Neural bases for attenuation of morphine withdrawal by Heantos-4: role of l-tetrahydropalmatine

Severe withdrawal symptoms triggered by cessation of long-term opioid use deter many individuals from seeking treatment. Opioid substitution and α2-adrenergic agonists are the current standard of pharmacotherapy for opioid use disorder in western medicine; however, each is associated with significant complications. Heantos-4 is a non-opioid botanical formulation used to facilitate opioid detoxification in Vietnam. While ongoing clinical use continues to validate its safety and effectiveness, a mechanism of action accounting for these promising effects remains to be specified. Here, we assess the effects of Heantos-4 in a rat model of morphine-dependence and present evidence that alleviation of naloxone-precipitated somatic withdrawal signs is related to an upregulation of mesolimbic dopamine activity and a consequent reversal of a hypodopaminergic state in the nucleus accumbens, a brain region implicated in opioid withdrawal. A central dopaminergic mechanism is further supported by the identification of l-tetrahydropalmatine as a key active ingredient in Heantos-4, which crosses the blood–brain barrier and shows a therapeutic efficacy comparable to its parent formulation in attenuating withdrawal signs. The anti-hypodopaminergic effects of l-tetrahydropalmatine may be related to antagonism of the dopamine autoreceptor, thus constituting a plausible mechanism contributing to the effectiveness of Heantos-4 in facilitating opioid detoxification.

withdrawal-associated reductions in firing rates of DA neurons in the midbrain along with attenuated DA efflux in the basal forebrain [18][19][20][21][22][23][24] . Based on these data, normalization of hypodopaminergia has been proposed as a pharmacotherapeutic strategy to facilitate cessation of opioid use 2,16,25 . Indeed, this therapeutic rationale is consistent with the findings of our initial preclinical assessment of Heantos-4, which confirmed in morphine-dependent rats the alleviation of naloxone-induced hypodopaminergia and significant attenuation of somatic withdrawal measures 19 . Similar outcomes were also observed with the α2-adrenergic agonist clonidine 18 , suggesting that normalization of DA function may be achieved even by indirect effects on the DA system.
The present investigation was guided by the hypothesis that Heantos-4 facilitates detoxification from opioids through direct modulation of the DA system. Employing in vivo microdialysis, we first conducted a series of neuropharmacological investigations in the nucleus accumbens (NAc), a key region of the mesocorticolimbic DA system implicated in opioid reward 13 as well as the expression of somatic and aversive features of opioid withdrawal 26,27 . The first experiment, using morphine-dependent rats, examined if Heantos-4 could reverse a hypodopaminergic state in the NAc precipitated by the µ-opioid antagonist naloxone after its onset and concurrently alleviate somatic correlates of withdrawal. We next considered the possibility of a therapeutic mechanism of action directed at the DA autoreceptor, which has a critical role in moderating DA transmission especially following exposure to opioids and other drugs of abuse 28 . Presynaptic DA autoreceptors (including the D2S short isoform and D3 subtypes) orchestrate the synthesis, storage and release of DA [29][30][31][32] . At appropriate low doses, D2/D3 agonists augment auto-inhibition of DA release, thereby provoking a hypodopaminergic state; in contrast, D2 antagonists block feedback inhibition, in effect, stimulating DA release [33][34][35][36] . It is conceivable that the anti-hypodopaminergic effects of Heantos-4 similarly involves blockade of the DA autoreceptor and a consequent upregulation of DA release (i.e., a DA autoreceptor antagonist). Thus, a second experiment examined a putative DA autoreceptor antagonist property of Heantos-4 against the hypodopaminergic effects of the D2/3 agonist quinpirole.
Heantos-4 is composed of extracts from twelve medicinal plants (see Table 1 in ref. 19 ), the qualitative analyses of which have identified 194 compounds representing the major classifications of bioactive phytochemicals [37][38][39][40] . In seeking to further substantiate a central dopaminergic mechanism of action, we identified l-tetrahydropalmatine (l-THP) as a phytochemical constituent of interest. Its analgesic properties and potential therapeutic value in substance misuse are attributed to modulation of DA receptors 8,41 . Specifically, l-THP behaves functionally as a D1, D2 and D3 antagonist 8,[42][43][44][45] and moreover, exhibits the prototypical property of DA antagonists to increase DA release from axon terminals 46,47 . Consistent with DA autoreceptor modulation, low doses of l-THP (1-5 mg/kg) have been shown to enhance DA synthesis, release and metabolism, and most importantly, reverse the DA release-attenuating effects of the D2 agonist apomorphine 43,47 . To demonstrate that l-THP derived from Heantos-4 crosses the blood-brain barrier, a third experiment employed ultra high-pressure liquid chromatography/mass spectrometry (UHPLC/MS) methodology to detect l-THP in blood plasma and cerebrospinal fluid following oral administration of Heantos-4. A fourth experiment tested the hypothesis that l-THP may serve as an antagonist of the DA autoreceptor, capable of reversing the hypodopaminergic effects of quinpirole. Finally, a fifth experiment assessed whether l-THP as a key neuroactive ingredient in Heantos-4 could replicate the therapeutic effects of its parent formulation in morphine-withdrawn rats.

Results
Experiment 1: Effects of Heantos-4 in morphine-withdrawn rats. Rats were systemically injected with morphine (10 mg/kg/day, i.p.) on Days 1-7 days and naloxone (10 mg/kg, i.p.) on Day 8 (Fig. 1a). Naloxone evoked a significant reduction in DA efflux compared to baseline in all rats ( Fig. 1b; vehicle-treated, Time − 40-150 min, F 18,90 = 1.76, p = 0.04; Heantos-4-treated, Time − 40-0 min, F 3,18 = 6.62, p < 0.01). In animals that subsequently received vehicle gavage, the hypodopaminergia persisted over a 3-h period. In contrast, administration of Heantos-4 (500 mg/kg, p.o.) 30 min after the onset of hypodopaminergia resulted an immediate increase in DA efflux that was sustained at values ~ 40% above pre-naloxone baseline levels for the duration of the experiment. Mean changes in DA (F 1,11 = 15.91, p < 0.01) and DOPAC (F 1,11 = 83.20, p < 0.01) efflux elicited by Heantos-4 were significantly higher than in the vehicle condition. The increase in dopaminergic activity following Heantos-4 was paralleled by a significant amelioration of naloxone-precipitated withdrawal signs (Fig. 1c). In comparison to vehicle, Heantos-4 treated rats showed a significantly lower occurrences of face and body grooming (estimate of difference = 5.14; 95% confidence interval 2.41-8.07; p < 0.01), wet dog shakes (estimate of difference = 1.74; 95% confidence interval 0.55-3.20; p = 0.01), abdominal stretching (estimate of difference = 3.10; 95% confidence interval 1.21-5.21; p = 0.049) and rearing (estimate of difference = 4.41; 95% confidence interval 2.72-6.44; p < 0.01). Experiment 2: Effect of Heantos-4 on quinpirole-induced hypodopaminergia. We next assessed the presynaptic DA autoreceptor as a possible site at which Heantos-4 modulates DA release. First, we established an in vivo functional assay in which the DA autoreceptor could be selectively activated by appropriately low concentrations of the D2/D3 receptor agonist quinpirole (D2 Ki = 4.8 nM, D3 Ki = 24 nM) 48 or inhibited by the D2 antagonist eticlopride (Ki = ~ 0.92 nM) 49 (Fig. 2a). In morphine-naïve rats, application of quinpirole (1 µM) directly into the NAc by reverse-dialysis was accompanied by a state of hypodopaminergia (Fig. 2b), with a significant reduction in basal DA levels reaching ~ 40% below baseline (F 18,108 = 17.30, p < 0.01). Notably, quinpirole at 1 µM (or 0.26 ng/µL/min over 150 min) is within the lower range of the dose-response analyses of quinpirole (3-300 ng) previously shown to selectively affect DA autoreceptor function 50 . In contrast, eticlopride (50, 100 nM) dose-dependently elevated DA efflux above baseline (F 18,126 = 3.37, p < 0.01 and F 18,126 = 6.99, p < 0.01, respectively). Co-application of quinpirole and eticlopride resulted in an initial decline in DA efflux below baseline, which within 20-30 min returned to baseline values with the lower dose of eticlopride ( The microdialysis-based DA autoreceptor assay was then used to assess the putative DA autoreceptor antagonist property of Heantos-4 in a separate group of rats (Fig. 2c). Intra-NAc reverse-dialysis of quinpirole (1 µM), followed 30 min later by vehicle gavage, resulted in a significant state of hypodopaminergia (F 18,144 = 29.371, p < 0.01) (Fig. 2d). Subsequent administration of Heantos-4 (125, 250 and 500 mg/kg, p.o.) evoked an increase in DA efflux that quickly but transiently approached baseline levels, with the highest dose evoking increased efflux above pretreatment values. Statistical comparisons of mean changes in DA (F 3,40 = 8.23, p < 0.01) and DOPAC (F 3,40 = 37.69, p < 0.01) efflux indicated a significant dose-dependent reversal of quinpirole-induced hypodopaminergia by Heantos-4. As l-THP shares a similar molecular weight and tetrahydroprotoberberine chemical scaffold with numerous compounds in Heantos-4 (including those in Table 1), the UHPLC/MS method was optimized to detect the unique fragmentation pattern of l-THP molecules (Fig. 3b,c). This permitted the selective detection and quantification of l-THP in blood plasma and cerebrospinal fluid samples (Fig. 3d,e). At each of the collection times (15,30,45, 60 min post-treatment), l-THP levels were clearly present in plasma and cerebrospinal fluid of Heantos-4-treated rats; in contrast, l-THP was not detected in any of the samples from vehicle-treated rats (Fig. 3f,g). Given the obvious group differences (and zero variance in the vehicle-treated control group), between-group statistical analyses were not conducted. l-THP was detectable in plasma and cerebrospinal fluid up to 60 min post-Heantos-4 administration, establishing a clear main effect of time in the Heantos-4-treated group in blood plasma (F 3,24 = 4.17, p = 0.02) and cerebrospinal fluid (2-tailed t-test, t = 2.48, df = 15, *p = 0.03). The concentration of l-THP were two orders of magnitude lower in the cerebrospinal fluid than in blood; nevertheless, the oral bioavailability of Heantos-4 as indexed by l-THP levels in cerebrospinal fluid (which ranged from 188 to 234 nM) were within the concentration range required for occupancy of D1 and D2 receptors (Ki = 124 ± 6 nM and 388 ± 78 nM, respectively) but not D3 (Ki = 1420 ± 220 nM) in the brain 42 .  that l-THP derived from Heantos-4 crosses the blood-brain barrier, we next tested the hypothesis that l-THP, like its parent formulation, modulates DA efflux through antagonism of the DA autoreceptor. In our hands, intra-NAc reverse-dialysis of l-THP (50 µM) significantly increased DA efflux (F 18,162 = 8.17, p < 0.01), whereas quinpirole (1 µM) induced a marked state of hypodopaminergia (F 18,108 = 26.16, p < 0.01) (Fig. 4a,b). When quinpirole was co-infused with l-THP, DA efflux remained near pretreatment baseline levels. Importantly, mean changes in DA (F 2,25 = 103.40, p < 0.01) and DOPAC (F 2,25 = 31.15, p < 0.01) efflux were significantly higher in the quinpirole + l-THP group compared to quinpirole alone, suggesting that the quinpirole-induced hypodopaminergia was antagonized by the effects of l-THP in the NAc (i.e., via presynaptic DA autoreceptors on axon terminals as distinct from somatodendritic sites). We then examined whether the effects of intra-NAc l-THP on the DA autoreceptor could be replicated by a systemic route of administration (Fig. 4c). Reverse-dialysis of quinpirole (1 µM) in the NAc, and followed by saline treatment (i.p.) 30 min later, resulted in a significant inhibition of DA efflux (F 18,108 = 17.12, p < 0.01) (Fig. 4d). Subsequent treatment with l-THP (5 mg/kg, i.p.) evoked an initial increase in DA efflux that transiently restored baseline levels, and then gradually subsided to ~ 30% below baseline. Systemically administered eticlopride (0.3 mg/kg, i.p.), which served as an internal control for a DA autoreceptor antagonist, resulted in a rapid and sustained increase in DA efflux that approached baseline. As indicated by the mean change in DA (F 2,24 = 6.98, p < 0.01) and DOPAC (F 2,24 = 19.45, p < 0.01) efflux, dopaminergic activity was significantly higher in rats treated with either l-THP or eticlopride compared to saline. In animals subsequently injected with saline, the hypodopaminergia persisted over a 3-h period, replicating the observations in Experiment 1. Subsequent treatment with either a single dose of l-THP (5 mg/kg, i.p.) or eticlopride (0.3 mg/kg, i.p.), the latter once again serving as an internal control for D2 antagonism, resulted in an immediate increase in DA efflux from a hypodopaminergic state. Both l-THP and eticlopride treatments evoked mean changes in DA (F 2,18 = 17.66, p < 0.01) and DOPAC (F 2,18 = 51.12, p < 0.01) that were significantly higher than in the vehicle condition. Importantly, , the moderate increases in DA efflux, induced by l-THP ranging from ~ 30 to 50% above baseline, were accompanied by a significant amelioration of naloxone-precipitated withdrawal signs (Fig. 5c). In comparison to saline, l-THP-treated rats showed a significantly lower occurrence of face and body grooming (estimate of difference = 3.00; 95% confidence interval 0.62-5.49; p = 0.02), wet dog shakes (estimate of difference = 2.00; 95% confidence interval 0.76-3.50; p < 0.01) and abdominal stretching (estimate of difference = 3.83; 95% confidence interval 2.17-5.78; p < 0.01). In comparison to saline, eticlopride-treated rats displayed significantly lower occurrences of wet dog shakes (estimate of difference = 1.83; 95% confidence interval 0.67-3.34; p < 0.01,) abdominal stretching (estimate of difference = 3.13; 95% confidence interval 1.47-5.13; p < 0.01) and rearing (estimate of difference = 0.83; 95% confidence interval 0.26-1.83; p = 0.01).
Basal DA levels in morphine-treated and -naïve rats (Experiments 1-5). D2 receptors are vulnerable to regulatory changes in response to withdrawal from repeated morphine treatments 51 . To assess whether DA autoreceptor-mediated feedback inhibition was altered by withdrawal from morphine treatment, we compared basal DA concentrations in morphine-dependent and -naïve rats. On microdialysis day, DA concentrations (mean ± s.e.m.) of samples collected during the baseline period prior to naloxone treatment in rats that received the 7th morphine injection 24 h earlier (Experiments 1 and 5, n = 33) and control rats naïve to morphine (Experiments 2 and 4, n = 145) were 1.05 ± 0.08 nM and 1.09 ± 0.05 nM, respectively. Analysis of the basal DA values observed in each of the experiments (Supplementary Table S1) revealed no statistical difference [H 5 = 9.16, p = 0.10].

Discussion
Current non-opioid detoxification therapies target the dysfunctional states of the dopaminergic and noradrenergic systems which are associated with affective and somatic correlates of opioid withdrawal 18,52 . For example, in morphine-dependent animals, clonidine reverses a hypodopaminergic condition 18 and noradrenergic hyperactivity as well as activating α2-adrenergic and imidazoline-1 receptors 53,54 . However, while the combined agonism of both receptor types synergistically prevents inhibition of midbrain DA neuron firing induced by naltrexone 54 , imidazoline-1 receptor activity is also associated with hypotensive side-effects that limit utility 55 . The present findings indicate that a modest upregulation of mesolimbic DA activity, and a consequent reversal of a hypodopaminergic state in the NAc, underlies the ability of Heantos-4 to attenuate somatic signs of withdrawal in morphine-dependent rats. The effectiveness of a single treatment with Heantos-4 was notable, as was its immediacy (on the order of minutes) in either preventing the appearance 19 or mitigating the hypodopaminergic and somatic features of withdrawal after their onset (present study).
The identification of l-THP as a key active ingredient, which crosses the blood-brain barrier (Fig. 3) and shows a therapeutic potency equal to its parent formulation in morphine-dependent rats (Figs. 1 and 5) further substantiates a central dopaminergic mechanism of action. In line with the known characterization of l-THP as a D2/3 antagonist 8,42-45 , application of a low dose directly in the NAc reverses hypodopaminergia induced by the D2/3 agonist quinpirole (Fig. 4) as well as the D2 agonist apomorphine 43,47 , consistent with the selective antagonism of DA autoreceptors in the NAc. Findings in mice with D2 receptor mutations suggest that the D2S www.nature.com/scientificreports/ isoform in particular may play an important role in opioid dependence 28 . Specifically, the capacity of D2S to presynaptically assert auto-inhibition on DA activity appears critical to dampening supra-physiological increases in DA release evoked by opioids; conversely, it is plausible that blockade of D2S autoreceptors could normalize basal DA activity from a hypoactive state. Based on these premises, we propose that the anti-hypodopaminergic effects of l-THP reflect a putative antagonist action on presynaptic DA autoreceptors, which could serve as a novel therapeutic strategy to 'normalize' dysfunction of the mesolimbic DA system in OUD. Importantly, enhancement of DA function in the NAc, either by drugs or endogenous opioids, is effective in the suppression of tonic pain 56 , as alleviation of pain contributes to successful detoxification from opioids 2 . Accordingly, the analgesic properties of l-THP 8 , mediated in part by elevation of DA release in the NAc may contribute to management of withdrawal-associated pain. It should also be noted that effects of l-THP (administered systemically) on somatic signs of withdrawal are unlikely to be mediated solely by the NAc; indeed, it www.nature.com/scientificreports/ will be important to determine the involvement of somatodendritic DA autoreceptors in regulating midbrain DA cell activity. Although our data suggest that the anti-hypodopaminergic effects of l-THP and Heantos-4 involve antagonism of the DA autoreceptor, other mechanisms, such as DA transporter (DAT) inhibition, could also contribute to this effect. It is notable that different DAT inhibitors are distinguishable by the DOPAC responses they evoke: both cocaine and d-amphetamine give rise to an increase in extracellular DA accompanied by a decrease in DOPAC levels [57][58][59] whereas methylphenidate elicits increases in DA levels, with relatively little to no change in DOPAC 60,61 . In this regard, the effects of Heantos-4 and l-THP to consistently enhance extracellular levels of DOPAC do not reflect patterns seen following disruption of DAT function. In fact, the robust and sustained elevation of both DA and DOPAC efflux (Figs. 1, 2, 4 and 5) resembles the upregulation of DA activity observed with D2 antagonists 62,63 .
The comparable potency of Heantos-4 and l-THP to reverse hypodopaminergia induced by either quinpirole in drug-naïve or naloxone in morphine-dependent rats, while also ameliorating key features of opioidwithdrawal, is compelling. Indeed, l-THP has been explored as a treatment option for heroin and cocaine-use disorders, with the caveat that doses higher than the recommended therapeutic range for humans (60-180 mg) are associated with adverse side-effects, such as sedation and anhedonia 41,42,78 . Higher, sedation-inducing doses of l-THP may involve antagonism of postsynaptic D2 receptors as well as a non-dopaminergic (benzodiazepine-like) mechanisms via positive allosteric modulation of GABA-A receptors 64 . It is also noteworthy that non-sedative doses of l-THP (below 10 mg/kg in rodents) [65][66][67][68][69]77 , administered in combination with a low dose of a µ-opioid receptor antagonist naltrexone is more effective than l-THP alone in blocking cocaine-seeking by rats 70 . While the precise quantity of l-THP in the Heantos-4 formulation in relation to the threshold for inducing side-effects remains to be determined, our behavioral observation of Heantos-4-treated rats indicate that l-THP within the Heantos-4 formulation is well-tolerated, with no adverse side-effects aside from a mild state of somnolence (present study and ref. 19 ).
The use of combined botanical extracts with a range of pharmacological effects is a prominent feature of many botanical formulations 71 , and may account for the clinical success of Heantos-4 to mitigate opioid withdrawal. While the present study focused on l-THP as a mediator of therapeutic effects of Heantos-4, this formulation contains other phytochemicals in the tetrahydroprotoberberine family with known effects on brain dopaminergic activity and opioid dependence (Table 1). Further investigation of these and other botanicals within the Heantos-4 formulation, with a focus on interactions between the dopaminergic and noradrenergic systems 52 as well as on both somatic and affective features of opioid withdrawal 2,13 are warranted in order to gain a fuller understanding of factors that may explain the promising clinical effects of Heantos-4 in facilitating opioid detoxification. Furthermore, we recognize that the present emphasis on somatic signs of withdrawal from opioid dependence represents only one of several important features of OUD. In particular, the relationship between hypodopaminergia and the dysphoric effects of opioid withdrawal merit further study 13 , and experiments are underway to extend our earlier preclinical assessments of affective measures of opioid withdrawal (e.g., naloxone-precipitated conditioned place avoidance; see ref. 19 ). Future studies utilizing more sophisticated models of addiction, including self-administration paradigms, will explore the potential effectiveness of acute and repeated Heantos-4 treatments on opioid craving, seeking and relapse.
In summary, the present preclinical insights into the neuropharmacological, neurochemical and behavioral effects of Heantos-4 in morphine-dependent rats complement the promising clinical observations in Vietnam and provide a compelling rationale for further investigations of this non-opioid based formulation as a gateway treatment for OUD.

Methods
Subjects and surgery. All experimental procedures were conducted in accordance with the ethical standards set by the Canadian Council on Animal Care and approved by the University of British Columbia Animal Care Committee.
Male Sprague-Dawley rats (250-275 g) from Charles River (St. Constant, Canada) were pair-housed upon arrival in a colony room that was maintained at ~ 21 °C with a 12-h light/dark cycle (lights on at 7 p.m.).
Rats in microdialysis experiments were implanted with bilateral guide cannulae directly over the NAc (from bregma, + 1.7 mm anterio-posterior, ± 1.1 mm mediolateral; from dura, − 1.0 mm dorso-ventral) as previously described 19 . During recovery (7 days minimum) rats were individually housed, with food and water available ad libitum.
In vivo microdialysis and HPLC. Microdialysis probes were concentric in design and assembled in-house as previously described 19 . Prior to implantation, probes were connected to an Instech liquid swivel (Plymouth Meeting, PA, US) and thoroughly flushed with the perfusion medium, artificial cerebrospinal fluid which contained (in mM): sodium phosphate buffer (10), CaCl 2 (1.2), KCl (3), MgCl 2 (1) and NaCl (147) at pH 7.4. The probes were then inserted via the guide cannulae into the NAc (2 mm membrane length spanned − 4.8 to − 6.8 mm below dura) without the use of anesthesia. Rats remained in a plexiglass chamber (40 × 40 × 40 cm) with food and water overnight. Probes were continuously perfused with artificial cerebrospinal fluid at 1 µL/ min. Sample collection began 16-18 h post-implantation at 10 min intervals. Once DA levels stabilized (with less than 10% fluctuation over four consecutive samples), each rat received its predesignated treatment and sampling continued for several hours.
Microdialysis samples were analyzed immediately upon collection using reverse-phase HPLC with electrochemical detection as previously described 19 . The identification and quantification of DA and Scientific Reports | (2020) 10:21275 | https://doi.org/10.1038/s41598-020-78083-x www.nature.com/scientificreports/ dihydroxyphenylacetic acid (DOPAC) content in microdialysis samples were based on a 3-point calibration of external standards prepared in artificial cerebrospinal fluid. The placement of microdialysis probes in the brain was verified using histological methods as previously described 19 . As shown in Supplementary Fig. S1, tracts formed by the membrane portion of microdialysis probe were localized to the shell/core border of the NAc, spanning 1.2 to 2.2 mm anterior to bregma.

Drugs and preparation.
Heantos-4, produced by the Sung lab, was received as a dry granular powder that was not readily soluble in water. Therefore, the drug was prepared daily as a suspension in 0.5% (w/v) carboxymethyl cellulose in H 2 O (vehicle) and administered 2.0-2.5 mL per os (p.o.) using a soft rubber feeding tube. The dose-range (125, 250 and 500 mg/kg) and the oral route of administration were based on the clinical treatment prescribed for opioid detoxification in Vietnam 11 .
Quinpirole hydrochloride, eticlopride hydrochloride and naloxone hydrochloride were obtained from Sigma-Aldrich (Oakville, Canada). Morphine sulfate and l-THP were obtained from Unipharm Wholesale Drugs Limited (Richmond, Canada) and Santa Cruz Biotechnology (Dallas, Texas, US), respectively. quinpirole, eticlopride and l-THP were prepared daily in 0.5% (v/v) dimethyl sulfoxide in artificial cerebrospinal fluid for application by intra-NAc reverse-dialysis. Morphine and naloxone were prepared daily in sterile saline for intraperitoneal (i.p.) injections. Eticlopride and l-THP were initially dissolved in ~ 100 µL 0.1 M H 2 SO 4 and brought to final volume with saline (pH was adjusted to 5.5 using NaOH). All i.p. injections were administered at 1 mL/kg. ments 1 and 5). The rat model of morphine dependence employed here is modified from a previously described protocol 18 . One week following surgery, morphine (10 mg/kg, i.p.) was administered on Days 1-7. Approximately 5 h after the final injection, rats were implanted with microdialysis probes in the NAc (left and right hemispheres counterbalanced). The following morning (on Day 8), with stable baseline DA levels established, the µ-opioid receptor antagonist naloxone (10 mg/kg, i.p.) was administered, followed 30 min later by either Heantos-4 (500 mg/kg, p.o.), vehicle (p.o.), l-THP (5 mg/kg, i.p.) or saline (i.p.).

Concurrent microdialysis and assessment of withdrawal in morphine-dependent rats (Experi-
Concurrent with microdialysis sampling, rats were assessed for naloxone-precipitated withdrawal signs at 0-20 min, 60-80 min and 120-140 min following administration of Heantos-4, l-THP, eticlopride, or control treatment (saline or vehicle). Using real-time monitoring and video recordings, a researcher blind to treatment conditions recorded the occurrences of face and body grooming, wet dog shakes, abdominal stretching and rearing.

Microdialysis/reverse-dialysis experiments in morphine-naïve rats (Experiments 2 and 4). Rats
with probes in the NAc were randomly assigned to one of the following treatment condition: intra-NAc reversedialysis of quinpirole (1 µM), eticlopride (50 or

Collection and UHPLC/MS analysis of blood plasma and cerebrospinal fluid samples (Experiment 3). Blood collection.
Following Heantos-4 (500 mg/kg, p.o.) or vehicle administration, anesthesia was induced with 4% isoflurane gas mixed with oxygen and maintained with 2.0-2.5% isoflurane for the remainder of the procedure. Blood from the lateral tail vein was collected using a previously described procedure 72 . Samples were collected serially at 15, 30, 45 and 60 min post-treatment (~ 1 mL /time point) from each rat (i.e., withinsubject design) and deposited into Eppendorf tubes coated with ethylenediaminetetraacetic acid (0.1 M). To isolate plasma, blood samples were centrifuged at 4 ℃ for 10 min at 1300×g and then filtered through an ultrafiltration cartridge (30 kDa cut-off) to remove proteins for UHPLC/MS analysis.
Cerebrospinal fluid collection. Following oral gavage of Heantos-4 (500 mg/kg, p.o.) or vehicle, rats were anesthetized with urethane (25 g/7 mL) and, in a prone position, the head was secured at a downward (~ 45°) angle from horizontal. The dissection of tissue to reveal the cisterna magna was performed according to a previously described procedure 73 . A 28 G ½ in. needle attached to a 1 mL syringe was inserted through the dural surface of the cisterna magna at a 30° angle. The cerebrospinal fluid was carefully aspirated into the syringe until ~ 100 µL of fluid was collected, deposited into an Eppendorf tube and stored at − 80 ℃. In comparison to blood, the volume of cerebrospinal fluid is significantly smaller, and permits collection of a single sample per time-point. Thus, a between-group design was employed to collect cerebrospinal fluid samples at 30 and 45 min post-gavage. Time-constraints related to experimental procedures (e.g., oral administration, induction of anesthesia and tissue dissection prior to cerebrospinal fluid collection) precluded sample collection at 15 min. UHPLC/MS system. Analysis of blood plasma and cerebrospinal fluid samples were conducted using an UHPLC/MS system consisting of an Agilent 1290 Infinity Binary Pump, Sampler, Thermostat, and Thermostatted Column Compartment (Mississauga, Canada) connected to an AB SCIEX QTRAP 5500 hybrid linear ion trap triple quadrupole mass spectrometer equipped with a Turbo Spray source (Concord, Canada). The mass spectrometer was operated in positive ionization mode, and data were acquired using the Analyst 1. 5 www.nature.com/scientificreports/ A Waters Acquity UHPLC BEH C18 column (1.7 µm particle, 2.1 × 100 mm; Mississauga, Canada) was used for chromatographic analysis. The mobile phase was composed of 0.1% formic acid in deionized water (Solvent A) and 0.1% formic acid in methanol (Solvent B). The flow rate was 200 µL/min with 15% solvent B as initial condition (t = 0 min), increasing to 60% solvent B to t = 4 min, then increasing to 85% solvent B to t = 6 min, then held for 1.5 min until t = 7.5 min. The gradient was then reverted back to original conditions of 15% solvent B from t = 7.6 min and stabilized for 1.5 min before the next injection. The total run time was 9 min. The injection volume was 10 µL.
The mass spectrometer was operated with electrospray ionization (ESI) in multiple reaction monitoring (MRM) mode with the following parameters: ionization voltage (4500 V), source temperature (450 °C), curtain gas (30 units), ion source gas 1 (40 units), ion source gas 2 (60 units), and collision gas was set to high. Nitrogen was used for all gases. Both Q1 and Q3 quadrupoles were at unit mass resolution, entrance potential was 10 and dwell time was 150 ms.
Identification and quantification of l-THP. The following method was developed to identify l-THP, as distinct from other tetrahydroprotoberberines and related small molecules with similar molecular weights (Table 1). A UHPLC/MS/MS full scan was performed using direct infusion of l-THP (10 ng/mL) into the mass spectrometer, which generated a mass spectra consistent with the known fragmentation pattern of l-THP molecules 74  Quantification of l-THP levels in plasma samples was based on a 5-point external calibration of l-THP. The standards were prepared in blood plasma sampled from normal untreated rats, processed in the same manner as experimental samples (i.e., ultra-filtration) and analyzed under the UHPLC/MS/MRM conditions described above. Quantification of l-THP levels in cerebrospinal fluid samples were based on a 6-point external calibration of l-THP standards prepared in cerebrospinal fluid from normal untreated rats and analyzed using the same UHPLC/MS/MRM conditions. Data presentation and statistical analyses. Microdialysis data. DA and DOPAC data for each subject were normalized to baseline values, with 100% representing mean DA or DOPAC content in the three samples prior to treatment. For statistical purposes, the mean value during Time 0-150 min was calculated for each subject. Between-group comparisons were conducted using a one-way analysis of variance (ANOVA). Within-group comparisons against baseline (mean of values during Time − 30-0 min) were conducted using a one-way repeated measures (RM) ANOVA. These were followed when appropriate by the Holm-Sidak method of multiple comparisons (overall significance level was p < 0.05). Basal DA levels were determined as the mean of content of DA during the baseline period (final 30 min), and the Kruskal-Wallis one-way ANOVA on ranks test was used to determine between-group differences.
l-THP levels in plasma and cerebrospinal fluid. Analysis of plasma levels of l-THP involved a one-way RM ANOVA followed by Holm-Sidak method of multiple comparisons (overall significance level was p < 0.05). Statistical analyses of cerebrospinal fluid levels of l-THP involved a two-tailed t-test.
Somatic withdrawal signs. Between-group comparisons of the incidence of individual withdrawal signs were conducted using a two-sample Poisson ratio test (two-tailed, 95% CL).
The 'box plot' method of identifying outliers was applied to all data sets 75 . SigmaPlot (version 12.3) and SigmaXL (version 9) was used to perform the statistical analyses.

Data availability
All data generated or analysed during this study are included in this published article (and its Supplementary  Information files).