One of the prevailing themes of addiction research has been investigating the neural substrates of individual differences in the risk for substance abuse. Nevertheless, while many environmental, congenital, and genetic factors associated with the risk of developing a substance use disorder have been identified, little is known about how vulnerable populations differ from others that are relatively protected. Here, we examined whether innate differences in dopamine (DA) neurotransmission predict the onset and course of cocaine seeking and consumption.

DA neurotransmission is an important component of reward and reinforcement processes [1], and numerous studies suggest that DA may differ between populations that are vulnerable vs protected from developing severe substance use disorders. For example, females across species are more prone to heavy cocaine consumption, and female rats acquire cocaine self-administration more readily [2,3,4], escalate cocaine intake to a higher degree [5], express greater motivation for cocaine [2] and are more vulnerable to some forms of reinstatement of cocaine-seeking [6]. Female rats also demonstrate greater DA release and higher maximal rate of DA uptake (Vmax) in the dorsal striatum [7, 8], and female mice exhibit an estrus-dependent increase in DA release in the nucleus accumbens (NAc) in vitro [9]. Similarly, mice and rats reared in isolation as juveniles/adolescents consume more cocaine in adulthood [10, 11], work harder to obtain cocaine [12], and may be more prone to cocaine-primed reinstatement [11] than group housed controls (but see [12]) without an apparent difference in the cocaine dose-response curve [12]. Isolation-reared rats also demonstrate an increase in “sensitivity” to the uptake inhibition effects of cocaine (i.e., greater apparent cocaine potency) as well as greater DA release and uptake in the dorsomedial striatum (DMS) and NAc [13].

Great strides have also been made by measuring DA neurotransmission based on behavioral phenotypes thought to predict the risk of cocaine-associated behavior. For example, rats that demonstrate a lower locomotor response to cocaine develop cocaine-induced CPP more readily [14] and expend more effort to obtain cocaine [15]. These rats have greater DAT binding in both the NAc and dorsal striatum, as well as a reduced DA response to cocaine [16]. Rats that instead demonstrate a higher locomotor response to novelty—one of the most widely used predictors of vulnerability to cocaine intake—are more sensitive to the psychostimulant effects of cocaine [17,18,19], acquire cocaine self-administration more readily [4, 19, 20], and demonstrate a positive vertical shift in cocaine intake relative to those that are less reactive [20, 21], although they do not demonstrate the high measures of cocaine-seeking believed to model the more severe aspects of cocaine addiction [22]. These “High Responders” do not differ from “Low Responders” in basal extracellular DA [23, 24] or baseline measures of DA release or uptake [19] in the NAc, although they evince an enhanced DA response to cocaine [19, 23, 24], and have greater vesicular DA [25]. Additionally, while they exhibit fewer D2 receptors (D2R) in the NAc and striatum they do not differ in the behavioral response to the D2/D3 receptor agonist, quinpirole [26].

Higher impulsivity is similarly associated with a greater risk of cocaine self-administration. Highly impulsive female and male rats are more likely to escalate cocaine self-administration [27] and have higher cocaine intake than low impulsivity rats [27,28,29] (but see [21]), although female rats do not differ in motivation for cocaine [27]. Importantly, the relationship between impulsivity and cocaine self-administration has been hypothesized to be mediated by D2R function and expression [29], which itself has been hypothesized to underlie the risk of cocaine addiction ([30, 31] but see [32,33,34]; [35] for review).

Thus, although cocaine-vulnerable phenotypes have been associated with differences in DA neurotransmission, the nature of those differences is unclear and may be confounded by the relatively indirect means of segregating animals through, e.g., prior cocaine exposure or unavoidable stress. Nevertheless, existing observations demonstrate the clear need to directly probe the relationship between inherent variability in DA signaling and the risk and severity of cocaine use disorders. Here, we directly measured the relationship between individual differences in DA release and uptake prior to cocaine exposure and subsequent cocaine-associated behaviors using a combination of neurochemistry and multiple tests of cocaine reward and reinforcement.



Male Sprague-Dawley (Envigo) rats weighing 325–350 g on arrival were maintained in a vivarium with controlled temperature and humidity under a 12:12 h light:dark cycle and were provided ad libitum access to food and water. Studies were limited to males in this initial study, as cycling levels of estrogen and progesterone have been demonstrated to affect both DA neurotransmission [9, 36, 37] and cocaine-seeking [2, 38, 39]. The potential influence of diurnal variability [40, 41] was minimized by collecting baseline DA at the same approximate time of day. All experiments and procedures were conducted during the dark cycle and in accordance with the National Institutes of Health Guide for Care and Use of Laboratory Animals (2011) under the supervision of the Institutional Animal Care and Use Committee at Drexel University. Additional details of experimental procedures are provided in Supplementary Methods.

Baseline in vivo voltammetry

Rats were placed into a stereotaxic apparatus under isoflurane anesthesia [42], and a carbon fiber microelectrode was implanted into the DMS while a stimulating electrode was aimed at the ventral tegmental area/substantia nigra (VTA/SN) as previously described [43, 44]. Electrode positions were adjusted to optimize release in the DMS before the carbon fiber microelectrode was lowered into the NAc. Rats undergoing cocaine self-administration were also implanted with a jugular catheter at this time. Each rat’s baseline DA release and uptake measures were determined based on the recording with maximal DA release [43, 45]. It was not always possible to achieve sufficient DA release to assess the rate of DA uptake in both the DMS and the NAc, thus only one measure is available for some rats. See Supplementary Methods.

Conditioned place preference

Cocaine conditioned place preference (CPP) experiments were conducted using chambers consisting of three compartments (Med-Associates; Fairfax, VT). Baseline compartment preference was determined for each rat (n = 14) and baseline estimates of locomotor activity were measured using Ethovision Software (Noldus; Leesburg, VA). Measures of time spent in each chamber and baseline “exploratory behavior” (chamber entries) during habituation [46, 47] were hand-scored by an investigator blinded to baseline measures of DA release and uptake. One rat demonstrating ≥20% preference for one side of the apparatus during the habituation was removed from the study. CPP for 20 mg/kg cocaine (i.p.) in the remaining 13 rats was determined as previously described [48], see Supplementary Methods.

Self-administration training

Rats (n = 70) initiated cocaine use under an FR1 schedule of reinforcement for a 0.75 mg/kg infusion of cocaine. Self-administration training was conducted daily in the homecage in 2 h sessions or until 40 infusions were received. Most (n = 52) were trained with a single lever as previously described [44, 49,50,51]. Rats (n = 18) used for extinction and reinstatement experiments were trained with an inactive lever as is typical for this procedure [52]. Responses on the inactive lever were recorded but had no programmed consequences. Rats were considered to have reached acquisition criteria when cocaine-reinforced responses were ≥10 for three consecutive sessions [52] and were subsequently advanced through different experiments (see Fig. S1). Rats trained with a single lever were considered to have failed to acquire after 21 sessions, while those trained with two levers were given 30 sessions, as we hypothesized the addition of the second lever could increase the number of sessions required to reach acquisition criteria. Nevertheless, acquisition incidence (Fig. S2a) and rate (Fig. S2b) did not significantly differ between rats self-administering with single- or dual-levers. Rats that failed to meet acquisition criteria were removed from further study.

Cocaine intake

To measure cocaine intake under low-effort conditions, a subset of rats that reached acquisition criteria (n = 16 single lever, n = 12 dual lever) continued on short-access FR1 for a further 12-13 sessions for a total of 15-16 sessions after acquisition criteria was first met. Measures of intake are based on the average of the final three sessions. Four rats were removed, one due to loss of patency and three due to sudden weight loss (n = 3 single lever, n = 1 dual lever). No differences in the dose of cocaine obtained under FR1 were observed between single- and dual-lever trained rats (Fig. S2c), thus data from these animals were combined.

Within-session threshold

After establishing cocaine consumption under FR1, rats (n = 15) trained with a single lever underwent the within-session threshold procedure. Under this schedule, the dose of cocaine is initially high and is then decreased in a stepwise, quarter-logarithmic fashion every 10 min for 110 min (435.7, 245, 137.8, 77.5, 43.6, 24.5, 13.8, 7.7, 4.4, 2.4, 1.4 µg/inf) [44, 53]. This procedure provides a measure of both free-consumption (intake when cocaine is essentially free, Q0) and “motivation” to maintain preferred levels of cocaine (the maximal “price” paid in lever presses/mg cocaine, Pmax) [54]. Rats self-administered cocaine under this schedule until reaching stability, defined as 3 days with Pmax within 15% variability without ascending or descending trends. One rat failed to achieve stability and was removed from the study, as was one that lost patency. Q0 and demand elasticity (α) over these three days were determined by generating cocaine demand curves using an automated script in RStudio (kindly provided by Dr. Erik Oleson) that fits a least squares regression curve to the natural log(mg intake) by unit price of cocaine at each epoch. Q0 was defined as the y-intercept and α was defined as the slope of the curve [55]; average demand curves were generated from the averages of these coefficients. Consistent with previous work, the first “epoch” was excluded from the analysis [54, 56].


Rats that did not undergo extended FR1 (n = 16) instead self-administered cocaine under the progressive ratio (PR) schedule. Under this schedule, the number of responses required for each subsequent cocaine infusion (0.75 mg/kg) increased exponentially over 6 h sessions (1, 2, 4, 6, 9, 12, 15, 20, 25, 32, 40, 50, 62, 77, 95, 118…) [50, 57]. The total number of infusions received under this schedule was defined as the “breakpoint”. Rats self-administered cocaine until reaching stability, defined as 3 days with breakpoints within 15% variability without ascending or descending trends. See Supplementary Methods.

Extinction and cocaine-primed reinstatement

During extinction (n = 11), responses on the active lever were recorded but had no programmed consequence. Extinction responding is defined as the number of responses on the active lever on Day 1 [52]. Reinstatement responding was measured as the number of lever presses on the active lever following an i.p. injection of cocaine. Two rats were removed prior to cocaine-primed reinstatement, one due to sudden weight loss and one due to experimenter error. All rats were first challenged with i.p. saline to mitigate the influence of stress upon receiving an injection for cocaine-primed reinstatement. Rats were then allowed to re-normalize to extinction levels of responding and were challenged again with 10 mg/kg i.p. cocaine (n = 9). Injections occurred immediately prior to initiation of the reinstatement session. See Supplementary  Methods.

Ex vivo voltammetry

Details of ex vivo voltammetry are based on previous work [42]. Briefly, naïve rats (n = 13) were lightly anesthetized using isoflurane [42], brains were rapidly removed, and a vibrating microtome was used to produce 400 µm-thick sections containing the DMS, as this region was the most robust predictor of cocaine-associated behavior in our studies. Using a bipolar stimulating electrode (Plastics One) and a carbon fiber microelectrode implanted in the DMS, DA release was evoked every 3 min using a single electrical pulse (4 ms, ~330 μA) and baseline levels of DA release and uptake were recorded until stable (≤15% variability). One rat was identified as an outlier using Grubb’s test (p < 0.05) and was removed from analysis. Cocaine (0.3–30 µM) was then cumulatively applied to the tissue and ensuing changes in DA release and uptake were recorded. Measures of uptake inhibition (appKm) were obtained using Michaelis–Menten modeling as previously described [42, 43, 49, 58, 59] and cocaine sensitivity (Ki) was calculated as Km divided by the slope of the dose-response curve [41, 60, 61]. See Supplementary Methods.

DAT knockdown

Four short hairpin RNAs (shRNAs) targeting different coding sequences of the rat DAT (rDAT) were tested for in vitro knock down of functional rDAT uptake activity. The most efficacious shRNA (CCAGTTACAATAAGTTCACCA) was then subcloned into an AAV vector (VectorBuilder Inc., Shenandoah, TX) and was packaged into an AAV2/8 using a Helper-Free system. Rats were injected with DAT-KD (n = 7 for voltammetry; n = 4 for western blots) or an AAV2/8-GFP control (n = 6 for voltammetry; n = 6 for western blots; AddGene) at ~100 nl/min in the ventral tegmental area (500 nl). Viral expression incubated for 3-4 weeks before ex vivo voltammetry experiments were conducted as described above. Western blots to confirm reduced DAT expression were performed on bilateral punches from the DMS. See Supplementary Methods.

Data analysis and statistics

DA dynamics were determined by a trained investigator and were confirmed by a second, blinded investigator. Self-administration data were collected using an automated program and processed automatically using a custom Excel macro. Cumulative lever pressing was binned in 5 min increments and represented as a percent of total lever presses (LPn/LPtotal). Inter-infusion interval was calculated as the time between two cocaine infusions, and post-infusion pause was calculated as the time between cocaine infusion and the next lever press. Only lever presses leading to an infusion were included in the calculation of the post-infusion pause to allow comparison with the inter-infusion interval. All rats obtained a minimum of 7 reinforcers, thus these measures were averaged across trials 2–7 to ensure inclusion of all animals. The first trial was excluded to allow for distinction between the post-infusion pause and response time during the inter-infusion interval. The rate of responding for each infusion was determined by subtracting the post-infusion pause from the inter-infusion interval and dividing the number of required responses by the result \((\frac{{Required\;Response}}{{Inter - infusion\;interval - Post - infusion\;Pause}})\).

Where noted, rats were divided into “slow” and “fast” uptake groups based on a median split within the experiment (i.e., upper 50% are designated as “fast”). Statistical analyses were conducted using SPSS (v25/26/27) and all data were assessed for normality using the Shapiro–Wilk test, and homogeneity of variance and sphericity using Levene’s and Mauchly’s tests, respectively. DA release and uptake for DMS and NAc were entered into regression analysis separately to facilitate detection of relationships. Corrections were applied where noted; reported p-values are two-tailed. Outliers were detected using Grubb’s test (GraphPad QuickCalcs).


Innate DA dynamics do not predict associative learning for cocaine

We used CPP to examine whether the formation of positive associations with cocaine was predicted by inherent variability in DA neurotransmission (Fig. S3a). As previously described, cocaine (20 mg/kg) produced significant CPP (Fig. S3b); however, the magnitude of preference for the cocaine-paired chamber was not correlated with either DA release or uptake in either the DMS (Fig. S3c, d) or NAc (Fig. S3e, f). We also evaluated behavior during the habituation test, as the response to a novel environment has been previously associated with DA neurotransmission [19, 23, 25, 62,63,64], see Supplementary Results (Fig. S3g–n).

We then measured acquisition of cocaine self-administration in a separate group (Fig. S4a, b). Baseline DA neurotransmission did not differ between rats that acquired and those that did not (Table S1), nor did DA dynamics in either the DMS or the NAc predict the number of sessions required for rats to meet acquisition criteria (Table S2).

Innate DA dynamics do not predict preferred levels of cocaine intake

Next, we measured the preferred level of cocaine intake under low-effort conditions in the subset of rats self-administering cocaine under FR1 for 2 weeks to determine whether inherent differences in DA release and uptake predicted consumption. No relationships were observed between DA release or uptake and cocaine intake under FR1 (Fig. S4c–f) in either the DMS or NAc. As FR1 responding for a moderate dose of cocaine incorporates both consummatory and appetitive aspects of intake, we then assessed cocaine free-consumption (Q0) using the within-session threshold procedure. Neither baseline DA release nor uptake in either region predicted Q0 (Fig. S4g–j).

DA uptake in the DMS, but not the NAc, predicts effort to obtain cocaine

We then determined whether DA dynamics predicted effort to obtain cocaine using the within-session threshold procedure (Fig. 1A, B). Simple regression analysis revealed that DA uptake in the DMS [F(1,10) = 14.732; r2 = 0.596, p < 0.005; n = 12], but not the NAc (n = 11), predicted the maximal price paid (Pmax; Fig. 1C, D). Pmax was not predicted by DA release in either area (Fig. S5a, b). Student’s t test revealed that rats defined as having “fast” baseline rates of DA uptake in the DMS based on a median split reached significantly higher unit prices for cocaine than those with “slow” baseline uptake rates [t(10) = −3.94, p < 0.005; Fig. 1E]. Linear regression demonstrated a similar relationship between DA uptake in the DMS and α [F(1,10) = 9.213, r2 = 0.48, p < 0.05; Fig. 1F]; rats with faster rates of uptake were less sensitive to changes in unit price (Fig. 1G; S5c, d). Note that a possible ceiling effect was observed in measures of Pmax. While this effect is likely mitigated by α, which is based on the slope of the demand curve, it is not known how animals that reached this ceiling would have responded to a higher unit price.

Fig. 1: DA uptake in the DMS predicts effortful responding for cocaine.
figure 1

A Experimental timeline showing that baseline DA release and uptake were measured in rats before testing under the within-session threshold procedure or under the progressive ratio (PR) schedule. B Example color plots and current vs time traces of stimulated DA release. DA uptake in C the DMS (n = 12) significantly predicts Pmax, and D rats with fast Vmax reach a significantly higher Pmax than rats with slow Vmax. E This relationship was not observed in the NAc (n = 11). F DA uptake in the DMS significantly predicts α. G Rats with fast uptake maintain cocaine intake across the session better than rats with slow uptake, reflecting their lower elasticity (α, see Fig. S5c). Average consumption (transparent lines) and modeled demand curves (bold lines) reflect this difference (average r2 = 0.75). H DA uptake in the NAc did not predict α. DA uptake in I the DMS (n = 15) significantly predicts breakpoints under the PR schedule. J Rats with fast Vmax tend to reach higher breakpoints than those with slow Vmax; K this was not observed in NAc (n = 12). The relationship between DA uptake and PR responding was especially strong in the first hour L, when rats with fast Vmax self-administered more cocaine than rats with slow Vmax M. N This relationship was not observed in the NAc. *p < 0.05, **p < 0.01. Bar graphs represent mean ± SEM.

We then used PR, a well-established model of motivation for cocaine, to corroborate this finding in a separate group of rats. Consistent with results under the within-session threshold procedure, regression analysis showed baseline rates of DA uptake in the DMS predicted average breakpoint [F(1,13) = 10.015; r2 = 0.435, p < 0.01; n = 15; Fig. 1I]. Rats designated as “fast uptake” tended to reach higher breakpoints [t(13) = 2.045, p = 0.062; Fig 1J]. Breakpoints were not predicted by DA uptake in the NAc (Fig. 1K), nor by DA release in either area (Fig. S5e, f). Regression analysis showed that the relationship between the number of infusions and the rate of DA uptake is particularly strong in the first hour of the session [F(1,13) = 30.355; r2 = 0.70, p < 0.001; Fig. 1L], when rats with inherently faster rates of DA uptake receive significantly more infusions than those with slower rates of uptake [t(13) = −4.03, p < 0.005; Fig. 1M]. This effect was not observed in the NAc (Fig. 1N).

Dopamine uptake in the DMS predicts topography of cocaine seeking under PR

Rats with slow and fast rates of uptake in the DMS made their final response under PR at the same approximate time [t(13) = 0.625; p = 0.543]; however, we observed a significant difference in cumulative responding (expressed as percent of total responses) between “slow” (n = 8) and “fast” (n = 7) uptake rats [repeated measures ANOVA, Greenhouse–Geisser correction; Time: F(2.244, 29.173) = 264.65, p < 0.001; Time X Uptake: F(2.244, 29.173) = 10.313, p < 0.001; Fast vs Slow Uptake: F(1,13) = 19.555, p < 0.005], suggesting a difference in response topography (Fig. 2A, B). To further explore this difference, we first compared the inter-infusion intervals between rats with inherently slow vs fast rates of DA uptake in the DMS (Fig. 2C). Rats with slower rates of DA uptake had longer average inter-infusion intervals than those with faster rates of uptake [unequal variances t-test: t(6.844) = 2.875; p < 0.05; Fig. 2D], with a moderate negative relationship between uptake in the DMS and the average inter-infusion interval [regression: r2 = 0.31; F(1,13) = 5.766; p < 0.05; Fig. 2E].

Fig. 2: Animals with faster rates of uptake in the DMS differ in the pattern of cocaine self-administration under the progressive ratio schedule.
figure 2

A Heat map depicting the rate of completion of total responses over the 6 h session for each rat. B Rats with fast Vmax (n = 7) differed in the rate of completion from rats with slow Vmax (n = 8). C Inter-infusion intervals and post-infusion pauses per reinforcer in rats with slow and fast Vmax. D Rats with fast rates of DA uptake in the DMS take shorter inter-infusion intervals, including shorter post-infusion pauses, than rats with slow rates of uptake. DA uptake in the DMS negatively predicts the E inter-infusion interval and F post-infusion pause. G Rats with fast (n = 6) and slow (n = 5) Vmax in the NAc did not differ in the rate of completion. *p < 0.05; ***p < 0.005. Bar graphs represent mean ± SEM.

The inter-infusion interval comprises the post-infusion pause—the time between the delivery of one reinforcer and the onset of effort to obtain the next one—and the time spent responding, from which the rate of responding may be determined; the presence or absence of a relationship with each of these measures may yield specific interpretations of the data. Post-infusion pause was predicted by DA uptake in the DMS [r2 = 0.29; F(1,13) = 5.361, p < 0.05; Fig. 2F] and was significantly longer in rats with slow vs fast rates of uptake [unequal variances t-test t(6.732) = 2.805; p < 0.05; Fig. 2D]; the rate of responding did not appear to correlate with uptake in the DMS or the NAc (Fig. S5g, h). This relationship did not appear to be present in the NAc, however, where dividing rats into fast (n = 6) and slow (n = 5) uptake did not reveal significant differences in response topography [repeated measures ANOVA, Greenhouse–Geisser correction; Time: [F(1.703, 15.326) = 121.03, p < 0.001; Time X Uptake: F(1.703, 15.326) = 1.338, p = 0.287; Fast vs Slow Uptake: F(1, 9) = 1.554, 0.224; Fig. 2G].

DA uptake in the DMS predicts cocaine-primed reinstatement, but not extinction

We then examined whether DA neurotransmission in the DMS predicted extinction and cocaine-primed reinstatement (Fig. 3A, B). Baseline DA uptake dynamics in the DMS did not predict responses on the active [r = 0.276, p = 0.412; n = 11] or inactive levers [r = 0.279, p = 0.406; not shown] on the first day of extinction, nor did DA release in the DMS [r = −0.44, p = 0.897; Fig. 3C, D]. Surprisingly, we did identify a significant relationship between DA release in the NAc and active-lever responding [release: r2 = 0.715, F(1,5) = 12.571, p < 0.05; uptake: r = −0.138, p = 0.768; n = 7; Fig. 3E–G], in which DA release was negatively associated with the number of lever presses. This relationship was not identified with the inactive lever [r = −0.439, p = 0.324; not shown].

Fig. 3: DA uptake in the DMS predicts cocaine-primed reinstatement.
figure 3

A Experimental timeline depicting DA release and uptake measures prior to testing for extinction and reinstatement of cocaine-seeking. B Mean ± SEM lever presses across consumption (for rats with 2 levers only), extinction, saline habituation (light blue squares) and i.p. cocaine challenge (red squares). The black dashed line represents the group average. Neither DA uptake (C) nor release (D) in the DMS (n = 11) correlated with responding on the active lever under extinction conditions. E DA uptake in the NAc did not correlate with extinction responding; however (F) DA release negatively predicted active-lever responding on the first day of extinction (n = 7), and (G) rats designated as “high” DA release in NAc showed significantly fewer responses on Day 1 of extinction compared to those designated as “low” DA release. Vmax (H) (n = 9) in the DMS predicted active-lever responding following a priming injection of cocaine. I Rats designated as “fast uptake” in the DMS made significantly more lever presses in response to a priming injection of cocaine than rats designated as “slow uptake”. J There was no observed relationship with DA release in the DMS. *p < 0.05, **p < 0.01. Bar graphs represent mean ± SEM.

We then used cocaine-primed reinstatement to model one aspect of relapse propensity. The rate of DA uptake in the DMS predicted active-lever responding following a challenge dose of cocaine [r2 = 0.8; F(1,7) = 27.353, p < 0.005; n = 9], with rats with faster rates of uptake exhibiting higher reinstatement responding than those with slower rates of uptake [t(7) = 5.404, p < 0.005] (Fig. 3H–J). As a group, there was a strong trend for cocaine to reinstate responding [paired t-test, t(8) = 2.294, p = 0.0509] that varied according to designation of “slow” or “fast” uptake; 100% of rats designated as “fast” uptake (n = 5) increased active lever responding compared to the previous day, whereas only 25% of rats designated as “slow” uptake (n = 4) did [Fisher’s Exact Test, p < 0.05].

DA uptake in the DMS mediates neurochemical sensitivity to cocaine

Results from these behavioral experiments suggest that baseline rates of DA uptake in the DMS predict cocaine seeking under conditions in which there is a pharmacological effect of cocaine, in a manner consistent with previous research into the effect of cocaine dose on behavior. For example, rats expend more effort for higher doses of cocaine under PR [65] and show a higher degree of reinstatement in response to higher doses of cocaine [66]. This may suggest that innate differences in the rate of DA uptake influence the neurochemical sensitivity to cocaine. We used ex vivo voltammetry to measure the relationship between DA uptake in the DMS and cocaine potency at the DAT (Fig. 4A) and identified a significant negative correlation between baseline Vmax in the DMS and Ki, a measure of apparent cocaine potency [r2 = 0.461, F(1,11) = 9.418; p < 0.05; n = 13; Fig. 4B]. As the D2R is implicated in stimulant use and is known to interact with DAT, we also assessed D2R function with respect to DA uptake. No relationship between baseline DA uptake and the response to quinpirole was identified (Fig. 4C), suggesting that baseline DA uptake is not a proxy for D2R function.

Fig. 4: Reduced rate of DA uptake in the DMS attenuates cocaine potency at the DAT.
figure 4

A Timeline showing coronal slices obtained from drug-naïve rats and exposed to increasing concentrations of cocaine and quinpirole. Vmax was negatively correlated with B cocaine potency at the DAT (Ki; n = 13) but not C the response to quinpirole. D Virus map and Timeline showing AAV-DAT-shRNA infusion followed by exposure of coronal slices to increasing concentrations of cocaine. E Examples of adjacent Western blots showing DAT and GAPDH protein content in GFP and AAV-DAT-shRNA-treated rats; F AAV-DAT-shRNA significantly reduced DAT expression compared to GFP-treated controls. G Examples of colorplots and current vs time plots from AAV-DAT-shRNA- and GFP-treated rats. H Vmax was lower in rats with DAT knockdown (n = 7) compared to GFP controls (n = 6), with a non-significant trend I for lower DA release. J DAT knockdown significantly increased Ki, indicating reduced apparent cocaine potency. *p < 0.05, **p < 0.01. Bar graphs represent mean ± SEM. ITR Inverted Terminal Repeat, CMV Cytomegalovirus promoter, EGFP Enhanced Green Fluorescent Protein, Intron-pA combined intron/poly-A sequences, shRNA short hairpin RNA.

We then used an AAV-DAT-shRNA to knock down the DAT to directly test whether experimentally disrupting DA uptake reduces cocaine’s apparent potency at the DAT (Fig. 4D and Fig. S6). DAT knockdown significantly reduced DAT expression [unequal variances t-test t(7) = 2.716, p < 0.05; Fig. 4E, F] and DA uptake [student’s t test: t(11) = 3.77, p < 0.005; Fig. 4G, H], with the average rate of uptake ~50% of GFP-control rats; there was also a non-significant trend for lower DA release [student’s t test: t(11) = 1.897, p = 0.084; Fig. 4I]. DAT knockdown also significantly reduced apparent cocaine potency [t(11) = −2.336, p < 0.05; Fig. 4J], indicating that significantly more cocaine was required to reach 50% inhibition of the DAT.


Here we used a combination of in vivo voltammetry and several tests of cocaine-associated behaviors to identify DA predictors of the risk and severity of future cocaine use. The current observations demonstrate that the rate of DA uptake in the DMS of cocaine-naïve rats significantly predicts future cocaine seeking in response to cocaine exposure, and further suggest that this effect may be produced by an enhanced sensitivity to cocaine in rats with inherently faster rates of DA uptake. Despite its broad implication in cocaine seeking, DA neurotransmission in the NAc predicted only extinction responding in these studies.

DA uptake in the DMS selectively predicts incentive value of cocaine

Our finding that the rate of DA uptake in the DMS of cocaine-naïve rats predicts their future effort expenditure for cocaine is consistent with some previous work in animals that innately expend more effort to obtain cocaine. For example, female rats work harder to obtain cocaine under a PR schedule of reinforcement [2] and demonstrate increased DA release and rates of DA uptake [7] as well as greater DAT density in the dorsal striatum [67], a finding that has been extended to humans [68]. Similarly, rats that have a lower locomotor response to intravenous cocaine have greater numbers of DAT binding sites in the dorsal striatum [16] and reach higher breakpoints [15].

As both PR and the within-session threshold schedules involve manipulating the ratio of effort to reward to measure cocaine-seeking, we expected to identify a similar relationship between the rate of DA uptake in the DMS and responding under extinction conditions (which has an infinite effort to reward ratio). Surprisingly, DA uptake in the DMS did not predict extinction responding—suggesting that the relationship between DA uptake and effort expenditure relies on the response to cocaine itself. This is supported by our finding that DA uptake in the DMS predicts the response to a priming injection of cocaine, with rats that demonstrated faster rates of DA uptake at baseline pressing the formerly cocaine-paired lever more than those with slower rates of uptake. Importantly, we did not identify a relationship with intake under low-effort conditions (i.e., two-weeks of FR1 and Q0), nor did we identify a relationship with CPP or the acquisition of self-administration. These findings suggest that increased effort to obtain cocaine does not extend from higher preferred level of cocaine intake or the formation of positive associations with cocaine.

The shorter inter-infusion interval observed in rats with faster rates of DA uptake in the DMS supports the overall finding that innate DA uptake predicts severity of cocaine use, as previous work identified inter-infusion intervals as an important predictor of addiction-like behavior [69]. In both uptake groups, the inter-infusion interval was dominated by the post-infusion pause—the period of time after the cocaine infusion before the rat begins seeking the next reinforcer. This measure has been observed to be ratio- and dose-dependent, with doses on the ascending limb of the dose-response curve generally resulting in shorter pauses [70]. As rats with faster rates of uptake in the DMS demonstrate shorter post-infusion pauses, this may provide further evidence that the rate of DA uptake in the DMS modulates sensitivity to the reinforcing effects of cocaine. Importantly, we cannot definitively state whether or how the dose-response curve to cocaine may have shifted, as we used only a single, moderate dose of cocaine typically found on the ascending limb of the dose-response curve for PR experiments, and increased Pmax and alpha may be attributable to either vertical or horizontal shifts in responding.

The lack of an observed relationship between innate DA neurotransmission in the NAc and cocaine self-administration was surprising, as considerable previous work heavily implicates this region in aspects of cocaine intake [71], motivation to obtain cocaine [44, 49, 72, 73], and in response to associated cues [74]. Further, previous work by Willuhn et al suggested that the early stages of cocaine self-administration rely on DA transients in the NAc, with reliance shifting over time to the dorsolateral striatum [75], and others have shown a relationship between NAc DA, the locomotor response to novelty, and cocaine self-administration [19, 23, 24, 62]. Nevertheless, our results are consistent with some prior work. We previously demonstrated that neurochemical sensitivity to cocaine—but not cocaine-naïve measures of DA release or uptake using voltammetry—differed between low- and high-responders to novelty [19], even though these high responders acquired cocaine self-administration more quickly than low-responders.

Notably, our results suggest that DA release and uptake in the NAc negatively predict future cocaine-seeking under extinction conditions and the exploratory (but not locomotor) response to a novel environment. This partially corroborates a previous report that DA uptake and DAT expression in prefrontal cortex, but not striatum, correlated negatively with response to inescapable novelty [76] and is potentially consistent with previous findings that high-responders have higher basal levels of DA in the NAc [23] (but see [24]), which may be expected to result from lower levels of uptake [24]. Both a novel environment [77] and unreinforced operant behavior [78, 79] are stressors, and cocaine itself induces release of stress hormones [80]. Therefore, this may eventually lend support to the hypothesized role of the NAc in integrating limbic and locomotor responses—although exactly how this interacts with baseline levels of stimulated DA release is unclear.

Innate DA uptake in the DMS predicts the dopaminergic response to cocaine

A potential basis for the relationship between DA uptake in the DMS and subsequent expression of motivation for cocaine is a difference in the behavioral response to cocaine mediated by an altered neurochemical response. To evaluate this, we used ex vivo voltammetry in the DMS of cocaine-naïve rats to measure DA release and uptake at baseline and in response to cocaine. Rats with innately faster rates of DA uptake demonstrated an enhanced “sensitivity” to cocaine—i.e., less cocaine was required to inhibit DA uptake. This effect was recapitulated in rats with DAT knockdown, demonstrating that reducing DA uptake reduces the apparent pharmacological potency of cocaine. Greater neurochemical sensitivity to cocaine has previously been associated with an increased motivation to obtain it; increasing the dose of a cocaine reinforcer—and, thus, DA uptake inhibition—increases effort output, with rats reaching higher breakpoints for higher cocaine doses [65, 72]. Taken with our neurochemical and behavioral observations, this may suggest that rats with faster rates of DA uptake in the DMS respond to cocaine as if it were a higher dose than rats with slower rates of DA uptake. Future work using projection-specific viral vectors to reduce DA uptake in specific brain regions will be vital to explore this relationship further.

Although we demonstrated that DA uptake in the DMS significantly predicts the incentive value of cocaine and the pre-synaptic DA response, it is unclear whether DA uptake per se regulates the behavioral response for cocaine or whether it might instead elicit alterations in other mechanisms in the DA system to produce its effects—particularly at the receptor level. For example, increasing the rate of DA uptake decreases extracellular, “tonic” DA release measured using microdialysis [81] while also increasing D1 receptor and D2R expression [82], and DAT knockdown has previously been reported to diminish D1 receptor function [83]. The observed increases in pre-synaptic sensitivity to cocaine in rats with inherently faster rates of DA uptake may therefore act on a larger number of available post-synaptic receptors to produce stronger effects. These would likely lead to other behavioral effects, including differences in the response to other substances of abuse or natural reinforcers. Future work will explore this possibility.


Specific aspects of cocaine self-administration may be reliably predicted by innate differences in DA neurotransmission prior to initial cocaine exposure, carrying important ramifications for future scientific and public health initiatives. First, these studies suggest that factors that increase or decrease the rate of DA uptake in the DMS may, respectively, increase or decrease the severity of future cocaine use disorders. Some research suggests environmental enrichment may reduce the rate of DA uptake in cortical [84,85,86] and potentially striatal [87, 88] areas and may reduce the risk of cocaine-associated behaviors [88, 89]—suggesting that non-invasive interventions may act on neurochemical substrates to reduce the impact of future cocaine use. Second, these studies indicate that DA uptake modulates the neurochemical sensitivity to cocaine. Innate DA uptake predicts cocaine-induced DA uptake inhibition, an effect recapitulated by knocking down the DAT using shRNA, suggesting that faster DA uptake increases the incentive value of cocaine by increasing the DA available to act post-synaptically. Together, these results suggest that individuals with faster rates of DA uptake in the DMS may be more sensitive to the effects of cocaine, and thus find cocaine more reinforcing, resulting in aberrant patterns of use. Future work will explore the impact of innate DA uptake on post-synaptic DA receptor responsivity and the pharmacological specificity of these results.

Funding and disclosures

This work was funded by National Institutes of Health grants MH106912 (OVM), DA044205 and DA049545 (MDB), and DA043787 and DA031900 (RAE). The authors report no biomedical financial interests or potential conflicts of interest.