Supp

Sep 13, 2023

Abstract With concurrent global epidemics of chronic pain and opioid use disorders, there is a critical need to identify, target and manipulate specific cell populations expressing the mu-opioid receptor (MOR). However, available tools and transgenic models for gaining long-term genetic access to MOR+ neural cell types and circuits involved in modulating pain, analgesia and addiction across species are limited. To address this, we developed a catalog of MOR promoter (MORp) based constructs packaged into adeno-associated viral vectors that drive transgene expression in MOR+ cells. MORp constructs designed from promoter regions upstream of the mouse Oprm1 gene (mMORp) were validated for transduction efficiency and selectivity in endogenous MOR+ neurons in the brain, spinal cord, and periphery of mice, with additional studies revealing robust expression in rats, shrews, and human induced pluripotent stem cell (iPSC)-derived nociceptors. The use of mMORp for in vivo fiber photometry, behavioral chemogenetics, and intersectional genetic strategies is also demonstrated. Lastly, a human designed MORp (hMORp) efficiently transduced macaque cortical OPRM1+ cells. Together, our MORp toolkit provides researchers cell type specific genetic access to target and functionally manipulate mu-opioidergic neurons across a range of vertebrate species and translational models for pain, addiction, and neuropsychiatric disorders. Introduction Mu-Opioid receptors (MORs) are inhibitory G-protein coupled receptors expressed across molecularly and functionally distinct cell types in the peripheral (PNS) and central nervous systems (CNS), where they act as the primary effector of endogenous opioid peptides to fine-tune neural transmission, cellular excitability, and influence gene transcription 1 . Exogenous opioid compounds, whether employed clinically or recreationally, leverage this evolutionarily conserved MOR system at non-physiological concentrations and timescales leading to intended benefits like analgesia, and unwanted detriments like opioid dependence and respiratory depression 2 . In order to disentangle the underlying mechanisms and specific cells involved in opioidergic functions, there is a need for a widely adoptable, robust, and reliable method that can be employed across numerous in vivo and in vitro model systems for neuroscientific investigations. The isolation and targeting of MORs, and the neuronal cell populations and circuitry that harbor them, initially relied on pharmacological and radiolabeled reagents to visualize and characterize physiological function in diverse species, such as mice, rats, pigeons, and non-human primates 3 , 4 , 5 . Recent improvements in the development of transgenic animals, including conditional Oprm1 knock-out, fluorescently tagged MORs and Cre-recombinase lines 6 , 7 , 8 , 9 , 10 , continue to refine our understanding of opioid receptor neurobiology and provide the ability to perform cellular level manipulations in complex behavioral tasks germane to pain and addictive behaviors, but are limited to rodent model systems. By contrast, recent advances in viral construct design and gene delivery have led to the production of highly efficient vectors for directing transgenes to specific neural structures and cell types by targeting unique genetic elements exclusive to target populations 11 , 12 , 13 , 14 . Taking advantage of these improvements in adeno-associated virus (AAV)-based gene delivery, we created two suites of constructs based on the mouse MOR promoter (mMORp) and the human MOR promoter (hMORp) elements that can transduce fluorescent reporters, chemogenetic actuators, biosensors, and recombinase effector transgenes in MOR+ neuronal cell types across multiple structures within the PNS and CNS. The size and design of these constructs make them amenable to the packaging limits of AAV vectors, and thus able to take advantage of the growing catalog of serotypes available for this vector that increase cell type transduction selectivity 15 . Here, we show that these viral tools are effective in rodents and other mammalian models of interest for the in vivo study of pain, addiction, emesis, and neuropsychiatric disorders involving opioid system dysfunction, including smaller mammals and non-human primates. We further show that these viruses can also be used in vitro when working with human iPSC-derived cell types, collectively demonstrating that these new viral constructs are capable of complementing and enhancing the current techniques and tools available to gain access to opioidergic cell types across germane to both basic and clinical research endeavors 16 . Results Designing viral constructs to target promoter sequences within the Oprm1 gene To design a new promoter system for viral-assisted, selective access to MOR+ cell types, we analyzed a 1.9 Kb genomic region immediately upstream of the translation start codon (ATG) of the mouse mu opioid receptor gene (Oprm1) for the presence of transcriptional initiation elements with complex transcription factor binding topology using PROMO 17 and the Eukaryote Promoter Database 18 (Fig. 1a , Supplementary Fig. 1a, b ). Prior reports on the putative Oprm1 promoter sequence describe a “proximal” (−450 to −249 bp upstream of the ATG site) and a “distal” (−1326 to −794 bp upstream of the ATG site) region, each with transcription start sites (TSSs), which respectively account for 95% and 5% of the overall activity of the gene 19 , 20 . Therefore, we designed four complimentary sequences that covered both the proximal and distal promoter regions within the mouse Oprm1 gene: mMORp1 (−1797 to −265, 1532 bp), mMORp2 (−1900 to −265, 1635 bp), mMORp3 (−1621 to −1, 1620 bp), and mMORp4 (−1621 to −18, 1603 bp; Fig. 1a , Supplementary Fig. 1c ). In addition to these sequences targeted to the murine Oprm1 gene, we also applied our construct design protocol to generate sequences targeted to the promoter region within the human OPRM1 gene, resulting in the production of hMORp1 (−1841 to −257, 1532 bp; Fig. 1a , Supplementary Fig. 1d ). All promoter sequences were assembled into plasmids encoding an enhanced yellow fluorescent protein (eYFP) reporter fluorophore and packaged into serotype 1 AAV vectors (AAV1). Administration of mMORp1-4 and hMORp1 AAVs in vitro into well plates containing primary cultured rat hippocampal neurons at concentrations of 1–3 × 1012 genome copies per milliliter (gc/mL) revealed both mMORp1 and hMORp1 constructs to robustly transduce cells, as evidenced by the presence of the eYFP reporter (Supplementary Fig. 2a, b ). The four mMORp and single hMORp constructs were then tested in vivo via intracranial injection at similar concentrations into the cortex of C57BL/6J mice. The AAV1-mMORp1-eYFP and AAV1-hMORp1-eYFP constructs once again produced robust transduction and expression of reporter eYFP, while no evidence of successful transduction was noted in tissue from animals injected with mMORp2-4, possibly due to the neuron-restrictive silencer element (NRSE) and transcription factor Sp1/Sp3 binding sites near the ATG site 21 . Further testing of mMORp1 and hMORp1 demonstrated similar successful transduction of other brain regions in vivo, including the central nucleus of the amygdala (CeA) and periaqueductal gray (PAG) (Supplementary Fig. 2c–g ). Using mMORp1 as a template, we then designed several constructs to drive the expression of different transgenes for cell labeling and circuit dissection approaches within MOR+ cell types (Fig. 1b ). These variants included constructs encoding a genetically encoded calcium indicator (mMORp-GCaMP6f), a chemogenetic inhibitor (mMORp-hM4Di-mCherry), a Cre recombinase (mMORp-mCherry-IRES-Cre), a Flp recombinase (mMORp-FlpO), and the red fluorescent protein oScarlet (mMORp-oScarlet), which can be used as a reliable alternative to eYFP for labeling putative MOR+ cells (Supplementary Fig. 2i ). Fig. 1: Development and validation of murine and human mu opioid receptor promoter (MORp) driven viral constructs. a DNA sequence for the murine Oprm1 (upper) and human OPRM1 (lower) promoter regions, including the approximate locations of several transcriptional elements such as repressor and activator transcription factors (TFs), transcription start sites and promoter elements, as determined via PROMO and Eukaryotic Promoter Database & UCSC Genome Browser analyses of the murine and human genes (Supplementary Fig. 1 ). The promoter region encoded by the four murine promoter constructs (mMORp1-4) and single human promoter construct (hMORp) are depicted beneath each sequence map. b mMORp and hMORp construct designs and packaging schema within adeno-associated viral (AAV) vectors of multiple different capsid serotypes. c Transduction efficacy from initial in vivo intracranial injections of the mMORp1-4-eYFP constructs into C57BL/6J mouse medial prefrontal cortex (mPFC), scale bar = 500 μm. d mPFC expression pattern with AAV1-mMORp1-eYFP across mPFC subregions, including the cingulate (Cg1), prelimbic (PL) and infralimbic (IL) cortex, scale bar = 200 μm. e Higher magnification images of the mPFC following transduction with the mMORp1 viral construct. Amplification of the eYFP signal, along with staining for both neuronal (NeuN) and microglial (Iba1) markers demonstrate selective transduction of neurons, with staining for additional glial markers to further verify this shown in Supplementary Fig. 5 . Cortical layer division markers (Layers 1–6) highlight viral spread and efficiency, scale bar = 100 μm. f Overlap of mMORp1-eYFP viral expression and endogenous mu opioid receptor (MOR) immunoreactivity within the central amygdala (CeA) (denoted by blue outline), but not surrounding amygdalar subregions of a C57BL/6J mouse, using a knock-out mouse-validated anti-MOR antibody (Supplementary Fig. 3 ). Basolateral amygdala (BLA), medial anterodorsal amygdala (MeAd), medial anteroventral amygdala (MeAv), basomedial amygdala (BMA), dorsal entopeduncular nucleus (EPd), ventral entopeduncular nucleus (EPv), intercalcated cells (ITCs), globus pallidus externa (GPe), substantia innominota (SI), and lateral hypothalamus (LH); scale bar = 200 µm. g RNAscope FISH in the CeA of a mMORp1-eYFP injected mouse examining co-localized of Oprm1 and eYfp mRNA transcripts. CeA co-localization of AAV5-mMORp1-hM4Di-mCherry and Oprm12A-Cre:Sun1-sfGFP reporter nuclei (h) or anti-Cre staining (i). Co-expression of AAV5-mMORp1-hM4Di-mCherry with anti-Cre immunoreactive cells in the CeA of Oprm1Cre mice (j), and the dorsomedial striatum (DMS) (k), mPFC (l) and ventral tegmental area (VTA, m) of Oprm12A-Cre mice. n Averaged number of mMORp1-mCherry+/anti-Cre+ cells compared to mMORp1-mCherry+/anti-Cre− cells quantified from successfully transduced brain regions of interest (from left or right hemispheres, or both) of Oprm1Cre and Oprm12A-Cre:Sun1 mice within the CeA (~90.2%, N = 5, n = 9), DMS (~89.65%, N = 5, n = 7), mPFC (~88.28%, N = 3, n = 4), and VTA (~85.24%, N = 5, n = 8). Detailed information of total counts and quantification for each region within individual mouse lines can be found in Supplementary Fig. 4 . Scale bar = 100 μm for g–m. White arrow heads denote cells in which co-labeling for Oprm1 and EYFP transcript (g) or mMORp1-mCherry and anti-Cre signal (h–m) is observed. mMORp viral construct shows selective expression on Oprm1/MOR positive neurons across multiple brain regions Before further examining the functional applications of the constructs described above, we first sought to confirm that the expression profile of our mMORp1 AAV was both restrictive and selective within predominantly neuronal cell types with putative Oprm1 promoter activity. To do so, we first intracranially injected C57BL/6J mice with AAV1-mMORp1-eYFP (3 × 1011 gc/mL), targeting the mPFC as a representative, MOR+ cell containing region (N = 3 male mice). The profile of transduced cells within the mPFC showed promising restriction across cortical layers 2, 5 and 6 within the cingulate (Cg1), prelimbic (PL) and infralimbic (IL) regions (Fig. 1d ). These virally transduced mPFC cells were co-labeled for the neuronal marker NeuN, while none appeared to overlap with glial markers, such as the microglial marker Iba1, a cell type that has previously been speculated to harbor active Oprm1 promoters 22 , 23 , 24 (Fig. 1e ). To determine the transduction efficiency and fidelity of mMORp1 expressed in neurons with endogenous Oprm1 promoter activity, we leveraged a rigorous methodological pipeline to assess mRNA, protein, and promoter-driven gene co-expression with mMORp encoded transgenes. First, we intracranially injected AAV1-mMORp1-eYFP into the central nucleus of the amygdala (CeA), the ventral tegmental area (VTA), and the dorsomedial striatum (DMS) of C57BL/6J mice, brain regions previously shown to express MOR, to observe viral spread and expression patterns. In the CeA, we found that most cells showed robust co-expression of mMORp-YFP and anti-MOR immunofluorescence (via a validated MOR antibody, Supplementary Fig. 3 ), with little expression in the adjacent basolateral amygdala (BLA) and other surrounding brain regions, which matches published reports of MOR expression in this subcortical region 25 (N = 3 male mice, Fig. 1f ). Since MOR antibody staining labels diffuse dendrites and axons, with hard-to-resolve somas, we next performed fluorescent in situ hybridization (FISH) on CeA tissue sections to determine if endogenous Oprm1 and transduced EYFP mRNA transcripts more clearly co-localized on CeA, DAPI-labeled nuclei. We observed a near total overlap of both transcript species within mMORp1 transduced cells (Fig. 1g ), suggesting a high degree of specificity for our construct at the level of mRNA. However, given the dynamic and circadian expression of Oprm1 26 , 27 , and the relatively short stability of the mRNA transcripts detectable by FISH at any one instance 28 , we next used two transgenic mouse lines: an Oprm1Cre knock-in/knock-out line developed by Dr. Richard Palmiter 29 and a bicistronic Oprm12A-Cre line developed by Dr. Julie Blendy using CRISPR knock-in 10 that both express Cre recombinase under the endogenous Oprm1 promoter. Mice from the two Oprm1-Cre lines, either heterozygous for Cre (Oprm1Cre, N = 2 male and 3 female mice) or crossed with the nuclear-envelop targeted fluorophore Sun1-sfGFP (Oprm12A-Cre:Sun1, N = 2 male and 2 female mice), were injected with AAV5-mMORp1-hM4Di-mCherry at a standardized volume and titer (400 nL, 1 × 1011 gc/mL) across four regions of interest (CeA, VTA, DMS, and mPFC), and tissue slices from successfully transduced regions were examined for mCherry co-labeling with anti-Cre immunofluorescence or Sun1-sfGFP. Sun1-sfGFP and anti-Cre staining were observed to overlap with mMORp1-mCherry signal within the CeA of Oprm12A-Cre:Sun1 mice (Fig. 1h, i ), while all four transduced regions within Oprm1Cre or Oprm12A-Cre:Sun1 mice showed similarly robust co-labeling of cells for both anti-Cre and mMORp1-mCherry signal (Fig. 1j–m ). Analysis of the transduced regions of interests pooled from both Cre lines revealed the majority mMORp1-mCherry+ labeled neurons to be co-labeled for anti-Cre staining (CeA = 90.2%%, n = 9 ROIs, from N = 5 mice; DMS = 89.7%, n = 7, N = 5; mPFC = 88.3%, n = 4, N = 3; VTA = 85.2%, n = 8, N = 5; Fig. 1n ; quantification for individual mouse lines available in Supplementary Fig. 4a–h , with sample quantification of a single region of interest [ROI] shown in Supplementary Fig. 4i ). Minimal expression of mMORp1-mCherry signal was observed on neurons negative for anti-Cre or anti-Cre and Sun1-eGFP, further indicating that successful transduction of cells via mMORp1 (hence referred to as mMORp) is restricted predominately to putative MOR+ cells-types with active genomic Oprm1 promoters. To further demonstrate that our mMORp constructs afford a level of selectivity and utility for targeting putative MOR+ cell populations that stand apart from other more generic promoter driven constructs, we performed co-injections of an AAV5-hSyn-mCherry reporter virus mixed one to one with a serotype-matched variant of our mMORp-eYFP virus (AAV5-mMORp-eYFP) into two of the representative MOR+ regions assessed in our previous specificity studies (CeA and VTA, Supplementary Fig. 2k, l ). In both the CeA and VTA, we noted that the vast majority of hSyn-mCherry and mMORp-eYFP cells comprised separate populations (CeA: mCherry = 61.6%, eYFP = 24.1%, mCherry/eYFP = 14.3%, n = 4 ROIs from N = 2 mice; VTA: mCherry = 37.3%, eYFP = 40.7%, mCherry/eYFP = 22.0%, n = 2, N = 1; Supp. Fig. 2m, n ), with mCherry + /eYFP+ cells only making up a small percentage of total mMORp transduced cells (CeA = 37.4%; VTA = 35.1%). Transduction patterns for co-injections into the CeA also showed noticeably greater spread of the hSyn-mCherry virus into the MOR- BLA, while mMORp-eYFP transduced cells were found to be more restricted to the general boundaries of the CeA itself, further supporting the transduction selectivity of the mMORp constructs over those which may utilize more common promoter sequences in their design. As a final test of fidelity and specificity, we wanted to determine if the design of our MORp constructs would allow for transduction to occur predominantly in neurons, as opposed to more broadly across neuronal and glial cell types. AAV1-mMORp-eYFP was injected in a cohort of mice into mPFC, CeA and VTA, and a glial marker IHC panel was conducted to examine the overlap of mMORp-eYFP signal with CC1 (oligodendrocytes), PDGFRα (oligodendrocyte precursors), GFAP (astrocytes) and Iba1 (microglia) antibody staining (Supplementary Fig. 5 ). In all cases, minimal to no signal for glial marker staining was noted on cells positive for mMORp-eYFP across all regions of interest (CC1: mPFC = 1.9%, n = 4 ROIs from N = 2 mice; CeA = 1.4%, n = 4, N = 2; VTA = 4.3%, n = 2, N = 1; Supplementary Fig. 5a, b ; GFAP: mPFC = 1.2%, n = 3, N = 2; CeA = 1.6%, n = 3, N = 2; VTA = 0.9%, n = 2, N = 1; Supplementary Fig. 5c, d ; PDGFRα: mPFC = 1.0%, n = 2, N = 1; CeA = 5.1%, n = 3, N = 2; VTA = 3.5%, n = 2, N = 1; Supplementary Fig. 5e, f ; Iba1: mPFC = 1.3%, n = 2, N = 1; CeA = 1.6%, n = 2, N = 1; VTA = 1.2%, n = 3, N = 2; Supplementary Fig. 5g, h ), indicating a higher transduction preference of our viral constructs for neurons. Additional gene expression analyses performed on cultured human neuronal (SHSY5Y), astrocytic (A172), microglial (C20) and murine microglial (N9) cell lines following transduction with AAV1-hMORp-eYFP or AAV1-mMORp-eYFP, respectively, revealed the normalized expression of m/hMORp-eYFP to be lower in C20 cells compared to SHSY5Y cells, and reduced across other cell lines (one-way ANOVA, F = 3.764, P = 0.0409, SHSY5Y v. A172: P = 0.073, SHSY5Y v. C20: P = 0.0251, SHSY5Y v. N9: P = 0.0645; Supplementary Fig. 6f ), while OPRM1/Oprm1 expression across glial lines was significantly reduced compared to SHSY5Y cells (one-way ANOVA, F = 413.5, P < 0.0001, SHSY5Y v. all: P < 0.0001; Supplementary Fig. 6e ). Apart from the underlying predilection of AAVs for infecting neurons over other glial cell types 30 , 31 , 32 , the exclusion of expression specifically in microglia may result from the inclusion of a PU.1 transcription factor binding region, which has been previously demonstrated to repress MOR expression in myeloid-lineage cells, including microglia 33 , 34 . mMORp viral construct displays a robust expression profile in multiple brain regions across small mammalian model organisms Following the above specificity and selectivity studies, we next wanted to examine the possible applications of our viral constructs in targeting populations of MOR+ neurons across several animal model systems utilized in the broader opioid research field. While many groups utilize mice to study aspects of opioid use disorder (OUD), withdrawal, and acute/chronic pain due to a growing wealth of transgenic lines 8 , 9 , 29 , tools 35 , 36 , 37 , 38 , and behavioral paradigms 16 , there remain key advantages in the use of additional mammalian model systems, in particular rats and shrews, for addressing important questions regarding the endogenous opioid system’s function and dysfunction 39 , 40 , 41 . Thus, we evaluated the ability of mMORp to transduce neurons within the brains of both Sprague-Dawley rats and Asian house shrews (Suncus murinus, a unique animal model to study the opioid system in the context of emesis and nausea 42 , 43 , 44 ), compared with additional C57BL/6J mice. We intracranially injected AAV1-mMORp-eYFP (1.4 × 1012 gc/mL) in two conserved, MOR+ representative structures in mice and rats: the CeA and VTA, and the area postrema/nucleus tractus solitarius (AP/NTS) within the dorsal vagal complex of the shrew, a structure known to express mu opioid receptors that contributes to feeding, emesis and the hypercapnic/hypoxic ventilatory response to low-oxygen 39 , 45 , 46 , 47 . Viral expression in mouse CeA and VTA (N = 3 male mice, Fig. 2a, b ) displayed similar patterns of restriction to targeted regions of interest to what we reported above, with additional staining once again showing much of the eYFP signal to be contained within NeuN+ cell bodies and not within any cells labeled for glial markers such as Iba1 once again. Tissue taken from rats targeted within the CeA and VTA (N = 3 male rats, Fig. 2c, d ) showed complementary transduction patterns to those observed in mice, with eYFP signal primarily overlapping with NeuN+ cell bodies via IHC. Additional FISH staining of CeA tissue also showed similar results to those in mouse tissue, with most transduced cell bodies positive for EYFP transcript co-labeled for native rat Oprm1 transcript (71.8%, n = 2, N = 1; Supplementary Fig. 7a–e ). Transduced neurons were once again observed to be restricted primarily to the targeted regions of interest, indicating nominal levels of off-target spread or transduction of MOR- neurons. To confirm that this expression profile was selective for neurons within rat tissue as well, a similar glial marker IHC panel was conducted using CC1, GFAP and Iba1 antibody staining. As in mice, cells within rats transduced by mMORp-eYFP showed little to no overlap with signals for glial marker antibody staining within CeA (CC1: 1.0%, n = 2, N = 1; GFAP: 0.6%, n = 2, N = 1; Iba1: 0.9%, n = 2, N = 1; Supp. Fig. 8a, b ) or VTA (CC1: 0.5%, n = 2, N = 1; GFAP: 1.5%, n = 2, N = 1; Iba1: 2.3%, n = 2, N = 1; Supp. Fig  8c, d ). Within shrew tissue, similar patterns of transduction were also found, with mMORp-eYFP restricted exclusively to non-Iba1+ cells (0.6%, n = 3, N = 2; Supp. Fig. 8e ), and overall spread of the virus noted to be relatively restricted to the borders of the AP/NTS structure where the majority of MOR+ cells reside 48 (N = 2 female shrews, Fig. 2e ). Taken together, these cross-species studies demonstrate that our constructs are highly viable in both rat and shrew models systems and show a similar pattern of restricted expression within brain structures known to harbor MOR+ neuronal populations and respond to MOR manipulation. Fig. 2: mMORp viral transduction within putative MOR+ cells in multiple brain regions across mammalian model organisms. a Mouse (C57BL/6 J) CeA expression of AAV1-mMORp-eYFP (titer: 1 × 1011 gc/mL). Left: overview of the CeA (blue borders) and surrounding regions (white borders). Right: higher magnification images of mMORp-eYFP (anti-GFP amplified) with neuronal (NeuN) and microglial (Iba1) cell type markers. b Mouse VTA expression of AAV1-mMORp-eYFP: Left: VTA (magenta borders). Right: anti-GFP amplified eYFP, NeuN and Iba1. c Rat (Sprague-Dawley) CeA expression of AAV1-mMORp-eYFP (titer: 1 × 1012 gc/mL). Left: overview of the CeA (blue borders) and surrounding structures. Right: anti-GFP amplified eYFP, NeuN and Iba1. d Rat VTA expression of mMORp-eYFP. Left: overview of VTA (magenta borders). Right: anti-GFP amplified eYFP, NeuN and Iba1. e Shrew (Asian house shrew) area postrema/nucleus tractus solitarius (AP/NTS) expression of AAV1-mMORp-eYFP (titer: 1 × 1012 gc/mL). Left: overview of NTS (purple borders) and surrounding structures. Right: anti-GFP amplified eYFP, NeuN and Iba1. Scale bars = 100 μm (far left), 200 µm (right) for a–e. Staining for additional glial markers to demonstrate transduction of predominantly neurons in rat and shrew tissue is shown in Supplementary Fig. 8 (including quantification for Iba1 and mMORp-eYFP staining demonstrated in representative images above). Yellow arrows indicated representative NeuN/eYFP positive cells within merged images. mMORp-hM4Di produces robust anti-nociception in spinal nociceptive circuits To demonstrate the application of our constructs for encoding transgene products useful for interrogating opioidergic circuits in pain/nociceptive pathways, we conducted a series of functional and behavioral studies using the mMORp-h4MDi-mCherry and mMORp-GCaMP6f constructs. Intrathecal morphine and other opioids produce strong anti-nociception when binding to spinal MORs that are expressed on the presynaptic terminals of primary afferent nociceptors and on descending inputs from the brain stem, as well as MORs expressed on intrinsic dorsal horn neurons 1 . These compounds engage inhibitory Gi signaling cascades that silence these different neural populations to disrupt the transmission of peripheral nociceptive information to supraspinal brain regions. However, behavioral pharmacology does not allow for testing the role of different MOR+ neural populations within this critical nexus of the nociceptive pathways regarding analgesia. Since the DREADD GPCR, hM4Di, couples to similar Gi cascades as MOR, we hypothesized that mMORp-hM4Di could mimic intrathecal morphine antinociception 49 , but in a selective manner that tests the necessity of spinal cord MOR+ neurons to induce analgesia, independent of afferent and brainstem terminal opioid-mediated inhibition 50 . Thus, for chemogenetic manipulation of spinal MOR+ neurons in mice experiencing nociceptive stimuli, we directly injected C57BL/6J mice with AAV1-mMORp-h4MDi-mCherry at the L4-L5 lumbar sections of the spinal cord (intraparenchymal, 400 nL, ~3 × 1012 gc/mL, per injection site; Fig. 3a ). Robust mMORp-hM4Di-mCherry expression was detected throughout the dorsal and ventral horns, primarily in lamina II inner, consistent with expression patterns observed in MOR-mCherry mice 25 , without detection in dorsal root ganglia axonal inputs (N = 3 male mice; Fig. 3b, c ). We then conducted a series of well-validated, common nociceptive behavioral assays on these mice with and without different DREADD agonists to measure anti-nociceptive effects (Fig. 3d ). We found that mMORp-hM4Di injected animals displayed an increase in mechanical threshold sensitivity in the von Frey filament Up-Down test following either intrathecal deschloroclozapine (DCZ, 10 pg; two-way ANOVA, F = 8.911, P = 0.0105; Supplementary Fig. 9a ) or systemic clozapine N-oxide (CNO, i.p., 3 mg/kg; two-way ANOVA, F = 8.521, P = 0.0112) compared to the within-subjects baseline thresholds and mCherry control mice (Fig. 3e ). Similarly, the mMORp-hM4Di group showed an increase in reflexive latency to withdrawal, and decreases in the duration of paw licking behavior and jumping or escape-related events on an inescapable hot plate (52.5 °C) compared to control animals following either DCZ (hot plate latency: two-way ANOVA, F = 19.81, P = 0.0007, duration: F = 2.058, P = 0.1751, jump/escape: F = 4.921, P = 0.0450; Supplementary Fig. 9b ), or CNO treatment (duration: two-way ANOVA, F = 9.514, P = 0.0081; jump/escape: 2-way ANOVA, F = 13.65, P = 0.0024; Fig. 3f ). Lastly, mMORp-hM4Di and mCherry groups received systemic CNO 30 min prior to an intraplantar injection of a 4% formalin solution to induce TRPA1-nociceptor hypersensitivity and subsequent central sensitization 51 , 52 . In the mMORp-hM4Di mice we observed significant decreases in nocifensive behaviors (paw licking, biting, escape behaviors, etc.) during both the first stage of behavioral observations (direct activation of afferent nociceptors) and the later second stage (reflective of the inflammatory and central sensitization component) of the test (Fig. 3g ; time: two-way ANOVA, F = 22.27, P < 0.0001; phase: two-way ANOVA, F = 401.5, P < 0.0001). In total, these tests demonstrate the functional validity of our chemogenetics-based viral constructs in behaviorally relevant tasks, as exemplified by the ability of mMORp-hM4Di to produce similar antinociceptive effects to those of spinal opioid agonist administration 22 . Fig. 3: mMORp-hM4Di spinal cord expression and chemogenetic-induced analgesia. a AAV1-mMORp-hM4Di-mCherry injection schema within the lumbar spinal cord in C57BL/6J mice to inhibit dorsal horn MOR+ cells and not MOR+ nociceptors or descending brain stem circuits. b mMORp-hM4Di-mCherry expression and spread across the L4 spinal cord; scale bar = 100 μm. c Location map and quantification of mMORp-hM4Di-mCherry+ cells across the Rexed laminae in the dorsal and ventral horns (N = 3 mice). d Experimental timeline for viral injections and chemogenetic behavioral testing. e Mechanical sensory thresholds (von Frey Up-Down testing) in mMORp-hM4Di-mCherry injected mice (N = 9 mice) compared with hSyn-mCherry injected controls (N = 7 mice) at baseline and 30 min following systemic CNO administration (3 mg/kg; Two-way ANOVA + Bonferroni: main effect: P = 0.011 [viral treatment × CNO treatment]; multiple comparisons: basal v. CNO, P = 0.990 [mCherry], P = 0.006 [hM4Di]). Average response changes per group shown as thick gray (mCherry) or red (hM4Di) lines. Individual mice are shown as thin gray and red lines. f Nocifensive behaviors observed on an inescapable 52.5 °C hot plate over a 30-sec trial for the same animals (N = 9 hM4Di-mCherry, N = 7 mCherry): latency (sec) to hind paw withdrawal (two-way ANOVA + Bonferroni: main effect: P = 0.085 [viral treatment], P = 0.162 [CNO treatment]; multiple comparisons: basal v. CNO, P > 0.999 [mCherry], P = 0.067 [hM4Di]), hind paw licking duration (two way ANOVA + Bonferroni; main effects: P = 0.008 [viral treatment × CNO treatment], P = 0.008 [viral treatment]; multiple comparisons: basal v. CNO, P > 0.999 [mCherry], P = 0.0007 [hM4Di]), and total jumping bouts (two way ANOVA + Bonferroni; main effects: P = 0.004 [viral treatment × CNO treatment], P = 0.002 [viral treatment], P = 0.006 [CNO treatment]; multiple comparisons: basal v. CNO, P = 0.0006 [mCherry], P > 0.999 [hM4Di]). g Time course (left) and cumulative global scoring (right) of nocifensive behaviors (licking, biting, jumping, etc.) observed in the formalin injection test during the first and second phases of testing in mMORp-hM4Di-mCherry (N = 9) and control (N = 7) animals at basal and post-CNO time points (Time course: two-way ANOVA + Bonferroni; main effects: P = 0.006 [time bin × nocifensive behaivors], P < 0.0001 [time bin], P = 0.001 [nocifensive behaviors]; multiple comparisons: mCherry v. hM4Di, P = 0.014 [5 min], P = 0.008 [20 min], P = 0.0002 [25 min], P = 0.48 [30 min]. Global scoring: two-way ANOVA + Bonferroni; main effects: P = 0.006 [stage × nocifensive behaviors], P < 0.0001 [stage], P = 0.001 [nocifensive behaviors]; multiple comparisons: mCherry v. hM4Di, P = 0.222 [1st stage], P < 0.0001 [2nd stage]). All data are presented as means ± SEM *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. mMORp-GCaMP6f permits readout of neuronal population activities in response to morphine, opioid withdrawal, and noxious stimuli We next assessed the ability of the mMORp-GCaMP6f construct for use in fiber photometry calcium imaging in awake, behaving animals to assess population activity changes of transduced MOR+ CeA neurons, a sub-region of the amygdala shown previously to express MORs and to be implicated in aspects of nociceptive processing 53 , 54 . We injected C57BL/6J mice with AAV1-mMORp-GCaMP6f (400 nL, ~2 × 1012 gc/mL; N = 12 male mice) or an eGFP encoding control virus (N = 3 male mice) in the right CeA (Fig. 4a ) prior to implantation of a fiber optic head post just dorsal to the injection site. As with the other mMORp constructs, mMORp-GCaMP6f expressed strongly, with relatively restricted expression in the CeA and minimal spread into the neighboring BLA or striatum (Fig. 4b ). Following a three-week incubation period, mice were run through a battery of behavioral experiments to gauge relative calcium activity changes in response to noxious stimuli (Fig. 4c ). Initially, mice were exposed to brief air puffs delivered to the abdomen (compressed air canister, 1 s puff) and hot water drops applied to the left hind paw (~55 °C, ~25 µl drop), and the recorded calcium mediated events were time-locked to the stimulus application. In response to air puffs, calcium activity in the CeA of mMORp-GCaMP6f injected mice increased significantly relative to eGFP control animals when examining both the area under the curve (AUC) and peak z-score of recorded events (AUC: GCaMP = 13.2, eGFP = 4.7, two tailed unpaired t-test, P = 0.0351; peak z-score: GCaMP = 15.9, eGFP = 3.7, two tailed unpaired t-test, P = 0.0112; Fig. 4d ), thereby indicating proper functionality of our GCaMP6f construct in CeA neurons, a region known for its elevated activity profile in response to unexpected external stimuli 55 , 56 . Similarly, hot water applications reliably produced robust, time-locked calcium mediated events when compared with controls (AUC: GCaMP = 13.0, eGFP = 5.7, two tailed unpaired t-test, P = 0.051; peak Z-score: GCaMP = 12.3, eGFP = 4.5, two tailed unpaired t-test, P = 0.009; Fig. 4e ). To assess if mMORp-GCaMP6f-trandusced CeA neurons are opioid-sensitive, we next administered morphine (i.p., 10 mg/kg) 30 min prior to another round of hot water hind paw stimulations. Morphine treated mMORp-GCaMP6f animals displayed a marked reduction in calcium mediated event AUC and peak amplitude (AUC: no morphine = 13.0, morphine = 7.7, two tailed unpaired t-test, P = 0.016; peak z-score: no morphine = 12.3, morphine = 7.6, two tailed unpaired t-test, P = 0.005; Fig. 4f ), indicative of a potential effect of morphine on the activity of transduced, putative MOR+ CeA neurons. Fig. 4: mMORp-GCaMP6f in vivo photometry recording of MOR+ neural population calcium activity in response to noxious stimulation, opioid administration, and opioid withdrawal. a In vivo fiber photometry recording set up and overall experimental design, as well as AAV1/5-mMORp-GCaMP6f injection schema in C57BL/6J mice. b mMORp-GCaMP6f viral targeting and transduction efficiency within CeA neurons (NeuN+) and fiber optic implant placement; scale bars = 500 μm (left), 100 m (right). c Photometry experimental timeline for mMORp-GCaMP6f (N = 12) and mMORp-eYFP (N = 3) injected mice. d Normalized calcium-mediated activity responses observed in the CeA of mMORp-GCaMP6f and mMORp-eYFP mice in response to 10 air puff applications (1-min inter-stimulation interval. Thick lines = averaged responses, thin lines = individual mice average responses to all 10 stimulations). Inset graphs depict differences in the area under the curve (AUC) quantified between averaged GCaMP6f and eYFP control response (two tailed unpaired t-test, P = 0.035) and peak z-score between groups (two tailed unpaired t-test, P = 0.011). e Plots for normalized, calcium-mediated responses to the application of 10 ~ 55 °C hot water droplets to the left hind paw (1-min inter-stimulation interval). Thick lines = average group response, thin lines = individual mice averaged response. Insets show AUC (two tailed unpaired t-test, P = 0.051) and peak z-score (two tailed unpaired t-test, P = 0.009) comparisons for eYFP (N = 3) and GCaMP6f (N = 12) expressing mice. f Normalized, calcium-mediated responses to an additional round of hot water hind paw stimulations in the mMORp-GCaMP6f mice (N = 12) before and 30 min after morphine (i.p., 10 mg/kg). Opioid-naïve responses (green) and post-morphine responses (blue) are shown as grouped averages (thick lines) and individual averages (thin lines) following 10 stimulations. Insets show AUC (two tailed unpaired t-test, P = 0.016) and peak z-scores (two tailed unpaired t-test, P = 0.005) comparisons between the pre- and post-morphine responses in mMORp-GCaMP6f mice. Gray lines track changes in AUC and peak z-score for individual mice in pre- and post-morphine treatment conditions. g Normalized, calcium-mediated responses from mMORp-GCaMP6f mice undergoing an escalating forced morphine drinking paradigm (Fig. 4c) for morphine drinking (red, N = 6 mice) or saccharin drinking (gray, N = 5 mice) subjects following naloxone induced precipitated withdrawal (i.p., 3 mg/kg). Insets show the global withdrawal behavior score (left) and total calcium-mediated events (right) post-naloxone between morphine and saccharin mice (W/D score: two tailed unpaired t-test, P = 0.0001; Ca2+ events: two tailed unpaired t-test, P = 0.0007). All data are presented as means ± SEM *P < 0.05, **P < 0.01, ***P < 0.001. We next sought to extend the utility of our mMORp-GCaMP6f construct to examining applications aimed at modeling chronic opioid use and withdrawal. Using the same cohort of mMORp-GCaMP6f mice, we placed animals on a forced morphine drinking paradigm using home cage water bottles containing either an increasing concentration of morphine, or the artificial sweetener saccharin (N = 5 male mice, saccharin; N = 6 male mice, morphine). Drug treated mMORp-GCaMP6f groups were supplied with water containing 0.3 mg/ml morphine, 0.2% saccharin for 3 days, and 0.5 mg/ml morphine, 0.2% saccharin for 4 days, while control mMORp-GCaMP6f groups were only provided 0.2% saccharin laced water (Fig. 4c ). After the final day of drinking, mice underwent a precipitated withdrawal challenge (naloxone, i.p., 3 mg/kg) prior to the start of behavioral testing and fiber photometry recording. Morphine drinking, naloxone treated mMORp-GCaMP6f animals displayed significantly greater characteristic opioid withdrawal behavioral phenotypes (wet dog shakes, biting, jumping/escape behavior) immediately following naloxone injection, with an associated, significant increase in the total number of calcium-mediated events scored within a 20-minute observation window. By comparison, saccharin drinking mMORp-GCaMP6f animals showed little to no changes in overt behavior and calcium-mediated activity profiles for the duration of testing (average post naloxone global withdrawal score: saccharin = 18.4, morphine = 88.5, two tailed unpaired t-test, P = 0.001; average post naloxone calcium events: saccharin = 41, morphine = 91, two tailed unpaired t-test, P = 0.0007; Fig. 4g ). Taken together, these use case experiments demonstrate that our mMORp-GCaMP6f construct can be paired with widely adopted techniques for recording in vivo neural activities in pain and OUD-related studies. mMORp driven recombinases and PHP.eB/PHP.s capsid packaging allows for intersectional genetic access to PNS and CNS opioidergic cell populations and circuits Increasingly, many contemporary neuroscience investigations aim to both manipulate specific neuronal subpopulations and circuits based on multiple dimensions, including molecular expression, connectivity, function, and location within the nervous system, with viral tools becoming an increasingly invaluable means to achieve these levels of specificity 15 . Determining which cell types are accessed can be partly controlled by the capsid proteins of a specific virus (see Supplementary Fig. 16 ), such as the recently engineered PHP.eB and PNS PHP.S capsids variants, which are capable of selectively transducing cells within the CNS or the PNS over cells present in non-neural tissue types, respectively 12 , 57 , as well as the inclusion of specific promoter and/or enhancer element into constructs packaged within AAVs. Furthermore, hundreds of existing Cre and Flp recombinase transgenic mouse lines are also in use in labs around the world which can be used to achieve cell and circuit specific genetic access 58 , 59 . To capitalize on these advances in targeting strategies and tool development, we created four example mMORp constructs useful for intersectional neuronal labeling and tracing studies: a mMORp-mCherry-IRES-Cre construct encoding Cre recombinase, a mMORp-FlpO construct encoding Flipase, and two mMORp-eYFP constructs packaged in AAV-PHP.eB and AAV-PHP.s capsids. Single-cell RNA sequencing studies show that Oprm1 is expressed in both glutamatergic pyramidal cell types as well as GABAergic interneurons 60 , 61 , 62 , 63 . Selective access to these different classes of cortical neurons however has not been possible, leaving gaps in our basic understanding of cortical opioidergic processes. Thus, to test our mMORp-driven recombinases for their effectiveness in targeting GABAergic subtypes of MOR+ cortical neurons, we first co-injected a mixture of AAV8-mMORp-mCherry-IRES-Cre (titer: 7 × 1012 gc/mL) combined with AAV9-hDlx-FLEx-eGFP (titer: 2.2 × 1013 vg/mL) viruses into the mPFC and somatosensory cortex (S1) of C57BL/6 J mice (N = 3 male mice). The latter construct uses the human Dlx enhancer element to promote viral transduction in forebrain GABAergic cells and contains a FLEx switch making eGFP transgene expression dependent on the presence of Cre recombinase within a transduced cell 13 . We observed the same expression pattern of mCherry in mPFC across the cortical layers as shown in Fig. 1d for our first round of mMORp-eYFP injections, which was mirrored in S1 brain sections, while eGFP expression was restricted to a subset of small cells with non-pyramidal morphology in both mPFC and S1 (Fig. 5a, b ). Next, we co-injected AAV1-mMORp-FlpO (titer: 1.4 × 1012 gc/mL) virus with a pan-neuronal, human Ef1α promoter-driven Flp-dependent mCherry reporter (AAV9-Ef1a-fDIO-mCherry, titer: 2.4 × 1013 vg/mL) into a separate cohort of C57BL/6J mice (N = 3 male mice) to demonstrate an additional recombinase-based targeting strategy to label putative MOR+ neurons. We found that mMORp-FlpO successfully drove Flp recombinase dependent recombination within Ef1a-fDIO-mCherry transduced cells and observed numerous mCherry labeled cells in both mPFC and S1 (Fig. 5c, d ). Taken together, these two strategies show that our mMORp constructs can also be combined with the ever-expanding catalogs of recombinase-based viral tools to access a multitude of cell types in the brain. Fig. 5: mMORp driven recombinases and PHP.eB/PHP.S capsid packaging support intersectional viral strategies for gaining genetic access to CNS and PNS opioidergic cells/circuits. C57BL/6J mice injected with a 9 µl:1 µl mix of AAV8-mMORp-mCherry-IRES-Cre and a Cre-dependent AAV9-hDlx-FLEx-eGFP with intersectional expression in putative MOR+/GABAergic neurons in mPFC (a) and somatosensory cortex (S1, b). Insets in a and b show higher magnification images of hDlx-FLEx-eGFP cells (green) overlap with mMORp-mCherry-IRES-Cre cells (magenta); scale bars = 200 μm and 100 μm for insets. C57BL/6J mice injected in mPFC (c) or S1 (d) with a 9 µl:1 µl mix of AAV1-mMORp-FlpO and Flp-dependent AAV9-Ef1α-fDIO-mCherry. Inset high magnification images show mCherry+ cells. e CNS expression in representative sections from the spinal cord (coronal) and brain (sagittal along the medial-lateral axis relative to Bregma; coronal sections are shown in Supplementary Fig. 10 ) of C57BL/6J mice injected retro-orbitally with AAV.PHP.eB-mMORp-eYFP virus; scale bars = 500 μm (spinal cord sections) and 1000 µm (sagittal sections). f PNS dorsal root ganglia (DRG) FISH from C57BL/6J mice with intracerebroventrical injection of either AAV.PHP.S-mMORp-eYFP (upper) or AAV.PHP.S-CAG-tdTomato (lower). Custom cDNA probes targeting Rbfox3 (NeuN), Oprm1, EYFP and tdTomato transcripts; scale bars = 50 μm. g Quantification of total Oprm1+ cells co-labeled for tdTomato in control animals (N = 5 mice, n = 14 ROIs) compared to total Oprm1+ cells co-labeled for EYFP in PHP.S-mMORp-eYFP injected mice (N = 4 mice, n = 11 ROIs) across treatment groups (two tailed unpaired t-test with Welsh’s correction, P = 0.0437). h Summary quantification of the percent total number of cells positive for EYFP transcript (i.e., transduced by the AAV.PHP.s-mMORp-eYFP virus) that were also either positive for Oprm1 transcript (mMORp-eYFP/Oprm1+, 82.5%) or negative for Oprm1 transcript (mMORp-eYFP+ only, 17.5%) to demonstrate AAV.PHP.s-mMORp-eYFP specificity within mouse DRG neurons. Data presented as means ± SEM, *P < 0.05. We next evaluated the use of the PHP.eB and PHP.s capids to deliver mMORp-eYFP into broad, CNS and PNS distributed MOR+ cells. In adult C57BL/6J mice, we delivered AAV-PHP.eB-mMORp-eYFP (titer: 8.6 × 1012 gc/mL) via retro-orbital injection (50 µl) in order to facilitate better systemic viral spread and distribution, and observed robust expression of eYFP in cells throughout the spinal cord and brain in both sagittal (Fig. 5e ) and coronal tissue preparations (Supplementary Fig. 10 ), with high magnification insets shown next to representative coronal sections to provide a clearer view of the mMORp-eYFP+ cells observed in several of the structures of interest targeted in our intracranial focal injection studies. In C57BL/6J P2 pups, we performed an intracerebroventricular injection of AAV-PHP.S-mMORp-eYFP (titer: 1.4 × 1013 gc/mL, Fig. 5f , upper) and after 6 weeks incubation, performed FISH on dorsal root ganglia sections for EYFP and Oprm1 mRNA transcripts. When compared to pups that had been similarly injected with a generic AAV-PHP.S-CAG-tdTomato encoding virus (titer: 2.1 × 1013 vg/mL, Fig. 5f , lower), cells counted in DRG sections from the mMORp injected animals (n = 11 ROIs) showed robust co-localization of EYFP and Oprm1 transcripts on/around the same DAPI delineated nuclei, while co-localization of tdTomato and Oprm1 transcripts (n = 14 ROIs) was significantly more sparse (Oprm1+tdTomoto/tdTomato = 0.60, Oprm1 + EYFP/EYFP = 0.83, two tailed unpaired t-test with Welsh’s correction, P = 0.0437, Fig. 5g ). Similarly, we noted that the majority of Oprm1 positive cells in DRG sections collected from mice injected with the AAV-PHP.S-mMORp-eYFP virus were co-labeled for EYFP transcript compared to those cells labeled for EYFP alone (EYFP/Oprm1+ = ~82.5%, EYFP+ only = ~17.5%, Fig. 5h ). Taken together, these representative findings highlight the use of our mMORp constructs in combination with viral capsid engineering to selectively target MOR+ primary afferents and large numbers of spinal and brain neurons. Human and mouse MORps provide unique access to opioidergic cells in non-human primates in vivo, and human-derived neuronal cell cultures in vitro Having characterized our MORp constructs in rodents and other small mammalian model systems, we next sought to determine the viability of our human promoter derived MORp construct (hMORp). We first performed a series of intracranial injections using the AAV1-hMORp-eYFP virus (titer: 1.17 × 1012 gc/mL) in a single male rhesus macaque, targeting a region anatomically complementary to one we had examined for transduction efficiency of the mMORp constructs in our rodent studies, the dorsal anterior cingulate cortex (dACC), as well as the insular cortex and medial thalamus, with additional injections of the AAV1-mMORp-eYFP construct (titer: 1.4 × 1013 gc/mL) performed in the amygdala 64 , 65 , 66 (Fig. 6a, b , Supplementary Figs. 11 and 12 ). Tissue sections taken from dACC revealed robust transduction of NeuN+ neurons throughout multiple layers of the cortex, with the greatest densities of eYFP+ neuronal cell bodies localized to cortical layers II, V and VI, and the processes of these neurons noted to extend into/throughout layers I, III, V and VI (Fig. 6c , Supplementary Fig. 12 ). Immunostaining in the dACC for microglia with anti-Iba1 showed no overlap with eYFP+ cells, suggesting that like the mMORp constructs, our hMORp construct preferentially targets putative MOR+ neurons and not glial cells (Fig. 6d ). To confirm hMORp selectivity in MOR+ neuron, we performed both a similar glial marker IHC panel and FISH on dACC sections and quantified the co-localization of transgene eYFP signal with that of glial marker staining, or EYFP mRNA transcripts with macaque OPRM1 mRNA transcripts across different regions of interest within the dACC. Glial marker staining with antibodies against CC1, GFAP, PDGFRα and Iba1 in macaque dACC transduced tissue (Supplementary Fig. 13a–c ) revealed similar results to those observed in mouse, rat and shrew tissue assessed for MORp-eYFP signal, in that few to no cells were noted to co-label for eYFP and any glial marker signal (CC1: 1.6%, n = 2 ROIs, N = 1 macaque; GFAP; 1.2%, n = 2, N = 1; PDGFRα: 1.5%, n = 2, N = 1; Iba1: 0.5%, n = 4, N = 1; Supplementary Fig. 13d–g ). For FISH staining, we noted that the majority of OPRM1 labeled cells throughout the transduced region of the dACC were co-labeled for EYFP transcript compared to those cells labeled for OPRM1 or EYFP alone (OPRM1+/EYFP+ = ~81.4%, EYFP+/OPRM1− = ~18.6%; Fig. 6e, f ; expanded quantification and individual ROI counts presented in Supplementary Fig. 14 ). Successful transduction of neurons expressing the OPRM1 promoter was also observed in the amygdaloid structure via the mMORp construct, as well as the insula and mediodorsal thalamus via the hMORp construct (Fig. 6g ), with tissue samples containing the insular cortex and/or the amygdala also showing robust expression of the eYFP tag within neuronal cell bodies and processes in both regions (Supplementary Fig. 12 ), suggesting a broad application for hMORp to provide genetic access to MOR+ cells in multiple regions of the brain in non-human primate, genetically intractable subjects. Fig. 6: hMORp viral constructs drive robust transduction of putative opioidergic cells in non-human primate neural tissue. a 3D reconstruction of the skin, skull and underlying vasculature of a rhesus macaque for pre-operative intracranial injection planning. b Post-injection manganese-enhanced MRI for AAV1-hMORp-eYFP in vivo targeting accuracy assessement (virus mixed 1:100 with 100 mM manganese solution). c Left: Dorsal anterior cingulate cortex (dACC) expression of hMORp-eYFP; scale bar = 1000 μm. Right: dACC, higher magnification with cortical layer markers; scale bar = 200 μm. d Co-expression of hMORp-eYFP with neuronal (NeuN) but not microglial (Iba1) markers; scale bar = 100 μm. Staining for additional glial markers and relevant quantification to demonstrate transduction of predominantly neurons in macaque tissue is shown in Supplementary Fig. 13 . e RNAscope FISH in dACC tissue for co-expression of YFP, OPRM1, and SLC17A7 (VGLUT1) mRNA transcripts; scale bar = 100 μm, far right image = digital zoom of merged image. f Summary quantification of the total EYFP transcript positive cells quantified in sample regions of interest within dACC (upper, N = 1 dACC slice, n = 4 ROIs) with either EYFP+/OPRM1− (18.6%) or EYFP+/OPRM1+(81.4%) to demonstrate hMORp-eYFP virus expression specificity. g mMORp-eYFP expression within the basal nucleus of the amygdala, and hMORp-eYFP expression within the insular cortex and the mediodorsal thalamus following intracranial viral injections into these putative MOR expressing regions. Scale bar = 1000 μm. Lastly, we wanted to determine whether our MORp viruses would be able to provide genetic access to these same putative MOR+ cells within human derived model systems. To test this, we cultured and differentiated human iPSCs (LiPSC-GR1.1 line) to produce either nociceptor-like neuronal cells or non-neuronal cardiomyocytes. Differentiated nociceptor-like cells have previously been shown to possess gene expression profiles consistent with those of nociceptors and sensory neurons observed in vivo. These neurons also show distinct expression profiles of select genes during differentiation in culture 67 . For instance, OPRM1 specifically reached peak expression at day 21 and then decreased by day 28 (Supplementary Fig. 15a ) 67 . When treating these cells with direct administrations of either our AAV-PHP.S-mMORp-eYFP or a control CAG-tdTomato virus at four different titers (1 × 109 – 1012 gc/mL) around days 21–28 in culture, we observed broad transduction of cells with both viruses in nociceptors that increased with titer concentration, with robust expression of both reporters noted at the 1 × 1012 gc/mL titer most prominently (Fig. 7a–d ). By contrast, while expression of tdTomato signal remained prominent at both low and high viral titer concentrations in cultured cardiomyocytes, no eYFP signal was noted within these cultures following AAV-PHP.S-mMORp-eYFP treatment (Fig. 7e–h ). These results indicate that our current iteration of MORp constructs appear to show selectivity for human cell types known to express MOR (and/or the OPRM1 gene) when compared to cells shown to possess low expression of this receptor/gene, as demonstrated both from previous studies 68 , 69 and our own gene expression analyses of these two cultures, which found nociceptor cells to indeed express higher trending levels of OPRM1 than the cardiomyocyte cells in which our transduction experiments were performed (Supplementary Fig. 15b ). Fig. 7: mMORp viral transduction in human iPSC-derived MOR+ nociceptors. Representative low (left, scale bar = 50 μm) and high (right, scale bar = 100 μm) magnification images of cultured human nociceptors treated with higher titer (1 × 1012 gc/mL, MOI: 2 × 108 [nociceptors], 1 × 108 [cardiomyocytes]) AAV.PHP.S-mMORp-eYFP virus (a), or an AAV.PHP.S-CAG-tdTomato virus (b). High magnification sample regions are denoted in lower magnification images via a white box. c, d Similar low (left) and high (right) magnification images of cultured nociceptors treated with lower titer mMORp-eYFP (1 × 109 gc/mL, MOI: 2 ×105 [nociceptors], 1 ×105 [cardiomyocytes], c) or CAG-tdTomato virus (d). Images of cultured human cardiomyocytes treated with high titer mMORp-eYFP (e) or CAG-tdTomato (f) viruses, with regions boxed in white denoting high magnification sample areas shown on the right. Images of cardiomyocytes treated with low titer mMORp-eYFP (g) or CAG-tdTomato (h) viruses, at both low (left) and high (right) magnification. Evidence of OPRM1 gene expression within cultured nociceptor cells is demonstrated in Supplementary Fig. 15 . Discussion The use of up and downstream genetic elements to target virally deliverable transgenes to select neural structures, subtypes of neuronal populations and circuits within them represents a burgeoning area of research within the gene delivery field, with several recent studies utilizing the integration of specific enhancer and promoter elements into their construct designs to target cells expressing unique genetic or molecular markers of interest 14 , 15 . Our results demonstrate that these principles can be extended to targeting mu opioidergic neuronal populations throughout the CNS and PNS with a high level of specificity and selectivity, and that the study of these transduced cells via effector transgenes yields behavioral and physiological readouts consistent with MOR manipulation. These findings present broad implications for the use of these viral tools in the study of the neurobiology of opioid-related fields and open the door to the translational use of such tools for potential therapeutic development and screening platforms. The versatility of both the mMORp and hMORp constructs to transduce putative MOR+ neurons in multiple conserved brain structures across multiple species represents an important step towards expanding the use of different model systems in opioid circuit neuroscience investigations. Indeed, as alignment analyses of the sequences used to generate both of our constructs show (Supplementary Fig. 1e ) both the mMORp and hMORp sequences demonstrate a high level of homology to the native OPRM1/Oprm1 promoter sequence found in rat, macaque and human (no reference genome is currently available for the shrew), suggesting the possibility to significantly enhance and expand the investigation of the opioidergic system both within and outside of the murine research community. While five Oprm1-Cre transgenic mouse lines and a single Oprm1-Cre rat line 70 have been created, at present only one mouse line is commercially available (Jackson Labs, Strain:035574). These mice are haplo-insufficient or total knockouts for Oprm1 when hetero- or homozygous for the Cre allele, respectively, which needs to be accounted for when designing experiments to investigate MOR function itself 29 . The use of AAVs and our MORp constructs can provide a more cost-effective and rapid method for labs to dissect the mu-opioid receptor system in vivo than expensive and slow rodent breeding schemes. Importantly, there are no other transgenic species models available that provide genetic access to MOR cell types. The fact that non-human primate and other small mammalian systems lack the tools or transgenic animals to achieve the same level of granularity in the study of opioid neurobiology has placed an overreliance on mouse systems that may not be the most appropriate model of studies for areas such as pain perception, opioid addiction-like behaviors, or cognitive processes involving the endogenous opioid system. The success of our constructs to transduce MOR+ cells in vivo in mouse, rat, shrew and macaque model systems demonstrate them to be potentially useful for researchers working across species that wish to target these populations for anatomical and functional studies. Regarding the functional identity of the cells transduced with these viruses, and the utility of the effector transgenes encoded by them, our validation studies in mice suggest these cells to respond to both behavioral and pharmacological manipulation in a manner consistent with MOR agonism and/or antagonism. Within the spinal cord, the activation of mu receptors on cells in the DRGs has previously been demonstrated to produce a robust decrease in overall cellular activity and excitability 71 , 72 , 73 , consistent with MOR’s function as an inhibitory G protein coupled receptor 1 , as well as a reduction in nocifensive behaviors following intrathecal administration of MOR agonists 74 , 75 . Our chemogenetic studies using an hM4Di encoding mMORp construct produced similar effects when administering either systemic CNO or intrathecal DCZ, consistent with a more selective modulation of MOR and putative MOR+ cells within the region. Noted increases in the number of calcium mediated events in the CeA and the overall total withdrawal score for behavioral responses in mice injected with our mMORp-GCaMP construct were also consistent with previous studies demonstrating pain-responsive CeA neurons to increase their excitatory activity in response to noxious stimuli 76 , 77 , 78 , and the CeA in general to show an overall increase in molecular markers of neuronal activity in paradigms examining the effects of naloxone precipitated opioid withdrawal 78 , 79 , 80 , 81 . The potential shortcomings of a technique such as fiber photometry, which examines the bulk fluorescent signal from a population of cells transduced with a given calcium indicator, should be noted though. It is possible that the less than complete inhibition of signal from MORp-GCaMP noted in these studies may be attributed to the competing activity patterns of different MOR+ subtypes known to be present within a region such as the CeA 82 , 83 , 84 in response to select types of stimuli or overall differences in MOR receptor density that may be present on transduced cells (which may alter the responsive of certain cells to drugs like morphine, thus influencing the fluorescent signal noted for the broader population of transduced cells). As such, care should be taken when designing experiments using viral tools such as ours to attempt to both target and restrict expression of certain transgenes to the desired population or subpopulation of cells most germane to a specific research question, an approach that our suite of constructs may help to further facility when using them in combinations with other viruses and transgenic animals lines, as demonstrated through additional studies we’ve presented here. Despite this though, current IHC and ISH data clearly show that it is likely our viral constructs are indeed transducing the desired target neuronal populations within these nuclei. Indeed, the restricted expression of our viral constructs to predominantly neuronal cells is in agreement with not only the overall design of these constructs, which included transcriptional elements known to repress expression in myeloid-like cells such as microglia 33 , but also the natural predilections of AAVs to show greater transduction efficacy at neurons when compared to glia 85 , a detail supported by the findings presented in our glial marker staining panels across species. Additional gene expression assays used to address the concern of possible transduction events specifically in microglia seem to further support the higher preference for neuronal transduction inherent to our constructs, despite minimal upticks in mMORp-eYFP or hMORp-eYFP observed across cultured cell lines, which may be explained by cell line heterogenicity and/or minimal viral leak that may be present at higher titers. Despite the identity of the cells primarily transduced by our constructs across the select brain regions of interest discussed above appearing to be neurons, further testing will be necessary to ensure this to be the case in other regions of interest possessing MOR+ cells responsive to, and participatory in, the modulation of pain salient stimuli. Similarly, future testing will be equally necessary to determine which viral serotypes and titer concentrations will provide the greatest levels of transduction efficacy across brain regions, species, and sex variables outside of the sampling presented here. Outside of experiments in which the serotype of co-injected viruses was matched to reduce confounds from potential differences in transduction efficacy, the choice of serotype of a virus used for each of our studies was not informed by prior knowledge of viral tropism (ability of a virus to infect specific cells or tissue types 86 ) inherent to any specific serotypes as reported across different brain regions or neuronal cell types within the literature 87 . This question of tropism and determination of the correct serotype to employ to best transduce a specific tissue or cell type is not unique to our viral constructs, however, as consensuses on the specific capsid proteins and construct elements that provide the greatest transduction efficacy within different tissue types and cell populations is still a matter of debate and investigation, requiring more meticulous documentation and data sharing amongst researchers across disciplines 88 . Examining the extent of the combinatorial approaches that can be taken with the use of our viruses with other transgenic animal lines and tools for the intersectional targeting of unique CNS and PNS cell populations 59 or the dissection of mu opioidergic circuits within the brain will thus require careful testing and planning when using our MORp constructs across any and all applications. Regardless, we believe that the results of these initial use case studies highlight the power and versatility of our MORp tools in the study of the opioidergic system not only in regards to the modulation of physiological pain responses, but also potentially in the study of chronic opioid addiction, withdrawal and other emerging complex behavioral features of pain and OUD as well. While most of the work presented here delves into the applications of the mMORp constructs and their basic research utility in small animal models, the success of our hMORp and mMORp constructs to transduce MOR/OPRM1+ cells in rhesus macaque in vivo, as well as human iPSC-derived nociceptor cultures in vitro, respectively, suggests greater applications for these tools in translational research. Apart from potentially opening a door into the study of pain responsive neuronal populations in higher order animal models, the ability to directly target and manipulate opioidergic cells in human culture samples could greatly improve the specificity of drug screening studies for emerging therapeutics targeted to MOR in pharmaceutical research, as well as the prolonged effects of such compounds on specifically the cell populations of interest they are targeted to with chronic treatment paradigms. Indeed, the broadening of opioid research models is further extended to in vitro culture systems, as suggested in our exploratory use case for human-differentiated iPSCs. Patient- and disease-specific iPSC lines are currently under development as powerful in vitro disease models that provide unique exploration of nociceptor mechanisms of pain, including use in high-throughput screens for novel analgesics and as diagnostics to identify indivi