Inhibition of striatal indirect pathway during second postnatal week leads to long-lasting deficits in motivated behavior

Abstract

Schizophrenia is a neuropsychiatric disorder with postulated neurodevelopmental etiology. Genetic and imaging studies have shown enhanced dopamine and D2 receptor occupancy in the striatum of patients with schizophrenia. However, whether alterations in postnatal striatal dopamine can lead to long-lasting changes in brain function and behavior is still unclear. Here, we approximated striatal D2R hyperfunction in mice via designer receptor-mediated activation of inhibitory Gi-protein signaling during a defined postnatal time window. We found that G_i-mediated inhibition of the indirect pathway (IP) during postnatal days 8–15 led to long-lasting decreases in locomotor activity and motivated behavior measured in the adult animal. In vivo photometry further showed that the motivational deficit was associated with an attenuated adaptation of outcome-evoked dopamine levels to changes in effort requirements. These data establish a sensitive time window of D2R-regulated striatal development with long-lasting impacts on neuronal function and behavior.

Introduction

Several neuropsychiatric disorders including schizophrenia are thought to have a neurodevelopmental origin but the understanding of neurodevelopment in the context of disease is still limited. Schizophrenia is defined by positive, cognitive and negative symptoms. Decreased motivation is a negative symptom that causes significant functional impairments and generally does not respond to antipsychotic medication [1, 2]. In this context avolition is thought to be a core symptom as it affects other negative symptoms with distinct neurobiological mechanisms. It thus may provide a potential target for improved treatment strategies [3, 4]. An important aspect of motivation is to shape behavior towards outcomes by evaluating the associated benefit (value) and cost (work) [5] which is thought to be impaired in schizophrenia [3, 4]. Genetic and environmental studies suggest a neurodevelopmental origin of schizophrenia since many schizophrenia-associated genes affect brain development. Moreover, external risk factors are present during embryonic and postnatal development [6–9].

There is strong evidence linking abnormalities in striatal dopamine (DA) and D2 receptors (D2R) to schizophrenia. All effective antipsychotic treatments are D2Rs blockers [10, 11] and PET brain imaging studies consistently demonstrate increased striatal DA release and D2R occupancy in schizophrenia to be associated with positive symptoms [12–14]. In contrast, low ventral striatal DA has been found to correlate with negative symptoms including apathy [15]. Importantly, (18)F-DOPA PET imaging found elevated striatal DA release capacity early on in adolescent subjects with high risk for schizophrenia [16]. However, the origin of these early abnormalities is unknown. They could be a secondary consequence e.g. of prefrontal hypofunction or due to intrinsic changes in the dopamine system [17]. It is also unknown whether and how early changes in the DA system occur in patients before adolescence. Drd2 and other DA-related genes contain risk alleles that could affect brain development early on [18]. In rodents, medium spiny neurons (MSNs) of the striatum undergo a prolonged maturational process, in which hyperexcitable immature MSNs acquire their mature morphology, firing properties and synaptic input by the end of the fourth postnatal week [19, 20]. This process is triggered by the gradual rise in DA neurotransmission to MSNs and is dependent on Kir2 currents [21]. Also, early postnatal synaptic activity of maturing MSNs was shown to regulate corticostriatal synaptogenesis in a process likely to be influenced by DA signaling [22, 23].

DA neurotransmission in the Nucleus Accumbens (NAc) has long been associated with effort-related processes, as demonstrated by classical pharmacological and DA depletion studies in rodents that lead to reduced operant responding in high-effort tasks, shifting choice preference to low-effort/low-reward options [24–26]. More specifically, the striatal indirect pathway has been implicated in the regulation of effort exertion via its inhibitory action on the ventral pallidum (VP) [27, 28]. D2R overexpression studies in mice point to distinct motivational effects depending on whether it occurs developmentally or in adulthood [29, 30]. Virally induced D2R overexpression by indirect pathway MSNs (iMSNs) in adult mice enhanced their progressive ratio (PR) performance [31, 32]. This was mediated by decreased iMSN inhibitory output to the VP [31]. However, the opposite effect on motivation (reduced PR performance) occurred when D2R overexpression started from late embryogenesis throughout postnatal development (D2-OE_dev mice) [33, 34]. In D2-OE_dev adult mice, NAc DA release and Ventral Tegmental Area (VTA) neuronal activity were both reduced, suggesting that long-lasting effects on the DA system may underlie the motivational impairment [35–38]. These studies suggest an importance of the developing indirect pathway in the maturation of the adult DA system. Here, we addressed whether inhibition of the indirect pathway during a defined postnatal time window can lead to lasting changes in behavior and DA function.

To selectively target iMSNs during a defined postnatal time window, we expressed hM4Di in the indirect pathway of neonates. Like D2Rs, hM4Di is a Gα_i- and arrestin-coupled GPCR that reduces neuronal excitability and inhibits synaptic transmission when activated by a synthetic ligand [39]. Using this approach, Kozorovitskiy et al. inhibited the indirect pathway from postnatal day P8 to P15. This manipulation strengthened corticostriatal synaptic inputs, which was still detected 10 days later at P25 [22]. Here we show that developmental inhibition of iMSNs on P8–15 led to long-lasting deficits in Open Field (OF) activity and motivational performance in adult mice. Moreover, using fiber photometry in combination with a genetically encoded DA sensor we found that NAc DA release does not adapt as readily to changes in effort requirements. This establishes P8–15 as a sensitive developmental window for striatal and DA circuit development that shapes adult-motivated behavior.

Materials and methods

Ethics compliance statement

All experimental procedures were approved by Institutional Animal Care and Use Committees of the New York State Psychiatric Institute (Protocol #1621).

Animals

For all experiments, Homozygous Adora2a-Cre/C57Bl6 were crossed to C57Bl6 to obtain heterozygous Adora2a-Cre mice (RRID:MMRRC_036158-UCD). Mice were housed 1–5 per cage on a 12 h light/dark cycle in a temperature-controlled environment (72 °F, humidity 30–70%), with food/water ad libitum, unless otherwise noted. All experiments were conducted in the light cycle.

Surgical procedures

P1 male or female heterozygous Adora2a-Cre mouse neonates were injected bilaterally into the striatum with the Cre-recombinase-dependent viral vectors AAV5-hSyn-DIO-hM4D-mCherry (Addgene Cat. No. 44362-AAV5, 7 × 10¹² vg/mL, Cambridge, MA) or AAV5-hSyn-DIO-GFP (Addgene Cat. No. 50457-AAV5, 7 × 10¹² vg/mL). Injection procedures and coordinates are described in [40] and (Supplementary Methods).

DREADD (hM4DGi) developmental activation

For the first cohort used (Figs. 1 and 2), clozapine-n-oxide (CNO) dissolved in sterile 0.9% saline (1 mg/kg) was injected intraperitoneally from P8 to P15 twice daily. Sterile 0.9% saline was given to littermate controls. Mice expressing GFP as a control received identical treatment as hM4D-expressing mice [31, 41]. For a second cohort of mice the DREADD agonist JHU37160 (“J60”) was injected twice daily i.p. in 0.9% sterile saline at 0.1 mg/kg (as in [42]).

Fig. 1. Selective neonatal inhibition of striatal indirect pathway leads to reduced Open Field locomotion.

A Experimental timeline from stereotaxic viral hM4D (or GFP) injection, DREADD agonist treatment (CNO or J60), to behavioral assessments. Representative confocal images of a sagittal section of a P8 (B) and P15 (C) A2a-Cre/hM4D_dev mouse showing expression of hM4D-mCherry in the striatum (Hpc: Hippocampus). Distance traveled by P21 (D) and P90 (E) old A2a/hM4D_dev mice after developmental injection of CNO or saline from P8–15, respectively. P21: repeated measures 2-way ANOVA, n = 15, drug factor [F (1, 28) = 1.187, p = 0.2852], time factor [F (6.575, 184.1) = 12.47, p < 0.0001], drug × time interaction [F (11, 308) = 0.9838, p = 0.4610], P90: *(RM 2-way ANOVA, n = 15, treatment factor [F (1, 28) = 4.225, p = 0.0493], time factor [F (5.377, 150.6) = 32.52, p < 0.0001], treatment × time interaction [F (11, 308) = 0.7023, p = 0.7363]). Distance travelled by P21 (F) and P90 (G) old A2a/hM4D_dev mice after developmental injection of J60 or saline from P8–15, respectively. P21: (RM 2-way ANOVA, n = 10–11, time factor [F (3.815, 72.48) = 6.406, p = 0.0002], treatment factor [F (1, 19) = 1.295, p = 0.2692], treatment × time interaction [F (11, 209) = 0.7808, p = 0.6591]), P90: (RM 2-way ANOVA, n = 10–11, time factor [F (4.635, 88.07) = 17.44, p < 0.0001], treatment factor [F (1, 19) = 4.797, p = 0.0412]*, treatment × time interaction [F (11, 209) = 1.036, p = 0.4157]).

Fig. 2. Developmentally inhibited A2a-Cre/hM4D_dev mice display reduced willingness to work for food.

A Schematic of progressive ratio (PR) trial structure (ITI inter-trial interval). Sessions started with a ratio of two presses/reward and doubled after each completed trial. Sessions were timed out after 2 h or if mice did not press for 5 min. B–F PR results for CNO-treated A2a/hM4D_dev mice, n = 9–16. B Session breakpoints (earned rewards) for saline- and CNO-treated mice. Data expressed as median ± interquartile range, n = 9–16, 2-tailed unpaired Mann-Whitney test, U = 24.5, **p = 0.0042. C Summary of total presses per session for saline (982.8 ± 130.1 presses) and CNO-treated mice (575.1 ± 61.85 presses) expressed as mean ± SEM, 2-tailed unpaired t-test, n = 9–16, t = 3.210, df = 23, *p = 0.0039. D Survival analysis of percentage (%) of mice engaged in task as a function of session duration. CNO-treated mice: median survival = 56.04 min, controls: median survival = 113.8 min, Mantel–Cox Log-rank test, χ² = 5.139, df=1, p = 0.0234. E Mean (±SEM) press rate (s^–1) as a function of ratio, (Mixed-effects Analysis, n = 9–16, Ratio Fixed Effect [F (2.552, 47.53) = 22.75, p < 0.0001], Treatment Fixed Effect [F (1, 20) = 1.469, p = 0.2397]). F Mean (±SEM) latency to reward (s) as function of ratio, (Mixed-effects analysis, n = 9–16, Ratio Effect [F (2.392, 47.24) = 1.333), p = 0.2750], Treatment Fixed Effect [F (1, 23) = 0.2870, p = 0.5973], G–K PR results for J60-treated A2a/hM4D_dev mice. G Effort breakpoints (completed ratio) for saline- and CNO-treated mice. Controls: median rewards = 9, or ratio of 512, J60-treated group: median rewards = 8, or ratio of 256, 2-tailed unpaired Mann–Whitney test, n = 11–12, U = 28, p = 0.0104. H Total presses per session for saline: 830.4 ± 80.6 and CNO-treated mice: 1278 ± 126.2 presses, 2-tailed unpaired t-test, n = 11–12, t = 3.038, df = 21, p = 0.0062. I Survival analysis of % of mice engaged in task as a function of session duration. (Mantel Cox Log-rank test, χ² = 5.934, df=1, p = 0.0149). J Mean (±SEM) press rate (s⁻¹) as a function of ratio, n = 11–12, (Mixed-effects Analysis, n = 10–11, Ratio Fixed Effect [F (3.355, 59.13) = 7.003, p = 0.0003], Treatment Fixed Effect [F (1, 20) = 0.6637, p = 0.4248]). K Latency to reward: Mixed-effects Analysis, 10–11, Ratio Fixed Effect [F (3.297, 55.63) = 1.076, p = 0.3705], Treatment Fixed Effect [F (1, 20) = 0.3699, p = 0.5499], Ratio × Treatment [F (8, 135) = 1.243, p = 0.2792].

Behavioral training

Progressive ratio (PR)

After completion of dipper and lever press training (see Supplementary Methods) mice were trained on pseudo-random interval (RI) reinforcement schedules, in which a lever press was rewarded after a variable time elapsed since the previous reward. Over successive days, the mean intervals were 5, 10, 15, and 20 s (RI5, RI10, RI15, RI20). RI20 was repeated for 3 days, followed by PR the next day. In PR the press-to-reward ratio doubled after each trial (starting with two presses/reward) until the session ended at 120 min or terminated after 5 min of no presses [31].

Fixed ratio (FR) with fiber photometry

Cannula-implanted adult mice expressing dLight1.2 were food restricted to 90% of baseline weight and trained on the operant chamber while cable-tethered. Three levels of fixed ratio (FR) were tested: FR10, FR20, FR40, in which rewards were delivered after 10, 20, or 40 presses per trial (fixed ITI of 10 s). Each FR level (45 trials, 60 min) was repeated for 3 days (FR “block”) and data were pooled.

Stepping ratio (SR) with fiber photometry

Cannula-implanted adult mice expressing dLight1.2 were food restricted to 90% of baseline weight and trained on the operant chamber while cable-tethered, followed by RI schedules as above. Next, mice were exposed to a stepwise PR schedule (stepping ratio, SR), where each ratio (25, 50, 100, 200, and 400) was repeated for 10 trials (variable ITI with mean 20 s, maximum 90 min/session).

In vivo fiber photometry (FP): was performed as described in [43] and (Supplementary Methods).

Statistical analysis and economic demand equation: see Supplementary Methods.

Results

Indirect pathway inhibition during the second postnatal week decreases locomotor activity in adult mice

We first determined whether iMSN inhibition from P8–15 decreased locomotor activity as observed in developmental D2R-OE mice [33]. Mice were tested as juveniles (P21) and adults (P90) (Fig. 1A). The inhibitory DREADD, hM4Di, was virally expressed in the striatum of Adora2a-Cre neonates using a custom neonatal stereotaxic adaptor (as in [40]). P1 injection leads to robust expression at P7 (Fig. 1B) with a further increase in expression by P15 (Fig. 1C). We estimated the transfection efficacy in P15 mice at 44.49 ± 6.46% (mean ± SEM) (Fig. S1). Half of the mice were injected with the DREADD agonist clozapine-n-oxide (CNO; 1 mg/kg) and half with saline, twice daily. At P21, CNO-treated mice did not display differences in locomotion (distance traveled) compared to saline controls (Fig. 1D). At P90, an overall decrease in locomotion emerged (Fig. 1E). To control for non-specific effects of CNO due to clozapine conversion [44], a separate cohort expressed a Cre-dependent GFP virus and was treated with CNO or saline (P8–15). RM 2-way ANOVA analyses found no group differences in OF distance traveled at either P21 or P90 (Fig. S2A, B).

We attempted to replicate these results using the recently developed DREADD agonist, JHU37160 (J60), that is not metabolized to clozapine [42]. J60 (0.1 mg/kg) was i.p. injected twice daily (P8–15). Also, the J60 cohort showed decreased locomotor activity at P90 (Fig. 1G) but not at P21 (Fig. 1F). A separate GFP-expressing cohort treated with J60 or saline (P8–15) showed no group differences at P90 (Fig. S2C, D). Together, these results show that P8–15 inhibition of the indirect pathway decreased locomotor activity in the adult. No sex × treatment interaction was measured on OF performance (data not shown).

Indirect pathway inhibition during the second week decreases motivation in adult A2a-Cre/hM4D_dev mice

To assess whether the developmental inhibition leads to long-lasting changes in motivated behavior, we tested A2a/hM4D_dev mice in a PR schedule at age P110. First, mice were trained to press a lever for reward (evaporated milk) followed by a series of increasing random interval schedules as described in the methods. On the PR day, the press-to-reward ratio doubled after each trial, starting with 2 presses/reward and quickly reaching 2048 presses/reward after 11 trials. (Fig. 2A). The number of rewards reached by each animal was defined as the breakpoint.

Compared to controls developmentally inhibited (CNO) mice displayed a reduced median breakpoint (Fig. 2B). CNO-treated mice also showed significantly lower total lever presses (Fig. 2C) and shorter sessions (34.2% shorter) than control mice (data not shown: CNO: 66.17 ± 7.91 min, n = 9; SAL: 100.5 ± 8.6 min, n = 16; 2-way unpaired t-test, t = 2.811, df = 22, p = 0.0102). A session survival analysis confirmed that the CNO-treated mice abandoned the task earlier than controls (Fig. 2D). The press rates by both groups were sensitive to required effort in inverted “U-shaped” response curves, peaking at ratio 16 (Fig. 2E). A Mixed-effects analysis of the latency to reward revealed no effect of treatment or ratio requirements (Fig. 2F). To test for possible developmental effects of CNO alone, an independent cohort of A2a-Cre mice expressing GFP in the striatum was treated with CNO or saline and assessed as above (Fig. S3A–E). No significant group differences were detected between A2a/GFP_CNO and A2a/GFP_SAL mice (Fig. S3A–E).

Next, we evaluated the J60 cohort. Compared to saline controls, the J60-treated group showed a significantly reduced breakpoint (Fig. 2G). J60-treated mice displayed reduced overall pressing compared to controls (Fig. 2H) and a shorter session duration than controls (data not shown). The “survival” analysis confirmed that J60-treated mice disengaged from the task before controls (Fig. 2I). For both groups press rates varied in response to increasing ratios, but no statistical differences were detected between groups (Fig. 2J) and latency to reward was unaltered (Fig. 2K). J60 had no effects in A2a/GFP_dev control mice (Fig. S3F–J). We also fitted an exponential demand curve [45] finding an increased price sensitivity and decreased maximum point of demand in CNO and J60-treated vs Saline A2a/hM4D_dev mice (Fig. S4A–J). Together, the CNO and J60 results show that P8–15 inhibition of iMSNs impairs PR performance in adulthood. As in the OF experiments, we detected no sex x treatment interaction (data not shown).

Indirect pathway inhibition during the second week leads to reduced NAc dopamine release when effort requirements increase over days

In developmental D2R-OE mice decreased performance in the PR task is associated with decreased dopamine release in the NAc [36, 46]. We therefore measured NAc DA release in A2a/hM4D_dev mice. We expressed the genetically encoded DA sensor, dLight1.2, selectively in the NAc of A2a/hM4D_dev mice at P90, and inserted optic fibers for fiber photometry of task-evoked changes in DA (Fig. 3A) [47]. Task-evoked DA transients were obtained by aligning the fluorescent signal to lever extension or reward presentation (see Methods).

Fig. 3. Developmentally inhibited mice show effort-sensitive reduction of task-evoked NAc DA release across several days.

A Experimental timeline from neonatal hM4D viral injection, developmental inhibition with J60 injections (P8–15), dLight1.2 viral injections and optic fiber implantation (P90), to simultaneous DA monitoring (fiber photometry) during operant tasks. B Schematic of operant fixed ratio (FR) task structure. Each FR block was repeated for three consecutive days (45 trials per session, 135 total trials). A fixed ITI (inter-trial interval) of 10 s was used. C Anatomical localization of optic fibers. D Representative coronal section of viral dLight1.2 expression in striatum (anti-GFP immunohistochemistry) and optic fiber location (dashed lines delineate fiber shaft). E–I Results of DA release aligned to lever extension (Lever ON; time = 0). E–G Mean (±SEM) DA release (z scored dFF) aligned to lever extension during FR10, FR20, and FR40, respectively (n = 10). H Mean peak amplitude (±SEM) of DA release (z score dFF) across FR10, FR20, and FR40 shows declining peaks as work requirement increases for both groups (RM 2-way ANOVA, n = 10, ratio factor [F (1.682, 30.27) = 10.69, p = 0.0006], treatment factor [F (1, 18) = 3.43, p = 0.0805], ratio × treatment: (F (2, 36) = 0.2267, p = 0.7983)). I Mean (±SEM) latency to first press across FR10, FR20, and FR40 conditions showing both groups increasing latency with greater effort (RM 2-way ANOVA, n = 10, ratio factor [F (1.654, 31.42) = 33.84, p < 0.0001], treatment factor [F (1, 19) = 0.004594, p = 0.9467], Interaction [F (2, 38) = 0.4556, p = 0.6375]). J–N Results of DA release aligned to reinforcement delivery (Reward ON). J–L Mean (±SEM) DA release (z scored dFF) aligned to Reward ON during FR10, FR20, and FR40, respectively (n = 10). M Mean peak amplitude (±SEM) of DA release (z score dFF) across FR10, FR20, and FR40 shows increasing peaks as work requirement increases for both, but a blunted rise in peak amplitude for the J60 group (RM 2-way ANOVA, n = 10, FR block: F (1.159, 20.86) = 20.53, p = 0.0001; Treatment: F (1, 18) = 6.869, p = 0.0173; Interaction: F (2, 36) = 1.477, p = 0.2417). N Mean (±SEM) latency for rewarded head entries across FR blocks showing a mild positive relationship with effort, but no effect of treatment or interaction (RM 2-way ANOVA, n = 10, “Treatment” F (1, 19) = 3.109, p = 0.0939; “FR block” F (1.501, 28.53) = 4.321, p = 0.0323; “Interaction” F (F (2, 38) = 0.5149, p = 0.6017)).

Mice were trained in a series of FR experiments starting at 10 presses/reward (FR10), then FR20 and FR40. Each FR block repeated for 3 consecutive days, 45 trials each (Fig. 3B). At each trial initiation the lever is extended, eliciting a sharp fluorescent transient in both J60 and saline groups (Fig. 3E–G). With each ratio increase, the amplitude of these transients decreased, but there was no effect of treatment or interaction (Fig. 3H). We examined the latency to start pressing (latency to 1st press) as function of effort. Both groups displayed longer latencies with higher effort with no effect of treatment or interaction (Fig. 3I).

Next, we analyzed DA release aligned to reward presentation (“dipper up”). The “dipper”-aligned DA transient showed a biphasic structure, the first in response to dipper presentation, signaling reward availability, and a second transient, most likely related to reward consumption (Fig. 3J–L). The first DA transient aligned to the “dipper up” increased as a function of ratio (Fig. 3M). Both J60 and control groups exhibited an upward trajectory for the first DA transient, but this response appeared to be blunted in the J60 group (Fig. 3M). A RM 2-way ANOVA of the first-transient amplitudes across ratios was statistically significant for FR block (effort) and treatment with no interaction (Fig. 3M). Last, head entry latencies increased as a function of ratio with no significant effect of treatment (Fig. 3N). Overall, while in control mice peak DA release to a cue signaling reward availability is calibrated in response to increasing effort invested in the task, DA release is decreased, and the calibration seems to be blunted in A2a/hM4D_dev mice.

Indirect pathway inhibition during the second postnatal week leads to reduced NAc dopamine release when effort requirements increase within the same day

During PR, effort requirements increase exponentially with each trial within a day [48]. In contrast in the FR series above, effort increases over days and maximum presses per reward are much lower than during the PR schedule. We therefore attempted to measure DA during an effort-based task that approximates PR conditions. Since the measurement of DA transients benefits from multiple trials for signal averaging, we designed a compromise between changing FRs over days and the PR schedule where effort changes on each subsequent trial within a day. We designed a stepping ratio schedule (SR), where mice start at 25 presses/reward and repeat each ratio for 10 trials (“ratio step”) before advancing to the next ratio with double the requirement (Fig. 4A, B).

Fig. 4. Developmentally inhibited mice have less adaptable reward-aligned DA release in response to rapidly increasing effort requirement.

A Stepping ratio (SR) session structure. Each ratio was repeated for ten trials before advancing to the next ratio. B Schematic of SR trial structure. C, D Mean (±SEM) DA release (z scored dFF) aligned to lever ON (time 0 s) across ratios 25–400 for saline and J60 groups, respectively. E Mean (±SEM) peak DA release as a function of effort requirement, showing an overall effect of effort but not of treatment or interaction (RM 2-way ANOVA, n = 5–8, ratio factor [F (2.614, 28.76) = 3.742, p = 0.0263], treatment factor [F (1, 11) = 0.3206, p = 0.5826], interaction [F (4, 44) = 1.155, p = 0.3435]), F Mean (±SEM) latency to first press as function of increasing effort showing an inverted U overall relationship curve, but no statistical significance. G, H Mean (±SEM) DA release (z scored dFF) aligned to reward-delivery (Reward ON = time 0 s) across ratios 25 through 400 for saline (n = 5) and J60 (n = 8) groups, respectively. I Mean (±SEM) peak DA release as function of ratio, showing blunted DA responses as effort increases (RM 2-way ANOVA, n = 5–8, ratio [F (1.814, 19.95) = 39.61, p < 0.0001], treatment [F (1, 11) = 1.004, p = 0.3378], interaction [F (4, 44) = 3.375, p = 0.0171]), J Mean (±SEM) latency to head entry at each ratio; RM 2-way ANOVA did not show statistical significance for ratio, treatment or interaction. (RM 2-way ANOVA, n = 5–8, ratio [F (2.116, 23.27) = 2.371, p = 0.1131], treatment [F (1, 11) = 0.1504, p = 0.7051], interaction [F (4, 44) = 0.5798, p = 0.6788]).

DA responses aligned to lever extension (“Lever ON”) by both J60 and control mice varied significantly with the increasing ratios from 25 to 400 (Fig. 4C, D), but no overall treatment or interaction effects were detected (Fig. 4E). The latency to start pressing was neither affected by ratio nor by treatment (Fig. 4F). We then analyzed DA responses aligned to dipper elevation (“Reward ON”, Fig. 4G, H). As in Fig. 3 the signal had a “double peak” structure and the dipper-evoked DA transients increased with higher ratios, suggesting that DA release adapts to rapidly changing effort requirements. This adaptation or interaction between ratio and treatment was significantly attenuated in J60-treated mice (Fig. 4I). Head entry latencies seemed to be modulated across ratios for both groups, but no statistical significance was found (Fig. 4J).

Attenuated correlations between DA release at reward outcome and behavioral outcome measures

We analyzed the relationship between “dipper ON”, the reward availability cue, and head entry latencies throughout the entire session (all ratios). Control mice displayed a significant negative correlation, the higher the DA peak, the shorter the latency (Fig. 5A though not observed within highest ratio Fig. 5C), and this relationship was disrupted in J60-treated A2a/hM4D_dev mice (Fig. 5A). In line with this disruption 60% of control but only 25% of J60 mice showed a significant correlation (Fig. 5E, G as examples). We further observed that the time to complete each trial (latency between lever extension to dipper extension) correlates with DA release at dipper ON (Fig. 5B). This correlation was weaker in J60 mice (Fig. 5B) and could be measured during highest ratio in control but not J60 mice (for highest ratio: Fig. 5D). All J60 and control mice showed the correlation at the individual animal level (Fig. 5F, H as examples). Together, the correlational analysis supports the hypothesis that after developmental inhibition DA release associated with reward availability does not adapt as well to increasing effort or time requirements. Moreover, it is decoupled from the following head entry response.

Fig. 5. Correlation analyses show loss of positive relationship between reward-aligned DA and time to complete ratios in J60-treated A2a/hM4D_dev mice.

A Scatter plots for head entry latency and reward-aligned peak DA transients (“Dipper Peak”) for saline (blue) and J60 (red) groups including all ratios (25 through 400). Saline: Spearman r = -0.3499, p < 0.0001, J60: Spearman r = 0.0307, p = 0.5557. Straight lines represent linear regressions. B Scatter plots for time to complete ratios and reward-aligned peak DA transients (“Dipper Peak”) for saline and J60 groups including all ratios (25 through 400). Saline: Spearman r = 0.7538, p = 0.0001, J60: Spearman r = 0.4878, p = 0.0001. An F test was used to compare the two linear regression fits [4.413 (2, 602), *p = 0.0125]. C Scatter plots for head entry latency and reward-aligned peak DA transients for saline and J60 groups including ratio 400 only. Saline: Spearman r = 0.1728, p = 0.2798, J60: Spearman r = 0.0951, p = 0.4586. D Scatter plots for time to complete ratios and reward-aligned peak DA transient for saline and J60 groups including ratio 400 only. Saline: Spearman r = 0.5262, p = 0.0005, J60: Spearman r = 0.1340, p = 0.2911. E Scatter plot for head entry latency and reward-aligned peak DA transient from a representative individual saline-treated A2a/hhM4D_dev mouse (ID #2390) including all ratios. Spearman r = −0.3353, p = 0.0173. F Scatter plot for time to complete ratios and reward-aligned peak DA transient from mouse 2390—all ratios included. Spearman r = 0.7826, p < 0.0001. G Scatter plot for head entry latency and reward-aligned peak DA transient from a representative individual J60-treated A2a/hhM4D_dev mouse (ID #3813) including all ratios. Spearman r = −0.1477, p = 0.3060. H Scatter plot for time to complete ratios and reward-aligned peak DA transient from mouse 3813—all ratios included. Spearman r = 0.2850, p < 0.0449.

Discussion

We inhibited striatal iMSNs of A2a-Cre during the second postnatal week and assessed adult-motivated behaviors and NAc DA neurotransmission. Developmentally inhibited A2a/hM4D_dev mice exhibited reduced locomotor activity, but only in adulthood, suggesting that the long-term effects of the manipulation do not emerge until neurocircuits reach maturity. Moreover, J60-A2a/hM4D_dev mice showed impaired motivation under a PR schedule. Our analyses did not detect differences in press rate and latency to reach reward, suggesting that lower PR performance does not reflect a general motor impairment. Rather, our results suggest that J60-A2a/hM4D_dev mice have a higher sensitivity to rising costs dropping out earlier with increasing costs. Alternatively, this difference in PR performance could be attributed to an overall limit of effort spent (number of presses) in a session. However, in the FR40 task (total possible output of 1800 presses), both groups completed the task, suggesting that total effort spent does not explain differences in PR performance where the total output was below 1300 presses.

We first examined the effects of developmental iMSN inhibition on adult DA release in the FR block series. Lever extension at the trial start serves as a reward-predicting cue (contingent on pressing) induced large and sharp DA transients [43, 49]. DA release decreased as the FR increased for all groups. Conversely, when DA signal was aligned to dipper elevation, signaling reward availability, this relationship reversed and DA release increased along with rising ratios. These data confirm previous observations that cue-evoked and outcome-evoked DA in the NAc adapts with increasing effort requirements encoding upcoming and experienced effort or time cost [49]. While there was no significant difference between developmental inhibited and control mice on the lever induced DA release, dipper-induced DA release was decreased in developmentally inhibited A2a/hM4D_dev mice. This suggests that in developmentally inhibited mice NAc DA adapts less to spent effort or time.

In the stepping ratio (SR) task, mice experienced a steep rise in effort requirements within a session. In contrast to the PR task, we did not observe any behavioral deficit in the SR task (data not shown). This could be due to the tethering of the mice or because ratio requirements may have been too low to induce differences between the groups. As in the FR series, DA transients aligned to reward-delivery cue increased with increasing ratios and this adaption was blunted in developmentally inhibited mice.

The adaptation of DA release aligned to reward-delivery again suggests that it may be impacted by the spent physical effort [50]. In our PR, FR, or SR experiments, the explicit manipulation was effort as defined by the number of lever presses needed for reward-delivery. Time delay was not independently tested as another component of cost. Therefore, time to complete trials may contribute to the observed change in reward-delivery DA release consistent with [51]. Thus, with increasing time to complete the ratio the expectation of reward decreases leading to a positive prediction error that is reflected by increased dopamine at reward outcome. This adaptation is blunted in the developmental inhibited mice suggesting a deficit in the ability to encode a temporal difference-reward prediction error. Our correlational analysis in the SR task further supports the impaired adaptation to time since the time to complete each trial correlated with the outcome-induced dopamine release a relationship that was weaker in developmentally inhibited mice.

Our analysis showed that A2a/hM4D_dev mice do not display a general deficit of DA release, but that DA is released less only after high-effort spent. This contrasts with the findings of the original D2-OE_dev model of striatal D2R overexpression where DA release in the NAc was generally reduced [36, 46]. A decrease in VTA DA neuron firing and NMDA receptor subunit expression in mesolimbic DA neurons observed in anesthetized mice and post-mortem tissue pointed to a general intrinsic deficit in mesolimbic DA function [35]. However, Duvarci and colleagues also measured less recruitment of DA neurons in a cognitive working memory task and found that task parameters regulated DA neuron activity in D2-OE_dev mice, which suggested altered extrinsic regulation of DA neurons depending on the behavioral condition [46]. Since we did not analyze DA neuron activity or NMDA receptor expression in A2a/hM4D_dev mice, it is unclear whether the changes in DA release measured in the FR and SR schedules involve intrinsic, extrinsic or both mechanisms.

Postnatal developmental inhibition of direct and IPs (P8–15) has been shown to change excitatory corticostriatal inputs onto spiny projection neurons, indicating that recurrent circuit activity by both striatal projections via the cortex contribute to synaptogenesis [22]. While developmental inhibition of the direct pathway resulted in decreased frequency of excitatory drive onto and reduced dendritic spine density of striatal projection neurons on P15, the opposite was observed for inhibition of the indirect pathway. Moreover, inhibiting corticostriatal projections during the same period recapitulated the synaptic effects of inhibiting striatal projection neurons. This effect lasted up to 10 days after the manipulation (P25), suggesting that the circuit-level restructuring extends beyond the manipulation. Our results provide evidence that inhibiting the IP during the same developmental window is sufficient to decrease motivated behavior and OF locomotion well past the manipulation period, 75 days later in adult mice.

However, even if corticostriatal input strength is enhanced it is unclear how this may affect the downstream DA system since both direct and indirect pathways are equally affected. One possibility is that the postnatal IP inhibition is not affecting the DA system via recurrent cortical circuitry but by more direct disinhibition of dopamine neurons [52]. This early disinhibition may lead to an adaptive long-lasting decrease in DA neuron excitability, thereby reducing the ability of external inputs to activate DA neuron activity.

We did not observe behavioral sex differences in our experiments. However, our study was not powered to detect potential sex differences. In humans, early reports indicated an enhanced incidence of schizophrenia in males but more recent evidence suggests that the prevalence in males is due to an earlier age of incidence, which reverses with age [53–55]. Whether changes in basal ganglia circuitry during development are related to the higher early incidence in males remains to be studied.

PET imaging has shown that patients at risk for schizophrenia display enhanced DA release capacity during adolescence [16] and extensive DA function during adolescence has been discussed in the etiology of schizophrenia [56]. However, it is unclear how early abnormalities in the DA system arise in schizophrenia since both schizophrenia risk and changes in the DA system are difficult to assess in children. Human genetic studies identified a combination of moderate-risk polymorphisms in DA-related genes and disruptive early life events that contribute to the risk of schizophrenia [57, 58] and mutations in Drd2a gene have been associated to cortical thinning and anhedonia [59]. However, causal links between the genetic risk, neural development, and disease pathology remains to be shown. Here, we provide evidence that time-defined early postnatal Gi-activation of iMSNs, mimicking D2R hyperfunction, leads to long-term deficit in motivated behavior, a core negative symptom of schizophrenia [3], opening the possibility that early changes in the DA system may have long-lasting pathological consequences.

Author contributions

PO, PB, and CK conceived the experiments, PO, MS, and SM performed all experiments. AH developed MATLAB data analysis scripts. MG performed the economical modeling. RR developed protocol and analyzed striatal cell counting. PO and CK wrote the manuscript with inputs from PB.

Funding

This work has been supported by K08MH127379 to PO and R01MH093672, R01MH124858 to CK and MH068073 to PB. Additional support also provided by the Leon Levy Foundation Neuroscience Fellowship to PO.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41386-024-01997-x.