats under both ShA and LgA conditions and that anxiety was increased 24 h into withdrawal from both conditions. This implies that increases in both regular and compulsive cocaine intake is, at least in part, driven by increased anxiety, adding critical information with respect to the neural mechanisms involved in the pathophysiology underlying addictive behavior.
The medial prefrontal cortex (mPFC) plays an essential role in drug seeking, abstinence, and relapse, as inferred from rodent studies. [7][8][9] In addition, it is well established that the mPFC plays also a critical role in decision-making and behavioral flexibility. mPFC receives excitatory glutamatergic inputs from the sensory systems and adjusts them to the signals received from the hippocampus and amygdala. 10 When the activity of mPFC is dysregulated, the balance between the promotion or inhibition of a given function may be altered as well. 11 The glutamate system plays a key role in the functioning of the mPFC. For instance, withdrawal from cocaine selfadministration is associated with an increase in extracellular glutamate levels 12 and an upregulation of the GluN2B receptor subunit. 13 Furthermore, we have previously shown that repeated cocaine injections render the glutamatergic synapse of the mPFC sensitive to stress, an event that is suggested to play a central role in the reinstatement of cocaine seeking. 14,15 To this end, investigating the role of mPFC to drug seeking is critical as, for instance, cortical hypofrontality appears to be associated with cognitive impairments in cocaine addicts contributing to compulsive drug seeking. [16][17][18][19] From an anatomical point of view, the mPFC is divided into infralimbic (ILc) and prelimbic (PLc) cortices. Notably, although these two brain regions are in close proximity within the brain, they play different roles, sometimes even opposite, highlighting a functional dichotomy of these two brain regions. For instance, the PLc is recruited for the initiation of a conditioned fear response, whereas the ILc initiates, through extinction training, the suppression of conditioned fear responding. 20,21 Similar to fear conditioning, these two brain regions serve distinct functions in the response to the psychostimulant cocaine. In fact, the PLc appears to be critical for the initiation of cocaine seeking, whereas the ILc is engaged by extinction learning to suppress cocaine seeking. [22][23][24] Surprisingly, we know very little about cocaine-induced plasticity in the ILc and PLc, especially in terms of changes of excitatory signaling.
We have previously investigated glutamate homeostasis in the habenula of SERT −/− rats. 25 It has been established that cocaine enhances the glutamatergic transmission in those neurons projecting from habenula to the rostral tegmental area 26 that, in turn, influence the serotonergic tone via inhibition of the dorsal raphe nucleus. 27 We found that the deletion of SERT altered cocaine-induced glutamate homeostasis of the habenula. However, the habenula does not function in isolation, and evidence exists showing that glutamatergic determinants of the mPFC are altered in SERT −/− rats, both at baseline and following cocaine exposure. 5,28 To the best of our knowledge, no evidence exists on the role of SERT in the regulation of the glutamatergic synapse in the IL versus PL cortical subregions.
To fill this gap, SERT +/+ and SERT −/− rats were exposed to both ShA (1 h/day, 0.5 mg/kg/infusion) or LgA cocaine self-administration protocol (6 h/day, 0.5 mg/kg/infusion) to mimic limited consumption or the loss of control over drug intake, respectively. 29,30 By comparing cocaine-naive rats to rats exposed to ShA or LgA of cocaine, we evaluated the effects of SERT removal on critical determinants of glutamate homeostasis such as (1) the glial glutamate transporter 1 (GLT-1), which is responsible of the termination of glutamate neurotransmission by mediating reuptake of glutamate back into the presynaptic terminal and (2) the different subunits of the NMDA receptor (GluN1, GluN2A, and GluN2B), which represent the main glutamate receptor complex responsible of calcium influx into the cell, 31 as well as the scaffolding proteins PSD95 and SAP102, which are responsible of the anchoring of NMDA receptor tight to the postsynaptic density. 32 We evaluated the expression of these molecules in the whole homogenate, which informs us primarily about translational changes, as well as in the postsynaptic density (PSD), which gives us a clue of synaptic localization of receptors and the respective scaffolding proteins. In doing so, we took advantage of unused brain material collected from naïve and ShA and LgA cocaine SERT −/− as well as wildtype rats in a previous study 25 and examined the glutamate synapse in the ILc and PLc.

| Cocaine self-administration
Briefly, 1 week after surgery, rats were trained to self-administer cocaine (0.5 mg/kg/infusion) under a fixed ratio 1 (FR1) schedule of reinforcement for details 6,9 Two days after cocaine self-administration training, rats were allowed to self-administer cocaine during daily 6-h sessions (extended or LgA group of rats) or 1-h sessions (limited of ShA group of rats) for a total of 15 days. 30 Additional groups of cocainenaive SERT −/− and SERT +/+ rats also underwent intravenous catheterization, were handled daily, and received daily infusion of heparinized saline but were not exposed to the self-administration chambers. 6,9

| Tissue collection
Twenty-four hours following the last cocaine self-administration session, rats were sacrificed by decapitation, brains were quickly collected and stored at −80 C. Using the rat brain atlas, 36 the PLc and ILc (coordinates between bregma +4.20 mm and bregma +2.52 mm, Figure 1) were punched from frozen brain sections of 220 μm using a sterile 1-mm-diameter needle. 37 Punches from the right and left hemisphere were pooled. Prelimbic and infralimbic tissue was stored at −80 C until being processed for molecular analysis (see below).

| Protein extraction and western blot analyses
Proteins were extracted as previously described 38  Results were standardized using β-actin as the control protein, which was detected by evaluating the band density at 43 kDa.
Immunocomplexes were visualized by chemiluminescence using the ChemiDoc MP Imaging System (Bio-Rad Laboratories). Gels were run two times each, and the results represent the average from two different runs.

| Statistical analysis
Data were collected in individual animals (independent determinations) and are presented as means ± standard errors.
To enable visual comparisons across genotypes with different degrees of expression of glutamatergic synapse components, values are presented as percent of the control group, namely, the SERT +/+naive group that was not exposed to cocaine self-administration.
Molecular changes produced by genotype and cocaine exposure alone as well as by their combination were analyzed using a two-way analysis of variance (ANOVA), with genotype and cocaine selfadministration as independent variables. When dictated by relevant interaction terms, Fisher's least significant difference (LSD) test was used to characterize differences among individual groups of rats.
However, when no interaction between genotype and cocaine selfadministration was observed, only the main effects were reported.
Significance for all tests was assumed at p < 0.05.

| Cocaine intake
As we reported previously, 25 no genotype differences were observed during the acquisition of cocaine self-administration. Under ShA conditions, the daily number of cocaine infusion was higher in SERT −/− versus SERT +/+ rats, leading to a higher total cocaine intake in SERT −/− (172 ± 24 infusions) versus SERT +/+ (81 ± 13 infusions) rats.
When the rats were allowed to self-administer cocaine under LgA, the daily number of cocaine infusions was higher in SERT −/− versus SERT +/+ rats, leading to a higher total cocaine intake in SERT −/−   3.4 | Expression levels of GluN2A and GluN2B NMDA receptor subunits in the ILc and PLc homogenate and postsynaptic density following ShA and LgA to cocaine in SERT +/+ and SERT −/− rats Next, we investigated the expression of two accessory subunits of the NMDA receptor: GluN2A and GluN2B (Figures 4 and 5, respectively).
Two-way ANOVA of GluN2A and GluN2B expression in the homogenate of ILc revealed a main effect of cocaine access (GluN2A:    F I G U R E 6 Interaction between serotonin transporter (SERT) deletion and cocaine self-administration on the NMDA receptor-related scaffolding protein SAP102 in the infralimbic (ILc) and prelimbic (PLc) cortices. Protein levels of SAP102 in the homogenate and postsynaptic density of ILc (Panel A,B) and PLc (Panel C,D) are expressed as percentages of SERT +/+ -naive rats. Below the graphs, representative immunoblots are shown for SAP102 (102 kDa) scaffolding protein in the homogenate and postsynaptic density of ILc (Panel A,B) and PLc (Panel C,D), respectively. Histograms represent the mean ± SEM of five to six rats per group. * p < 0.05, ** p < 0.01 versus SERT +/+ naive; # p < 0.05 versus SERT +/+ ShA; § p < 0.05, § § § p < 0.001 versus SERT −/− naive; $ p < 0.05, $$ p < 0.01 versus SERT −/− ShA; ££ p < 0.01 versus SERT −/− LgA (two-way ANOVA followed by Fisher's least significant difference [LSD] multiple comparisons test)

| DISCUSSION
Our data show that deletion of SERT influences the homeostasis of the glutamatergic synapse in both the ILc and PLc and confers a different reactivity to the two different cocaine exposure regimens. Taking together our behavioral and molecular findings, we depict a situation that points to the SERT as a master regulator of basal glutamate homeostasis, primarily in the ILc. Notably, its removal dictates a profile of responsivity to cocaine that is different based on the two subregions of the prefrontal cortex, the different modality of cocaine self-administration as well as the cellular district taken into account.
In the ILc of SERT +/+ rats, we found that ShA cocaine did not affect the glial glutamate transporter GLT-1, whereas LgA cocaine intake caused a decrease in GLT-1 expression levels. Previous studies reported a reduced GLT-1 expression in the nucleus accumbens after cocaine self-administration for 2 h/day for 2 weeks combined with 3 weeks of extinction. 39 Because this was associated with a decrease in glutamate uptake, 39 it is possible that a decrease in GLT-1 expression leads to an overflow of glutamate. Although cocaine exposure in a conditioned place preference test also decreased GLT-1 in the nucleus accumbens, it did not affect GLT-1 in the dorsomedial prefrontal cortex. 40 Here, we extend the nucleus accumbens findings to the ILc. 41 In the ILc of SERT −/− rats, GLT-1 protein levels were unaffected under baseline conditions, increased after ShA cocaine intake and nor- hypothesize that the overall high peak in the expression of these glutamatergic molecular determinants following ShA may represent a (mal)adaptive response presumably secondary to decreased glutamate levels (see above), which may lead to increased synthesis of these glutamatergic proteins extrasynaptically. This may serve to regulate volume glutamate transmission but, perhaps, not translate to altered synaptic glutamate transmission. Thus, although the expression of NMDA receptor subunits is increased in the ILc postsynaptic density, such upregulation is not paralleled by an increased expression of the scaffolding protein specific for the NMDA receptor complex, that is, SAP102. This presumably leads to unstable synapses. Taken together, these results do indicate that, in the ILc of SERT −/− rats, there is an overall dysregulation of the glutamatergic synapse.
The PLc is characterized by a different situation. In SERT +/+ rats, no changes in any of the glutamatergic components (except for GluN1 in the postsynaptic density) were found. In the homogenate of naïve SERT −/− rats, no changes in glutamatergic system determinants were observed as well. Yet in the postsynaptic density, we found an overall trend towards increased expression of glutamatergic markers following ShA, which was significant for GluN2A and GluN2B. Because GLT-1 levels were not altered in this brain subregion, we cannot point to the receptor upregulation as a neuroadaptive response to glutamate efflux. Hence, a different mechanism may come into play to fos- LgA and suggests that, in this subregion, ShA-exposed rats are able to mount a neuroadaptive response that wanes following LgA, both in term of synthesis or receptor localization. Focusing instead on LgA- In conclusion, our findings suggest that SERT influences the homeostasis of the glutamate synapse both under basal conditions and following repeated cocaine self-administration. From our data, it appears that removal of SERT reorganizes and destabilizes the cortical glutamate synapse, presumably providing a ground for the higher intake of cocaine exhibited by SERT −/− rats. Because SERT −/− rats consume a higher amount of cocaine, our study does not inform whether it is removal of SERT, exposure to cocaine, or a combination of both that have generated the observed pattern of glutamatergic changes in the Ilc and PLc. In case of cocaine effects, cocaine-exposure-specific changes in 5-HT receptor expression may very well have contributed to the observed ShA and LgA effects in the glutamate system. 43 Future research should address in more detail the mechanisms leading to the glutamatergic changes. The ILc is generally considered to inhibit subcortical areas, and the PLc to stimulate subcortical areas. 44 Previous studies showed that impaired fear extinction in SERT −/− rodents is associated with reduced expression of the neuronal activity marker c-Fos in the ILc but not PLc, 45 increased theta synchronization between the ILc and lateral amygdala, 46 and an increase in the apical dendritic branches of ILc pyramidal neurons. 47 These findings do point out that deletion of SERT affects the structure and function of the ILc, potentially comparable to the finding of reduced connectivity between the prefrontal cortex and amygdala reported in humans carrying the short allelic variant of the serotonin transporter promoter polymorphism. 48 Because SERT −/− rats not only show impaired fear extinction, but also an impairment in the extinction of cocaine-seeking behavior after ShA and LgA cocaine self-administration, 4,49 functional changes in the ILc may contribute to their difficulties to refrain from cocaine intake and thereby their increased cocaine intake specifically under LgA conditions. Further changes in the PLc postsynaptic density, such as decreased GluN2B receptor expression, may aggravate this effect.

ACKNOWLEDGMENT
This work was supported by the Dipartimento delle Politiche Antidroga (Rome, Italy) through the ERANID Grant "STANDUP" awarded to F.F. and by ZonMW through the ERANID Grant "STANDUP" awarded to J.H., as well as by grants from MIUR Progetto Eccellenza.

CONFLICT OF INTEREST
The authors declare no conflict of interest in relation to the work herein described.