Single Dose of Amphetamine Induces Delayed Subregional Attenuation of Cholinergic Interneuron Activity in the Striatum

Abstract Psychostimulants such as amphetamine (AMPH) target dopamine (DA) neuron synapses to engender drug-induced plasticity. While DA neurons modulate the activity of striatal (Str) cholinergic interneurons (ChIs) with regional heterogeneity, how AMPH affects ChI activity has not been elucidated. Here, we applied quantitative fluorescence imaging approaches to map the dose-dependent effects of a single dose of AMPH on ChI activity at 2.5 and 24 h after injection across the mouse Str using the activity-dependent marker phosphorylated ribosomal protein S6 (p-rpS6240/244). AMPH did not affect the distribution or morphology of ChIs in any Str subregion. While AMPH at either dose had no effect on ChI activity after 2.5 h, ChI activity was dose dependently reduced after 24 h specifically in the ventral Str/nucleus accumbens (NAc), a critical site of psychostimulant action. AMPH at either dose did not affect the spontaneous firing of ChIs. Altogether this work demonstrates that a single dose of AMPH has delayed regionally heterogeneous effects on ChI activity, which most likely involves extra-Str synaptic input.


Introduction
Psychostimulants such as amphetamine (AMPH) target dopamine (DA) neuron terminals (Sulzer, 2011) and engender dose-dependent behavioral effects. DA release in the ventral striatum/nucleus accumbens (Str/NAc) is associated with hyperlocomotion, whereas DA release in the dorsal Str is associated with stereotypies (Robinson and Becker, 1986;Kalivas and Stewart, 1991;Gaytan et al., 1998;Yates et al., 2007). DA neurons modulate the activity of cholinergic interneurons (ChIs), which comprise ,2% of striatal (Str) neurons, and yet strongly control the Str circuitry (Goldberg and Wilson, 2010;Gonzales and Smith, 2015;Abudukeyoumu et al., 2019). Modulation of ChI activity is critical for the processing and reinforcement of reward-related behaviors (Atallah et al., 2014;Gonzales and Smith, 2015). ChIs in the ventral Str are crucial for psychostimulant-dependent behaviors (Sofuoglu and Mooney, 2009;Witten et al., 2010;Lee et al., 2020;Lewis and Borrelli, 2020). However, whether AMPH has subregional effects on ChI activity has not been elucidated.
Previous studies have shown that the phosphorylated form of the ribosomal protein S6 at serine 240 and 244 residues (p-rpS6 240/244 ) reports activity of ChIs under different pharmacological and/or behavioral conditions (Bertran-Gonzalez et al., 2012;Kharkwal et al., 2016;Matamales et al., 2016a,b). The phosphorylation of rpS6 can be induced by multiple signaling cascades; mTORC1 pathway and/or mTORC1-independent pathways such as the PKC, the MAPK or the cAMP/PKA pathways (Valjent et al., 2011;Bertran-Gonzalez et al., 2012;Gangarossa and Valjent, 2012). The phosphorylation of rpS6 appears to occur sequentially at five serine residues: in the order 236, 235, 240, 244, and 247 (Knight et al., 2012;Biever et al., 2015a). Bertran-Gonzalez and colleagues showed a clear p-rpS6 240/244 signal preferentially expressed in ChIs, in contrast to a much weaker signal of p-rpS6 235/236 (Bertran-Gonzalez et al., 2012). Pharmacological modification of ChI firing leads to changes of p-rpS6 240/244 intensity in ChIs (Bertran-Gonzalez et al., 2012;Matamales et al., 2016b). To address regionality in AMPH modulation of ChI activity, we mapped p-rpS6 240/244 intensity in ChIs throughout the entire rostrocaudal axis of the Str after a single low-dose or high-dose of AMPH at two time points: 2.5 h postinjection (2.5h pi ) and 24 h postinjection (24h pi ). This revealed that AMPH induces a delayed regionally heterogeneous dose-dependent attenuation of ChI activity in the ventral Str/NAc.

Ethics
This research was performed in accordance with the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health, under a protocol approved by the Institutional Animal Care and Use Committee of New York State Psychiatric Institute (#NYSPI-1494).

Experimental animals
Mice were 129 Sv/C57BL6J mixed background, backcrossed to C57BL6J at least five times and kept inbred. Mice were group housed and maintained on a 12/12 h light/dark cycle with lights on at 7 A.M. in a temperaturecontrolled room with food and water provided ad libitum. The DAT-IRES-Cre/1;ROSA26-flox-STOP-CAG-ChR2-YFP double mutant strain (The Jackson Laboratory, RRID: IMSR_JAX:006660, RRID:IMSR_JAX:024109) were used, with the same genotype as previous studies (Chuhma et al., 2014(Chuhma et al., , 2018Mingote et al., 2015Mingote et al., , 2017. The presence of Cre is not essential for the present study; the IRES-cre transgene insertion in the DA transporter (DAT) locus modestly reduces DAT expression and AMPH responsiveness (Bäckman et al., 2006;Chohan et al., 2020).

Drug treatment
D-AMPH hemisulfate (Sigma-Aldrich, A5880) either low-dose (2 mg/kg) or high-dose (16 mg/kg) was dissolved in 0.9% NaCl immediately before use. Injections were done intraperitoneally at a volume of 10 ml/kg body weight.

Behavioral monitoring
Mice were habituated to handling for 2 d before the drug administration. Monitoring took place under bright ambient light conditions during the light phase. On the injection day, mice were placed in the open field, equipped with infrared motion detectors (Plexiglas activity chambers, 40.6 cm long Â 40.6 cm wide Â 38.1 cm high; SmartFrame Open Field System, Kinder Scientific) for 1 h for habituation. Baseline activity was monitored for 30 min preinjection, then mice were injected intraperitoneally either with saline, 2 or 16 mg/kg AMPH, and observed for a 2-h postinjection period. Locomotor activity was recorded automatically in 10-min bins. Stereotyped behaviors, orofacial stereotypy (mouth movements, lick, bite, self-gnaw, taffy pull, jaw tremor, yawn) and grooming, were scored for 1 min every 5 min as previously described (Kelley, 2001).
Raw 16-bit images were analyzed using Fiji/ImageJ (version 2.0.0., NIH, RRID:SCR_002285). Z-projected images were obtained by taking pixels with the maximum intensity in a stack. The outer boundary of the Str and its anatomic subregions, NAc core and shell, dorsomedial (DM) Str and dorsolateral (DL) Str, were manually delineated in accordance with the mouse brain atlas (Paxinos and Franklin, 2008), and their areas (mm 2 ) in each coronal section were obtained.
Particle analysis detected all ChAT-positive neurons in the Str and the total number of ChIs, perimeter (mm), area (mm 2 ), and circularity (a circularity value of 1 indicates a perfect circle while values approaching 0 indicate more elongated shapes) of each ChI were measured. Density of ChIs (neurons/mm 2 ) in each Str subregion was calculated as ChI number in a subregion divided by area of the subregion. For each coronal section, the ChAT image was superimposed on the p-rpS6 240/244 image, and the ChATpositive neurons were used as a mask for p-rpS6 240/244 intensity analysis. Fluorescence intensity of the corpus callosum was used for background subtraction.
Location of each ChI was defined by coordinates of the centroid. The p-rpS6 240/244 intensity of each ChI was normalized to the maximum and the minimum intensities for each cohort, 2.5h pi or 24h pi , and color-scaled. All color-scaled ChIs were 3D plotted with outlines of the Str using a customized script in MATLAB (MathWorks; RRID:SCR_ 001622) as previously described (Matamales et al., 2016a).
Distributions of p-rpS6 240/244 fluorescence intensity were standardized to the corresponding saline group for each time point and subregion by calculating z-scores: z = (x -m)/s , where x is the p-rpS6 240/244 signal in individual ChIs, m and s are the mean and the SD, respectively, of p-rpS6 240/244 signal in the corresponding saline group.
ChIs were identified visually by large soma size, confirmed by spontaneous firing, shallow resting membrane potentials (around 60 mV) and voltage sag by À400 pA current injection (700 ms in duration; Chuhma et al., 2014Chuhma et al., , 2018. Recording patch pipettes were fabricated from standard-wall borosilicate glass capillary with filament (World Precision Instruments). Pipette resistance was 4-9 MX and series resistance was 7-32 MX. Composition of intracellular solution was 135 mM K 1 -methane sulfonate (MeSO 4 ), 5 mM KCl, 2 mM MgCl 2 , 0.1 mM CaCl 2 , 10 mM HEPES, 1 mM EGTA, 2 mM ATP, and 0.1 mM GTP; pH 7.25. Recording was done with an Axopatch 200B amplifier (Molecular Devices) in fast current clamp mode. All recordings were done at 32-34°C (TC 344B Temperature Controller, Warner Instruments). No more than four cells were recorded per animal.
Data were filtered at 5 kHz using a four-pole Bessel filter, digitized at 5 kHz (Digidata 1550A, Molecular Devices) and recorded using pClamp 10 (Molecular Devices; RRID: SCR_011323). Electrophysiological data were analyzed with Axograph X (Axograph Science; RRID:SCR_014284). Firing frequencies were calculated as average frequency in a 2 s window obtained from 10 consecutive traces.

Statistical analysis
Sample sizes were determined using G*Power 3.1 with effect sizes based on similar experiments (G*Power, RRID: SCR_013726), setting a = 0.05 and power = 0.8 (Cunningham and McCrum-Gardner, 2007;Faul et al., 2007). For the immunocytochemistry experiments, we used Cohen's d = 0.97 as an effect size, resulting in 5 mice per group. For the electrophysiological experiments, we used Cohen's d = 0.32 as an effect size, resulting in 12 mice per group.
Statistical analyses were performed using Prism 8 (GraphPad Prism, RRID:SCR_002798) or SPSS 26 (SPSS; RRID:SCR_002865). p , 0.05 was considered as significant for all analyses. Data are presented as mean 6 SEM. Parametric tests were used here because datasets followed a normal distribution (D'Agostino-Pearson normality test, p . 0.05). ANOVA was used for comparison among conditions. Where significance was detected, multiple pairwise comparisons with Bonferroni correction were performed as post hoc tests.

Dose-dependent effects of AMPH on locomotor activity and stereotypy
Behavioral observations were used to confirm the dose-dependent effects of AMPH. Mice received a single low-dose (2 mg/kg) or high-dose (16 mg/kg) of AMPH, and their brains were extracted for analysis either after 2.5h pi , when acute behavioral effects had subsided, or at 24h pi to assess enduring effects on ChI activity in the Str. One low-dose AMPH-injected mouse, in the 2.5h pi cohort, was excluded from the study as its locomotor activity decreased after injection.

AMPH attenuation of ChI activity in vivo
We mapped AMPH effects on ChI activity using p-rpS6 240/244 as a reporter. Double immunostaining showed colocalization of ChAT and p-rpS6 240/244 (Fig. 4). P-rpS6 240/244 signal was also present in other Str cells, so ChAT staining was used to extract the signal specifically deriving from ChIs. We quantified p-rpS6 240/244 intensity as the average pixel intensity in each ChAT-positive neuron, in sections from the saline, low-dose and high-dose AMPH-injected mice, at 2.5h pi or 24h pi (n = 5 animals/ treatment, 10 hemisections/animal). Individual ChI locations were plotted in coronal hemisections of the Str and p-rpS6 240/244 intensities were color-scaled (Fig. 5A,C).
The two doses did not affect ChI p-rpS6 240/244 intensity in the dorsal subregions (Fig. 6B), nor was there any difference between the medial and lateral subregions. In the ventral subregions, the NAc core showed a similar profile of attenuation to the NAc shell. This medio-lateral concordance in the dorsal and ventral Str reinforces the differential effect in the ventral Str. The difference between the 2.5h pi and 24h pi cohorts reveals a time-dependent effect of a single dose of AMPH on ChI activity in the Str.

Spontaneous firing of ChIs is not affected by AMPH
To investigate possible mechanisms underlying the observed decrease in p-rpS6 240/244 , we recorded spontaneous firing of ChIs in slices in the four Str subregions after saline, low-dose or high-dose AMPH at 24h pi (Fig. 7A). ChIs were identified visually by large soma size, confirmed by spontaneous firing and voltage sag in response to hyperpolarizing-current injection (Fig. 7B), as described previously (Chuhma et al., 2014). Although firing frequencies of ChIs varied significantly among Str subregions, AMPH did not affect firing frequencies of ChIs in any Str subregion (two-way ANOVA; treatment effect, F (2,134) = 1.21, p = 0.30; location effect, F (3,134) = 13.30, p , 0.001; treatment Â location interaction, F (6,134) = 1.12, p = 0.36; Fig. 7C). Thus, neither low-nor high-dose AMPH affected the intrinsic firing of ChIs in the deafferented slice, at 2.5h pi or 24h pi , suggesting AMPH effects on ChI activity are because of extra-Str synaptic input.

Discussion
ChIs are principal targets of DA neurons and subject to regionally heterogeneous modulation. Here, we mapped the downstream effects of a single AMPH dose on ChI activity using p-rpS6 240/244 as a ChI-preferential activity-dependent marker. The single dose of AMPH did not affect the distribution, overall morphology, or spontaneous firing of ChIs in any Str subregion, arguing against neurotoxic effects of AMPH. While AMPH had no effect on in vivo ChI activity at 2.5h pi , it significantly attenuated ChI activity at 24h pi in the ventral Str/NAc. In the NAc, the attenuation in ChI activity after low-dose was greater than after highdose. In the dorsal Str, no significant difference in ChI activity was observed after either low-dose or high-dose AMPH. Thus, a single dose of AMPH has delayed regionally heterogeneous effects on ChI activity, with a dose-dependency in the NAc.

Distribution, morphology, and spontaneous firing of ChIs in the Str
In rodents (Gonzales and Smith, 2015), non-human primates (Brauer et al., 2000) and humans (Holt et al., 1996), the average size of ChIs in the NAc is smaller than in the dorsal Str. Here, we found that ChIs in the NAc core were significantly smaller and more elongated compared with those in other Str subregions, and that the morphology of ChIs soma (area, perimeter and circularity) differed among Str subregions. We also confirmed the differential distribution of ChIs in Str subregions (Gonzales and Smith, 2015). ChIs are denser in the NAc medial shell, as previously described in mice (Matamales et al., 2016a), rats (Phelps and Vaughn, 1986), and primates (Brauer et al., 2000).
A single injection of AMPH, either low-dose or highdose, did not affect ChI distribution or soma morphology in any Str subregion, at either time point, showing these doses were not neurotoxic. Although AMPH neurotoxicity on DA neurons has been known for some time (Wagner et al., 1980;Ricaurte et al., 1984;Ryan et al., 1990;Miller and O'Callaghan, 1996;Krasnova et al., 2001Krasnova et al., , 2005Granado et al., 2018), no study has focused on downstream neurotoxic effect on Str ChIs. To cause a significant toxic effect on ChIs, a higher dose of a more potent psychostimulant appears to be required; a single high- Figure 6. Comparison of p-rpS6 240/244 intensity in ChIs 2.5h pi and 24h pi AMPH. A, B, Comparison of ChI p-rpS6 240/244 intensity z-scores compared to the mean intensity in the corresponding saline-injected animals at 2.5h pi (A) and 24h pi (B) of saline (0 mg/kg), low-dose (2 mg/kg), or high-dose (16 mg/kg) AMPH in each Str subregion. Dots in bar graphs show the average measurements per animal; *p , 0.05. dose (30 mg/kg) of methamphetamine was found to induce a loss of 29% of ChIs in the dorsal Str (Zhu et al., 2006;Goodwin et al., 2009).
Although ChI spontaneous firing rates differed among Str subregions (Chuhma et al., 2014;Gonzales and Smith, 2015), a previous study found that a single dose of AMPH at 2.5h pi did not affect intrinsic firing of ChIs in any Str subregion (Chuhma et al., 2014). Here, we have found that a single dose of AMPH at 24h pi , either low-dose or high-dose, did not affect the spontaneous firing of ChIs in the slice, arguing that the effects of AMPH on ChI activity, measured with p-rpS6 240/244 at 24h pi , involve extrinsic synaptic input to the Str.
Single dose of AMPH affects ChI activity P-rpS6 240/244 signal reports the integrated activity and p-rpS6 240/244 intensity changes appear to be detected 60 min after pharmacological or behavioral manipulations (Bertran-Gonzalez et al., 2012), suggesting that p-rpS6 240/244 is suitable to study ChI activity at 2.5h pi or later (Knight et al., 2012). Therefore, the lack of AMPH effect at 2.5h pi is not because of temporal limits of p-rpS6 240/244 measurement. Stress increases p-rpS6 240/244 intensity (Knight et al., 2012;Biever et al., 2015a), this may be reflected in the greater p-rpS6 240/244 intensity in the 2.5h pi compared with the 24h pi saline controls.
In the present study, p-rpS6 240/244 intensity in ChIs was not affected at 2.5h pi after low-dose or high-dose AMPH, while ChI activity modulation via DA neuron glutamatergic cotransmission is dose dependently attenuated after a single dose of AMPH 2.5h pi (Chuhma et al., 2014). This discrepancy could be because of differences in the measurements; p-rpS6 240/244 reflects the tonic in vivo activity of ChIs, which also receive cortical and thalamic glutamatergic inputs in addition to DA neuron inputs (Lim et al., 2014), in contrast to the short phasic firing control of ChIs by DA neuron synaptic inputs.
Psychostimulants, including cocaine, methamphetamine and AMPH, are associated with an overall downregulation of DA transmission, both DA release and D2 receptor levels (Ashok et al., 2017). So, we should have expected an increase in ChI activity because of the loss of D2 receptor inhibition. In contrast, the attenuation of ChI activity at 24h pi argues for polysynaptic effects extending beyond direct effects on DA neuron presynaptic terminals. Indeed, AMPH-induced DA release has an onset of minutes and lasts for about 1 h in rodents (Sulzer, 2011), in parallel with behavioral activation. Tonic attenuation of cortical or thalamic glutamatergic inputs may be caused by polysynaptic modulation, resulting in delayed attenuation of ChI activity. Since AMPH does not affect p-rpS6 240/244 levels or protein synthesis in the Str within 2 h following injection (Rapanelli et al., 2014;Biever et al., 2015b), 2.5 h does not appear to be sufficient to cause long-term circuit changes.
Polysynaptic mechanisms that could contribute to observed decreases in ChI activity in the ventral Str/NAc may involve AMPH effects on other neurotransmitters besides DA. Glutamate efflux in the ventral tegmental area (VTA) is affected by AMPH administration, although both an increase (Xue et al., 1996) and a decrease (Wolf and Xue, 1998) of glutamate efflux have been observed. Acute AMPH exposure induces attenuation of excitatory glutamatergic synaptic transmission in the VTA by activation of serotonin receptors (Jones and Kauer, 1999). AMPH also indirectly affects DA release by stimulating the trace amine-associated receptors (TAAR1) expressed in DA neuron presynaptic terminals (Underhill et al., 2021).

ChIs in psychostimulant-induced changes
In the present study, low-dose AMPH significantly attenuated ChI activity in the ventral Str/NAc, a crucial site of psychostimulant action (Russo et al., 2010;Sulzer, 2011). DA neurons projecting to the ventral Str/NAc that corelease glutamate Stuber et al., 2010) can drive burst firing in ChIs (Chuhma et al., 2014;Mingote et al., 2019). A single dose of AMPH attenuates glutamate cotransmission (Chuhma et al., 2014), and mice with conditional reduction in glutamate cotransmission show an attenuated sensitization to repeated AMPH (Mingote et al., 2017). Similarly, we found here that AMPH attenuated ChI activity at 24h pi only in the ventral Str/NAc, suggesting that DA neuron glutamate cotransmission may be one of the factors responsible for NAc-selective attenuation of ChIs by low-dose AMPH, in addition to attenuation of phasic firing control through direct synaptic connections of DA neurons.
Although psychostimulant addiction involves repeated use, a single dose of AMPH can induce enduring Str Figure 7. Spontaneous ChI firing 24h pi AMPH. A, Whole-cell recordings were made from ChIs in the four Str subregions. B, An example of ChI firing recorded in the DL Str shows the characteristic spontaneous firing (black trace), and the prominent sag in response to hyperpolarizing-current injection (gray trace). C, Spontaneous firing frequencies of ChIs in each Str subregion are shown after saline (0 mg/kg, n = 30 animals), low-dose (2 mg/kg, n = 22 animals), or high-dose (16 mg/kg, n = 28 animals) AMPH at 24h pi . Dots in bar graphs show measurements for individual animals; the numbers of ChIs recorded were 12-13 cells/Str subregion/treatment; ***p , 0.001. circuit changes, drug-dependent behavior and negative affective states, such as anhedonia, depression and anxiety (Vanderschuren et al., 1999;Koob and Le Moal, 2001;Xia et al., 2008;Kameda et al., 2011;Li et al., 2017;Jing et al., 2018;Rincón-Cortés et al., 2018;Jayanthi et al., 2020). Interestingly, even a single dose of AMPH has been found to induce behavioral and neurochemical sensitization, which appears to increase over weeks (Robinson, 1984;Vanderschuren et al., 1999). Our results, in line with these previous findings, point to the relevance of a single dose of AMPH for elucidating drug-induced plasticity. Enduring alterations in ChI activity following acute AMPH exposure point to ChIs as a key component of drug-induced plasticity in the Str circuitry. Further studies using mice with restricted expression of opsins in ChAT neurons will be required to explore whether this reduction in NAc ChI activity is important in subsequent drug-dependent behavior.