Morphine coordinates SST and PV interneurons in the prelimbic cortex to disinhibit pyramidal neurons and enhance reward

Opioids, such as morphine, are clinic analgesics which induce euphoria. Morphine exposure modifies the excitability and functional interactions between neurons, while the underlying cellular and molecular mechanisms, especially how morphine assembles heterogeneous interneurons (INs) in prelimbic cortex (PrL) to mediate disinhibition and reward, are not clear. Using approaches of optogenetics, electrophysiology, and cell type-specific RNA-seq, we show that morphine attenuates the inhibitory synaptic transmission from parvalbumin+ (PV)-INs onto pyramidal neurons in PrL via μ-opioid receptor (MOR) in PV-INs. Meanwhile, morphine enhances the inhibitory inputs from somatostatin+ (SST)-INs onto PV-INs, and thus disinhibits pyramidal neurons via δ-opioid receptor (DOR)-dependent Rac1 upregulation in SST-INs. We show that MOR in PV-INs is required for morphine-induced behavioral sensitization, while DOR as well as Rac1 activity in SST-INs is required for morphine-induced conditioned place preference and hyper-locomotion. These results reveal that SST- and PV-INs, functioning in PrL as a disinhibitory architecture, are coordinated by morphine via different opioid receptors to disinhibit pyramidal neurons and enhance reward.


Introduction
Addictive drug-induced long-lasting modifications in the brain are associated with the neuronal plasticity in reward circuits, which depends on dopamine neurons in the ventral tegmental area (VTA) and their downstream targets, including nucleus accumbens (NAc), anterior cingulate cortex, medial prefrontal cortex (mPFC), etc. [1,2]. mPFC responds diversely to reward-predictive cues and is critical for motivated behaviors [3]. Prelimbic subregion of mPFC (prelimbic cortex, PrL) has been implicated in regulation and innervation of addictive processes [4][5][6][7][8][9]. Both VTA and NAc are main recipients of the glutamatergic inputs from the PrL, indicating this nuclei is crucial for reward processing and the expression of drug-induced sensitization [3,9,10].
The information processing of cortical circuits depends on the functional interactions between the excitatory and inhibitory connectivity. Diverse types of GABAergic interneurons (INs) can receive, integrate, and encode information to stringently control the projective outputs [11]. Somatostatin (SST) and parvalbumin (PV) INs are two major subtypes of inhibitory neurons in PrL of rodent and human cortex, and they target distal dendritic and perisomatic regions of postsynaptic excitatory neurons, respectively, to exert their distinct inhibitory effects on excitatory neurons [12][13][14][15]. Functional connections between different subtypes of INs are observed and thus INs collaboratively regulate information integration in neural network [16][17][18]. The plasticity of PV-INs changes after fear conditioning and environment enrichment, and the dysfunction of PV-INs is found in schizophrenia mouse model [19,20]. Abnormality of SST-INs in cortex is found in the Alzheimer's mouse model [21]. These data indicate that the plasticity of distinct INs in cortical circuits participates in the process of cognition and psychiatric disorders.
Opioids, such as morphine, are clinical effective analgesics, but they also induce euphoria and adaptive changes of reward circuits [22]. Morphine acts through Gprotein coupled opioid receptors to modulate presynaptic and postsynaptic ion channels [23][24][25][26] and disinhibit the inhibitory control to modulate pain and reward [27,28]. Recent studies indicate that morphine exposure causes complex modifications of anatomical and functional connections in reward circuits, especially between excitatory neurons [29]. However, the effect and the biological basis of opioids on the connectivity among the excitatory and the inhibitory neurons in PrL, are not fully understood.
In this study, we combined morphological tracing, electrophysiology, and optogenetic methods, investigated morphine-induced alterations of inhibitory inputs from SST-INs and PV-INs onto local pyramidal neurons. Our results reveal a disinhibitory architecture consisting of SSTand PV-INs in the PrL, which is coordinated by morphine via different subtypes of opioid receptors to disinhibit pyramidal neurons, thus enhance morphine reward and related behaviors.

Morphine attenuates the inhibitory transmission to pyramidal neurons from PV-INs, but not SST-INs in PrL via a MOR-dependent pathway
To investigate morphine-induced disinhibition of pyramidal neurons mediated by different INs in PrL, PV-INs or SST-INs expressing hChR2-H134R were activated by serial laser pulses, and the responsive inhibitory postsynaptic currents (IPSCs) were measured in nearby pyramidal neurons (Fig. 1a, b). The responsive probability was comparable upon PV-INs or SST-INs activation (Fig. 1c, d). All lightevoked IPSCs could be abolished by 20 mM bicuculline (Fig. 1e, g). Morphine treatment (10 mg/kg, i.p.) significantly reduced the peak amplitude and the half-width of responsive IPSCs upon optogenetic activation of PV-INs, while did not change the 10-90% rise time or synaptic latency (Fig. 1e, f, and Supplementary Fig. 1a). Optogenetic activation of SST-INs in PrL evoked much weaker IPSCs in the nearby pyramidal neurons than activation of PV-INs, while the responsive probability and the peak amplitude of responsive IPSCs in pyramidal neurons were not changed by morphine exposure (Fig. 1g, h).
To investigate whether μ-opioid receptor (MOR) in the INs mediates the disinhibitory effect of morphine on PrL pyramidal neurons, we co-infected AAV-DIO-(hChR2-H134R)-mCherry with AAV-Flex-MOR-shRNA-EGFP or AAV-Flex-Scramble-shRNA-EGFP in the PrL of PV-Cre or SST-Cre mice (Fig. 1a). The downregulation of MOR in PV-and SST-INs was evaluated by immunostaining (Supplementary Fig. 2a-c). Laser evoked action potentials (APs) in PV-INs and SST-INs were not different between the Scramble or MOR-shRNA expressing cells ( Supplementary  Fig. 2d, e). Selective expression of MOR-shRNA in PV-INs abolished the inhibition by morphine on responsive IPSC amplitude in pyramidal neuron to laser stimulation of PV-INs (Fig. 1e, f). However, conditional knockdown of MOR in SST-INs had no effect on the response probability or the properties of the responsive IPSCs in pyramidal neurons to laser activation of SST-INs (Fig. 1d, g, h, and Supplementary Fig. 1b).
These results show that there is a broad connectivity between pyramidal neurons and PV-or SST-INs in the PrL, and PV-INs are able to evoke larger IPSCs in nearby pyramidal neurons than SST-INs. Morphine decreases the strength of synaptic inputs from PV-INs, but not SST-INs onto pyramidal neurons via a MOR-dependent pathway.

Morphine increases neurite complexity of SST-INs and inhibitory transmission onto PV-INs in PrL
Addictive drugs are shown to regulate the density of dendritic spines and the electrophysiological activity of neurons in the mPFC [30]. To evaluate morphine-induced morphological changes in PrL INs, we used reporter mice (SST-Cre::EYFP and PV-Cre::EYFP) and injected a fluorescent dye into SST-INs and PV-INs for morphological tracing ( Supplementary Fig. 3a). The results showed that 12 h after a single or five consecutive morphine injections, the neurite complexity (Sholl intersections) and total neurite length of SST-INs were significantly increased compared with the saline control group ( Supplementary  Fig. 3b, c). However, no significant difference in neurite complexity (Sholl intersections) and total neurite length in PV-INs were detected after morphine exposure (Supplementary Fig. 3d, e).
To examine synaptic transmission in these INs in PrL, we performed whole-cell patch-clamp recordings in PV-INs or SST-INs 12 h after saline or morphine exposure (Supplementary Fig. 4a). The amplitude of miniature excitatory postsynaptic currents (mEPSCs) recorded from SST-INs and PV-INs were both moderately decreased, while no difference in the frequency of mEPSCs was observed after morphine exposure ( Supplementary Fig. 4b-e). Morphine exposure did not affect the amplitude or frequency of the miniature IPSCs (mIPSCs) in SST-INs (Fig. 1i, k, m), but increased both the frequency and the amplitude of mIPSCs in PV-INs (Fig. 1j, l, n), indicating that morphine enhances inhibitory inputs onto PV-INs in PrL. In addition, morphine increased the number of induced spikes in SST-INs ( Fig. 1o, p), while decreased the number of induced spikes in PV-INs (Fig. 1q, r). The intrinsic electrophysiological characteristics of SST-INs and PV-INs did not change after morphine treatment (Supplementary Table 1).
These results suggest that morphine differentially regulates neurite complexity, inhibitory synaptic transmission, and membrane excitability of SST-INs and PV-INs in PrL.

Morphine enhances the inhibitory synaptic transmission from SST-INs to fast-spiking (FS) PV-INs in PrL
Accumulating evidence indicates that the disinhibitory microcircuits in the cortex involve interactions among different subtypes of INs [31][32][33][34]. We thus investigated whether morphine regulates the strength of synaptic inputs from SST-INs onto PV-INs. We took advantage of LhX6-EGFP transgenic mice, in which the majority of INs derived from the medial ganglionic eminence are labeled with EGFP [35,36], and bred this transgenic line with SST:: tdTomato line (LhX6-EGFP/SST-tdTomato alleles) to distinguish SST-INs (tdTomato + ) from other types of INs (EGFP + /tdTomato − ) (Fig. 2a).
Immunostaining result showed that 56.62% of EGFP + /tdTomato − INs is PV positive (Fig. 2b). Since FS is the most prominent electrical property of PV-INs, we recorded EGFP + /tdTomato − INs with FS character. To examine synaptic transmission from SST-INs onto FS PV-INs, AAV-DIO-(hChR2-H134R)-mCherry was infected in the PrL of LhX6-EGFP/SST-Cre mice, and EGFP + /tdTomato − FS-INs nearby tdTomato + SST-INs were recorded upon optogenetic activation of SST-INs (Fig. 2c, d). The connection probability did not change after morphine exposure (Fig. 2e). The responsive amplitude, but not the 10-90% rise time or half-width of light-evoked IPSCs in FS-INs, was increased 12 h after morphine exposure (Fig. 2f-h), suggesting that morphine exerts long-lasting effect on the inhibitory synaptic transmission from SST-INs to FS-INs.
We next assessed the presynaptic release probability by analysis of paired-pulse ratio (PPR) and the coefficient of variation in FS PV-INs upon optogenetic stimulation of SST-INs. Morphine significantly reduced PPR ratio at the second and third stimulations (Fig. 2i, j). The coefficient of variation of responsive IPSCs did not significantly change (Fig. 2k). Combined with result of the increased mIPSCs frequency in PV-INs after morphine exposure, these data suggest that morphine increases the presynaptic release probability of SST-INs to PV-INs, and indicate that morphine-enhanced inhibitory GABAergic transmission from SST-INs onto PV-INs involves presynaptic mechanisms.
Cell type-specific RNA-seq reveals that morphine upregulates Rac1 pathway specifically in SST-INs, but not PV-INs of PrL Given the observed difference of morphine-induced neuronal plasticity between SST-INs and PV-INs, the neuronalspecific molecular mechanisms between these two subtypes of neurons were assessed. SST-Cre and PV-Cre mice were crossed to RPL22-HA reporter mice to produce mice that express HA-tagged ribosomal protein (ribotag) specifically in SST-or PV-INs (Fig. 3a). Transcripts associated with ribosomes (in the process of de novo protein synthesis) were isolated from PrL 12 h after saline or morphine injection and sequenced (Fig. 3b). Analysis of the ribosome-associated transcripts in these two groups showed that Pvalb and Sst were respectively enriched in PV-INs and SST-INs ( Supplementary Fig. 5a), indicating the successful enrichment of interneuron subtype-specific transcripts. Morphine injection induced more transcriptional alterations in SST-INs, compared with PV-INs (translational change in 1558 vs. 328 genes; Fig. 3c and Supplementary Fig. 5b).
To infer potential intracellular pathways altered by morphine, ClueGO was used for regulatory network construction of these two subtypes of INs. The analysis indicates that SST-INs exhibited significant differences in genes involved in the pathways of cAMP and insulin signaling, addiction, and neurotrophic factor signal transduction, etc., which have  Morphine binds preferentially to MOR, while prolonged stimulation of neurons with morphine, both in vitro and in vivo, markedly increases recruitment of intracellular δ-opioid receptor (DOR) to the cell surface [40]. To explore the potential role of MOR and DOR in morphine-induced upregulation of Rac1 and Arhgef6 mRNAs in SST-INs, we infected AAV-Flex-DOR-shRNA-EGFP or AAV-Flex-MOR-shRNA-EGFP into PrL of SST-Cre mice. Cre-dependent DOR downregulation in SST-INs and PV-INs was verified by RNAscope ISH (Supplementary Fig. 7). We found that 12 h after morphine exposure, the fluorescent intensity of Rac1 and  Fig. 8). Moreover, we infected Cre-dependent Rac1-DN and hChR2-mCherry viruses into the PrL of LhX6-EGFP/SST-Cre mice, and performed whole cell recordings in FS PV-INs or pyramidal neurons ( Fig. 4i, m). Expressing Rac1-DN in SST-INs did not affect the light-evoked responsive probability in FS PV-INs (Fig. 4k), while attenuated the light-evoked responsive amplitude in PV-INs (Fig. 4j, l). However, expressing Rac1-DN in SST-INs decreased the light-evoked responsive probability (Fig. 4o), while did not affect the responsive amplitude in pyramidal neurons (Fig. 4n, p). In addition, pyramidal neurons nearby the SST-INs were significantly disinhibited by morphine, reflected by the increased current pulses, while the increased current pulses were abolished by expressing Rac1-DN in SST-INs (Fig. 4q, r). These data indicate that morphine increases the strength of the inhibitory transmission from SST-INs onto PV-INs via DOR and Rac1 signaling in SST-INs, and may thus attenuate the inhibitory effect of PV-INs onto pyramidal neurons in PrL. Kolmogorov-Smirnov test for cumulative frequency. Data are presented as mean ± SEM; *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001

Distinct opioid receptor pathways in PrL SST-INs and PV-INs mediate morphine-conditioned place preference (CPP) and behavioral sensitization
Prefrontal cortex has been implicated in the reward processing and the development of addictive-drug-induced behavioral sensitization [3]. To investigate the effect of MOR and DOR signaling pathways in specific INs on the rewarding properties and locomotor-activating effects of morphine, SST-Cre and PV-Cre mice were infected with AAV-Flex-MOR-shRNA-EGFP, AAV-Flex-DOR-shRNA-EGFP or AAV-Flex-Scramble-shRNA-EGFP in PrL (Fig. 5a). Results showed that knockdown of DOR in SST-INs significantly inhibited morphine-induced CPP and hyper-locomotion, while knockdown of MOR in SST-INs had no such effects (Fig. 5b, c). We then injected AAV-EF1α-DIO-Rac1-DN-mCherry or AAV-EF1α-DIO-mCherry into the PrL of SST-Cre and PV-Cre mice (Fig. 5d). Expressing Rac1-DN in SST-INs abolished morphine-induced CPP and hyper-locomotion (Fig. 5e, f), while expressing Rac1-DN in PV-INs had no effect (Supplementary Fig. 9). These results indicate that DOR-Rac1 pathway in the SST-INs, but not in the PV-INs, is involved in the reward properties and locomotor-activating effects of morphine.
Knockdown of MOR in PV-INs had no effect on morphine-induced CPP, but decreased locomotor activity 90 min after the initial morphine injection (Fig. 5h, i) and behavioral sensitization after repeated morphine exposures ( Supplementary Fig. 10), suggesting that MOR of PV-INs in PrL is involved in both the initiation and expression of behavioral sensitization to morphine. Knockdown of DOR in PV-INs had no effect on morphine-induced CPP and hyper-locomotion (Fig. 5h, i).
Taken together, these data reveal that DOR-Rac1 pathway in SST-INs is required for morphine-induced CPP and hyper-locomotion, while MOR pathway in PV-INs is involved in behavioral sensitization, and indicate that morphine, via distinct opioid receptors, coordinates an architecture consisting of SST-INs and PV-INs to disinhibit pyramidal neuron in PrL and enhance reward.

Discussion
GABAergic INs, as a minority of the cortical neuronal population in the forebrain, are crucial in fine-tuning cortical microcircuits. The interactions of GABAergic INs with excitatory glutamatergic neurons maintain balanced electrical activity and normal cortical functions [41]. Morphine activates presynaptic GABAergic neurons in VTA and disinhibits dopaminergic neurons, increasing dopamine release and inducing reward [42]. However, the molecular targets and the inhibitory architecture recruited by morphine to promote reward and behavioral sensitization are unclear.
PV-INs includes FS basket and chandelier cells. FS-INs are the largest population of INs in the neocortex. They regulate action potential firing and form complex structural contacts between themselves to promote synchronization of electrical activity [14,43,44]. In contrast to PV-INs, SST-INs are dendritic targeting and they mediate double-synapse inhibition on nearby pyramidal neurons [45]. We examined the amplitude of responsive IPSC from PV-INs and SST-INs onto pyramidal neurons by optogenetic stimulation, and observed that as compared with SST-INs, PV-INs showed stronger predominant inhibitory inputs onto pyramidal  (Fig. 1). SST-INs showed an increased intrinsic membrane excitability, while PV-INs showed a decreased intrinsic membrane excitability upon morphine exposure. Consistently, SST-INs exhibited more translational and morphological changes upon morphine exposure ( Fig. 3 and Supplementary Fig. 3), indicating that morphine-induced alterations are interneuron subtype-specific and likely via differential modulatory mechanisms.
SST-INs not only innervate pyramidal neurons but also strongly innervate other types of INs in the cortex [33,34,46]. The increased frequency of mIPSC in PV-INs indicates an increased inhibitory input. Utilizing LhX6-EGFP/SST-tdTomato mice, we identified PV-INs by EGFP + /tdTomato − labeling and FS features to study the synaptic transmission from SST-INs to FS PV-INs. Our data suggest that morphine exposure increases the inhibitory transmission from SST-INs to FS PV-INs in PrL and this is dependent on a presynaptic mechanism. We postulate that after morphine stimulation, SST-INs extend their neurites to form extensive reciprocal connections and stronger inhibition to adjacent FS PV-INs, and this is accompanied with reduced FS PV-INs firing, leading to disinhibition of nearby pyramidal neurons and behavioral changes.
Morphine exhibits affinity to MOR, DOR, and KOR subtypes, but has higher affinity to MOR and thus preferentially binds to MOR [47]. The results from MORand DOR-knockout mice indicate that MOR is essential for both of analgesia and tolerance of morphine [48], while DOR is required for the development of sensitization and tolerance to the locomotor-activating effects of morphine [49]. Previous histological data showed that MORs are expressed prominently in PV-INs, whereas DORs are expressed prominently in SST, neuropeptide Y, and corticotrophin releasing factor (CRF) INs, as well as in pyramidal neurons in hippocampus [50]. It is interesting to know how morphine remodels the circuits to promote reward by activating signaling pathways mediated by different opioid receptors in different INs.
In this study, we focus on MOR and DOR signaling and the inhibitory transmission from PV-and SST-INs onto pyramidal neurons in PrL. Our results reveal that morphine exerts its disinhibition function on pyramidal neurons via neuronal subtype-and opioid receptor-specific signaling pathways. The activation of MOR-signaling pathway in PV-INs by morphine attenuates the inhibitory inputs to pyramidal neurons directly, while activation of DOR signaling pathway in SST-INs by morphine enhances the inhibitory inputs to PV-INs and thus further disinhibits pyramidal neurons nearby (Fig. 5j). Our results showed that morphine specifically increases the neurite complexity and upregulates  Research on the structural and functional connectivity between inhibitory INs and pyramidal neurons is important for understanding on how inhibitory architecture in PrL gates neuronal network excitability. The circuitry for behavioral sensitization includes glutamatergic projections from the mPFC to the NAc [3,51], and the prefrontal glutamate release into NAc mediates drug-seeking behaviors [52]. The output circuits of PrL guide conditioned reward seeking through divergent PFC → NAc and PFC → PVT encoding [53]. mPFC → NAc population dynamics predict individual reward seeking or suppression decision [54]. These results indicate that glutamatergic projection in PrL, which is modulated precisely by local INs, is required for behavioral sensitization and reward. Our data indicate that morphine remodels SSTand PV-interneuron plasticity, which in turn induces behavioral changes via distinct molecular pathways in the two types of INs. The MOR and DOR signaling pathways in the INs play important regulatory roles in behavioral sensitization and reward processing. . These mice were bred to C57BL/6 J for more than 6 generations. SST-Cre::EYFP or PV-Cre:: EYFP alleles were generated by crossing SST-Cre or PV-Cre mice with Rosa 26 reporter mice; SST-Cre::RPL22-HA or PV-Cre:: RPL22-HA alleles were generated by crossing SST-Cre or PV-Cre mice with Ribotag mice; SST-Cre::tdTomato alleles were generated by crossing SST-Cre mice with Ai14 mice; LhX6-EGFP/SST-Cre:: tdTomato mice were generated by crossing LhX6-EGFP mice with SST-Cre::tdTomato mice; LhX6-EGFP/SST-Cre were generated by crossing LhX6-EGFP mice with SST-Cre mice. 6-10-week-old male offsprings were used in the experiments, and randomly assigned to groups. Mice used for the experiments were housed in groups on a 12 h light/dark cycle (light on from 8 a.m. to 8 p.m.) with access to food and water ad libitum. All experiment procedures were strictly in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals, and were approved by Animal Care and Use Committee of the animal facility at Fudan University.

Stereotaxic surgery
Mice were anesthetized with 2% isoflurane and placed in a stereotactic instrument (Stoelting, Kiel, WI, USA). Microinjections were performed using 33-gauge needle connected to a 10 μl Hamilton syringe. The intended stereotaxic coordinates for PrL were: AP + 2.0 mm; ML ± 0.3 mm (with an angle of 14°from the middle to the lateral); DV -2.0 mm. Each site was injected with 0.5 μl of purified and concentrated AAV (10 12 IU/ml) with a slow injection rate (0.1 μl/min). All mice were given at least 3 weeks to recover before behavioral experiments or electrophysiological recordings, and the efficiency of viral infection and the shRNA knockdown was verified by immunostaining. The histology slides were examined blindly to check the expression of EGFP or mCherry in PrL. Only the mice with virus infection in correct place were chosen for further analysis.
For light-evoked postsynaptic currents whole-cell recordings, ChR2 was excited with a 473 nm LED blue light source (Xcite-110) delivered through the epifluorescence pathway and fed into a 60× water-immersion objective lens (Olympus BX51). Photo-stimulation (2-6 mW/mm 2 , 1-2 ms duration) was controlled by a TTL input (HEKA Instruments). Pyramidal neurons or FS INs were clamped at −70 mV. For recordings, the pipettes were filled with intracellular solution (60 mM K-gluconate, 66 mM KCl, 2 mM MgCl 2 , 10 mM HEPES, 0.2 mM EGTA, 4.5 mM MgATP, 0.5 mM Na 3 GTP, 10 mM Na-phosphocreatine, 0.25% neurobiotin, pH 7.25, 300 mOsm). D-APV (100 mM) and CNQX (20 mM) were added to block excitatory currents. A train of light pulses (10 Hz) was delivered to the presynaptic cells, and the postsynaptic responses were recorded 20-30 repeated trails at 15 or 20 s interval. Data were analyzed off-line with Clampfit 10.3 (Molecular Devices, Union City, CA, USA) or Mini Analysis Program (Synaptosoft, Fort Lee, NJ, USA). The peak amplitude was calculated by subtracting the baseline. The synaptic latency was determined as the duration from current onset time to peak time. The rise time of evoked IPSCs was assessed from the 10 to 90% rising phase, and the half-width of evoked IPSCs was defined as the duration at the half amplitude. For the paired-pulse ratio calculation, the averaged peak amplitude of the first IPSC was defined as the basal level of synaptic strength. The variable coefficient was assessed from the amplitudes of each sweep.

Morphological reconstruction and quantitation
For neuronal reconstruction and morphological analysis, SST and PV INs in PrL were randomly selected, patched, and filled with 2% lucifer yellow (L0259, Sigma-Aldrich, St Louis, WA, USA) for at least 10 min. After 10 min additional diffusion, slices were fixed in 4% PFA in 0.1 M phosphate-buffered saline (PBS) overnight. Sections were blocked in 10% serum and 0.1% Triton-X in PBS, and incubated with an anti-Lucifer Yellow antibody (Invitrogen, #A-575c, Carlsbad, CA, USA) overnight at 4°C. Z-series images were taken at 2 μm interval using an Olympus FV1000 confocal laser scanning microscope with a 60× objective (Olympus). Full cell 3-dimensional reconstructions and analysis were made by Neurolucida (Micro-BrightField, Williston, VT, USA).

Immunohistochemistry
Mice were anesthetized with isoflurane and perfused with saline followed by 4% paraformaldehyde in 0.1 M PBS. The brains were removed, fixed in 4% paraformaldehyde overnight and subjected to dehydration in 30% sucrose at 4°C for 72 h before slicing 30 μm per slice. Slices were incubated with diluted antibodies in blocking solution containing 0.2% Triton X-100 (Sigma-Aldrich) and 3% goat serum (Jackson ImmunoResearch, West Grove, PA, USA) at 4°C overnight. The primary antibodies used were: anti-SST (Santa Cruz, #sc-47706, Dallas, TX, USA), anti-PV (Merck Millipore, #MAB1572, Darmstadt, Germany) and anti-MOR (Abcam, #ab10275, Cambridge, MA, USA). Slices were rinsed in 0.1 M PBS then incubated in Cy3 antimouse, Alexa 647 anti-rat or Cy3 anti-rabbit IgG antibodies (Jackson ImmunoResearch) for 1 h at room temperature, then mounted after rinsing with 0.1 M PBS. Images were acquired on a Nikon A1 microscope (Tokyo, Japan) using 20× air or 60× oil objective lens. The observer analyzing the expression of MOR in SST or PV INs was blinded to the group allocation.

Ribo-tag purification
Purification of ribosome-associated mRNA was performed as described previously with slight modification [57]. Mice were decapitated, and the brains were removed immediately. The PrL were dissected in ice-cold PBS. The brain tissue was homogenized in 1 ml Supplemented Hybridization Buffer (25 mM Tris pH 7.0, 25 mM Tris pH 8.0, 12 mM MgCl 2 , 100 mM KCl, 1% Triton X-100) containing 1 mM DTT, 1 × protease inhibitors (Roche, Upper Bavaria, Germany), 200 U/ml RNase inhibitor (Promega, Madison, WI, USA), 100 μg/ml cycloheximide (Cayman, Ann Arbor, MI, USA), and 1 mg/ml heparin (Sigma-Aldrich). The supernatant was incubated with 10 μg anti-HA antibody (Sigma-Aldrich, #H6908) and 100 μl Dynabeads Protein G (Invitrogen) for 12 h. Purified mRNA was eluted from the Dynabeads using TRIzol LS (Invitrogen) according to the manufacturer's instructions with the inclusion of a DNase digestion step. The Agilent RNA 6000 Pico Kit (Agilent, Santa Clara, CA, USA) and Agilent 2100 bioanalyzer were used to evaluate the quality of purified mRNA. Samples with RIN number > 7 were used.
Next-generation sequencing mRNA was enriched using NEB Next Poly(A) mRNA Magnetic Isolation Module (NEB, E7490S, Ipswich, MA, USA). Library was prepared with NEB Next Ultra RNA Library Prep Kit (E7530S) and sequenced on a HiSeq 4000 (Illumina) by Novogene Technology Co. Ltd (Beijing, China). Raw reads were quality checked and trimmed with FASTX-toolkit to remove adapter contamination and lowquality reads (quality score < 28). The clipped reads were aligned to mouse reference sequence (GRCm38/mm10) using HISAT2. Mapped reads for each transcript were counted using HTseq and differential expression analysis was performed with DESeq2. Genes with more than twofold expression changes, and were significantly different (P < 0.05) were selected for further analysis. ClueGo [58] was used for signaling pathway and network construction.

Reverse transcription, and quantitative real-time PCR (qRT-PCR)
Reverse transcription was completed using the PrimeScript RT reagent Kit (RR037A, Takara Biotechnology, Dalian, China). The cDNA was subjected to qRT-PCR using SYBR Premix Ex Taq (RR420A, Takara) and Eppendorf Mastercycler PCR System (Eppendorf, Hamburg, Germany). The primers are listed in Supplementary Table 2.
Locomotion test An activity monitor system (43.2 cm length × 43.2 cm width × 30.5 cm height, Med-Associates, St. Albans, VT, USA) was used to detect morphine-induced locomotor activities and behavioral sensitization. Each mouse was placed in the center of the open field and allowed to explore freely for 30 min (baseline). After given an intraperitoneal injection of morphine (10 mg/kg) (Shenyang 1st Pharmaceutical Company, Shenyang, China), the mice were confined to the open field for 120 min. To evaluate morphineinduced behavioral sensitization, mice were placed in chamber for 10 min, then injected with morphine (10 mg/kg, i.p.) and placed in chamber for 1 h. The total distance traveled was recorded.

Conditioned place preference
Morphine-induced CPP was performed using a two chamber (15 × 15 × 20 cm) apparatus with distinct tactile environments to maximize contextual differences. A manual guillotine door (15 × 20 cm) separated the two chambers. The observer was blinded to the group allocation. On the first day, mice were allowed to freely explore the entire apparatus for 15 min (pretest). The mice staying in one chamber for more than 10 min were excluded from the experiment. From the second to the sixth days, mice were daily given an intraperitoneal injection of morphine (10 mg/kg, i.p.) and confined to one of the chambers (drugpaired) for 30 min, and 6 h later, they received an i.p. injection of saline (equivalent volume to that of morphine) and confined to the other chamber for 30 min (conditioning). On the seventh day, mice were allowed to freely explore the entire apparatus for 15 min (test). The time spent in each chamber was recorded during the pretest and test sessions. CPP score was defined as the time (in seconds) spent in morphine-paired chamber minus the time spent in saline-paired chamber.

Statistical analysis
Data were analyzed with SPSS 20 software (IBM, Armonk, NY, USA). Sample size estimation was conducted on alpha value of 0.05 and desired power of 0.80. Comparisons between groups were made by unpaired or paired two-tailed student's t test, Mann-Whitney U test, χ 2 test, one-way ANOVA, or twoway ANOVA. Two-sample Kolmogorov-Smirnov test was used for analyzing the cumulative distribution. Results of locomotion and neuronal excitability were analyzed by twoway repeat-measure (RM) ANOVA followed by the Bonferroni's post-hoc test. Statistical significance was represented as *P < 0.05; **P < 0.01; ***P < 0.001, and ****P < 0.0001. All data are presented as mean ± SEM.

Data availability
All data needed to evaluate the conclusions in the paper are present in the paper and/or the supplementary materials. Additional data related to this paper may be requested from the authors. Raw and processed NGS data are deposited in the National Center for Biotechnology Information Bio-Project database under accession number (PRJNA508422). (2018YFC1004500 to QL), Shanghai Municipal Science and Technology Major Project (2018SHZDZX01 to LM and FW) and ZJLab.
Author contributions FW and LM designed and supervised the study. CJ, FW and LM planned the experiments and analyzed the data. CJ carried out the electrophysiology experiments. CJ, FW and GH carried out the immunohistochemical experiments and RNAscope ISH. QL and CL carried out the ribo-tag purification and bioinformatics analysis. XW and PL contributed to the acquisition of the behavioral data. XW carried out the morphological reconstruction and quantitation. ZW constructed the AAV vectors. CJ drafted the paper. FW, PZ and LM revised the paper.

Compliance with ethical standards
Conflict of interest The authors declare that they have no conflict of interest.
Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons. org/licenses/by/4.0/.