The life and times of endogenous opioid peptides: Updated understanding of synthesis, spatiotemporal dynamics, and the clinical impact in alcohol use disorder

Drugs targeting opioid GPCRs remain the most effective treatments for acute intense pain ( Bagley and Ingram, 2020 ; Corder et al., 2018 ; Lindsay et al., 2021 ) and alcohol use disorder (AUD), and are in clinical trials to treat major depressive disorder (MDD; Krystal et al., 2020 ; Pizzagalli et al., 2020 ). Yet after decades of “rational” development, available opioid-targeted therapies continue to have substantial liabilities. Mu opioid receptor (MOR) agonist drugs cause life threatening depression of respiration ( Aghabiklooei et al., 2013 ; Gerak et al., 2018 ) and disruption of gut motility ( Holzer, 2009 ). Continuous use can lead to rapid development of tolerance and risks development of opioid use disorder (OUD; Marie, 2019 ). OUD-related costs in the US exceed $1B/year ( Luo et al., 2021 ) and accidental toxic opioid poisonings now contribute to over 70,000 deaths/year. A major scientific challenge in overcoming these liabilities is that in vitro systems, cell lines that overexpress the receptor of interest and a reporter signaling pathway, largely fail to predict clinical outcomes. For instance, full MOR agonists should produce reward and we expect all such compounds will be self-administered in animal models. Yet, a cyclized variant of endomorphin 1 (EM-1), ZH853, which has similar in vitro activities to other full MOR agonists, does not support self administration in rodent studies ( Zadina et al., 2016 ). In vivo models also often lack predictive power. To explain this and other unexpected observations, a range of mechanistic hypotheses and ligand features have been proposed, including biased agonism, partial agonism, ion-dependence, heteromeric complexes, alternative splicing, G protein selectivity, and polypharmacology. However, fundamentally, a complete explanation must recognize how opioid drugs interact with the endogenous opioid system. In turn, this requires a thorough characterization of the regulation and function of endogenous opioid peptides. Understanding how the endogenous opioid peptides act at the opioid receptors may also improve ligand targeting because the endogenous systems have evolved together and are highly conserved throughout vertebrate evolution ( Abbadie et al., 2002 ; Larhammar et al., 2015 ; Li et al., 1996 ; Stevens, 2009 ). Here we summarize basic facts, recent developments, and knowledge gaps about how the at least 26 endogenous opioid peptides are regulated and function in the central nervous system. We present information on the synthesis of the endogenous opioid peptides and what is known regarding how they interact with the different members of the opioid receptor family. We describe examples of different modes of subcellular neuronal distributions of the peptide precursors and putative signaling peptides, and how this relates to potential release sites and receptor target localization. We then describe what these distributions might mean for how we conceptualize the spatiotemporal dynamics of opioid neuromodulation in the brain and thus highlight some remaining fundamental questions about the properties and functions of the endogenous opioid peptide system. Finally, to illustrate the clinical significance of understanding and targeting the endogenous opioid system, we describe how naltrexone’s (NTX) interference with endogenous opioid peptide signaling treats AUD. We review NTX’s clinical efficacy and some of the subsequent research aimed at learning more about the underlying biological phenomena.

PDYN is the source of dynorphin peptides, which are broadly thought to underlie aversive signals through activation of the KOR. This precursor was first described as Proenkephalin B, due to its similarity to PENK ( Chavkin et al., 1982 ; James et al., 1982 ). Peptides generated from PDYN include dynorphin 1–8, dynorphin-A 1–17, dynorphin-B 1–13, dynorphin 2–17, big dynorphin, and α- and β-neo-endorphin ( Fig. 2 ). In addition to activating KORs, dynorphins do also appear to have relatively high affinity for MOR and DOR ( Fig. 1 ; Gomes et al., 2020 ). For instance, the affinities of dynorphin A 1–8, 1–17, and dynorphin B 1–13 are within an order of magnitude of MENK at MOR ( Gomes et al., 2020 ). For β-neo-endorphin, binding affinity and potency at KOR and DOR are not distinguishable ( Gomes et al., 2020 ; Mansour et al., 1995 ). On the other hand, in recent studies the in vivo impacts of selective stimulation of Pdyn neurons have been blocked by selective KOR antagonists (e.g., Abraham et al., 2021 ; Al-Hasani et al., 2015 ). Such selectivity may also be facilitated by neural structural and organizational features described below. Interestingly, the address end of these PDYN peptides is the LENK sequence; Pdyn contains 3 copies of the LENK sequence plus the appropriate cleavage sites to generate the pentapeptide ( Fig. 2 ). It remains to be determined if PENK, PDYN, or an unidentified precursor gives rise to LENK destined for signaling. Evidence supporting the possibility that LENK is a product of PENK alone includes early ex vivo release studies measuring MENK and LENK release from acute striatal brain slice preparations, which contain both Penk and Pdyn neurons. Here the peptides were detected at ratios of between 3:1 and 4:1, consistent with the predicted ratio of MENK to LENK synthesized from PENK alone ( Henderson et al., 1978 ). Mouse knockout studies are also consistent with the majority of LENK arising from PENK ( Gupta et al., 2021 ). On the other hand, studies supporting PDYN also being a source of LENK include observations that enzymes extracted from the striatum, hippocampus, and substantia nigra metabolize longer dynorphin peptides into LENK ( Sandin et al., 1997 ). The approach used in this study, however, did not limit detection to peptides that are likely released as signaling molecules. Further, destruction of the ventral striatal neurons that project to the substantia nigra that express Pdyn and not Penk decreased LENK abundance in the nigra ( Christensson-Nylander et al., 1986 ). Together, while evidence is mostly against PDYN being directly upstream of LENK release, it remains plausible that it may occur in specific brain regions or under certain conditions. Clear functional differences have been observed between dynorphin A and B. Kunselman and colleagues (2021) found that in PC12 cells and cultured striatal neurons, while dynorphin B leads to KOR internalization and then recycling to the plasma membrane, dynorphin A causes internalization with little recovery of receptor to the plasma membrane. They also found that when KOR is activated by dynorphin A, signaling continues after internalization to a much greater extent than when it is activated by dynorphin B. Together these observations indicate a lasting change in receptor signaling and availability that depends upon which dynorphin peptide is released. Another interesting phenomenon related to PDYN products is that sometimes the cleavage between dynorphin A (1–17) and dynorphin B fails, leaving a much longer functional peptide, big dynorphin ( Fig. 2 ; Merg et al., 2006 ). Rather than acting through opioid receptors, it appears that big dynorphin acts through glutamatergic NMDA receptors or acid-sensing ion channels type 1a (ASIC1a): using ex vivo electrophysiology in acutely dissociated trigeminal neurons, big dynorphin inhibited NMDA receptor mediated currents, and these effects were not blocked by the non-selective opioid receptor antagonist naloxone or the selective KOR antagonist norBNI ( Chen et al., 1995a , 1995b ). With in vivo pharmacology experiments, Kuzmin and colleagues (2006) found that intracerebroventricular injection of big dynorphin increased locomotion and anxiolysis; these effects were not blocked by norBNI, but were blocked by the NMDA receptor antagonist MK-801. They also showed that while dynorphin A and B increased the latency to escape in a hot plate test, big dynorphin did not, strengthening the possibility that there are separate signaling impacts of these peptides ( Kuzmin et al., 2006 ; Tan-No et al., 2002 ). In cortical neurons, big dynorphin limits the steady-state desensitization of ASIC1a and acid-activated currents, and this effect is blocked by a peptide that binds directly to ASIC1a ( Sherwood and Askwith, 2009 ). More recently it was found that big dynorphin binding to ASIC1a results in a closed resting conformation ( Borg et al., 2020 ). This effect at ASIC1a appears to be mediated by electrosta

PNOC encodes nociceptin/orphanin-FQ, the endogenous ligand for the opioid-like receptor 1/nociceptin receptor ( Houtani et al., 1996 ; Wang et al., 1996 ). Interestingly, although there are some sequence similarities between PDYN and PNOC ( Houtani et al., 1996 ), the peptides derived from PNOC have minimal affinity for the other opioid receptors ( Fig. 1 ; Zhang et al., 1997 ). The propeptide is processed to form 3 primary signaling peptides ( Fig. 2 ): nociceptin/orphanin-FQ, nocistatin ( Okuda-Ashitaka et al., 1998 ), whose target receptor(s) is unclear ( Avenali et al., 2016 ; Kuspiel et al., 2021 ; Osmakov et al., 2019 ), and NocII/NocIII that produce behavioral effects when administered intracerebroventricularly ( Florin et al., 1997 , 1999 ) but whose effector (s) remains unknown. Although in vivo experiments reported that nociceptin and nocistatin might work in opposition ( Okuda-Ashitaka et al., 1998 ), in whole cell electrophysiology in locus coeruleus (LC) neurons nocistatin application did not activate a conductance and it also did not block responses to nociceptin ( Connor et al., 1999 ).

3.1. Introduction to peptide distribution studies Opioid peptide and receptor distributions in the central nervous system (CNS) have attracted intense research efforts to identify the neural circuits that potentially contribute to both their beneficial and deleterious effects. Of note, opioid receptors are expressed in several non-neuronal organs including heart and immune cells as well. The highest concentrations of endogenous opioid peptides are found in the adrenal gland, where they are synthesized by adrenal medullary chromaffin cells that release the opioid peptides into the adrenal vein for systemic circulation ( Livett et al., 1981 ; Viveros et al., 1979 ). In brain, early on the highest concentration of MENK was found in the globus pallidus ( Kobayashi et al., 1978 ). Intense ENK immunoreactivity was detected in the periaqueductal gray (PAG), a key brain region associated with pain transmission and opioid analgesia ( Moss et al., 1983 ; Moss and Basbaum, 1983 ). In many brain regions ENK and dynorphin expression patterns have been found to be complementary, most easily discerned through mRNA detection (See Section 3.4 ). Much research has also focused on how levels of opioid precursor mRNA or peptide content in different brain regions change following a wide variety of in vivo experiences. The functional implications for peptide release of changes in tissue peptide content or mRNA levels, however, remain mostly unknown, as changes in mRNA levels do not necessarily correlate with changes in peptide expression levels (e.g., Greenbaum et al., 2003 ; Guo et al., 2008 ; Haque et al., 2017 ; Tang et al., 2009 ). Extracellular measurements to detect exocytosed peptide are critical for the field’s understanding, and more temporally precise methods are required to align observed peptide release with specific behavioral events ( Abraham et al., 2021 ; Calhoun et al., 2018 ; Conway et al., 2022 ; DiFeliceantonio et al., 2012 ; Girven et al., 2022 ; Mabrouk et al., 2011 ). More recently immunocytochemical labeling has revealed distinct expression patterns for EM-1 and EM-2. EM-1 is preferentially detected in brain in the parabrachial nucleus, the nucleus of the solitary tract (NTS), septum, diagonal band, bed nucleus of the stria terminalis (BNST), hypothalamus, PAG, nucleus accumbens (NAc), LC, and amygdala (summarized in Fichna et al., 2007 ; Martin-Schild et al., 1999 , 1997 ). In contrast, EM-2 is preferentially expressed in the spinal cord. Importantly, chemically stimulated release of EM-2 was detected in the spinal cord using HPLC and electrochemical detection in samples collected with push-pull perfusion ( Lisi and Sluka, 2006 ).

Genetic sequencing technology has advanced significantly such that high-resolution, high-throughput detection is now possible, including identifying low abundance signals at the single cell level ( Svensson et al., 2017 ). With appropriate controls, there is no practical limit to how many transcripts can be detected in each cell via RNAseq, providing a different, and potentially very exciting, kind of information compared to techniques that require higher abundance of target material for detection. By isolating a brain region, one can then determine the transcripted mRNAs from each isolated cell including neurons, glia, and epithelial cells. These techniques require careful gross dissection and isolation of the desired brain region. Some structures naturally lend themselves to accurate isolation, but others do not. This can make it challenging when attempting to interpret low-expression mRNAs or low-abundance clusters, as they may either represent small populations in the target brain region or contamination from neighboring brain regions. For higher abundance mRNAs or proteins, distributions can be subsequently confirmed with RNAscope or immunofluorescence. Spatial transcriptomics quality is also rapidly improving to provide topographical organization information ( Beauchamp et al., 2022 ). When used appropriately, these techniques can be incredibly powerful, both for confirming earlier observations and for detecting previously unknown expression patterns, thereby enabling the generation of novel hypotheses. For example, unbiased populations of VTA cells from male and female rats were recently profiled with single nucleus RNA sequencing by Phillips and colleagues (2022) . In their analysis of the dataset they focused on the expression of opioid peptide related genes. With respect to the conventional neurotransmitters associated with VTA neurons, dopamine, glutamate, and GABA, the combinatorial expression patterns that were previously reported (reviewed in Morales and Margolis, 2017 ; Barker et al., 2016 ) were also detected in this study. That is, there is anatomical, physiological, immunocytochemical, and conventional in situ hybridization evidence for VTA neurons that have the capacity to synthesize and signal through more than one neurotransmitter including dopamine/glutamate, glutamate/GABA, and dopamine/glutamate/GABA cell populations in the VTA. These different types of VTA neurons participate in specific neural circuits ( Morales and Margolis, 2017 ). One feature of this RNAseq dataset is that it expands the profiling of the genes that contribute to the synthesis, transport, and degradation of these transmitters, all of which provide additional evidence that neurons can in fact synthesize and release each neurotransmitter. For example, they examine several genes involved in the synthesis of tetrahydrobiopterin, which is essential in the first step of dopamine synthesis, generating L-DOPA from tyrosine ( Phillips et al., 2022 ). With respect to genes underlying endogenous opioid peptide production and the opioid receptors, their clustering shows that Penk is relatively highly expressed in the VTA glutamate-only neuron subcluster. Penk expression was also detected in small subpopulations of glutamate/GABA neurons, and a subcluster of GABA only neurons. Message for Pomc or Pdyn was detected just sparsely in VTA neurons, consistent with prior studies. Pnoc was detected among several cluster phenotypes of GABA only neurons and one cluster of glutamate/GABA neurons, consistent with transgenic mouse results ( Parker et al., 2019 ). It is possible that these patterns relate to the projection targets of the VTA neurons, as has been reported for clusters of gene expression detected when dopamine neurons have been studied in isolation ( Poulin et al., 2018 ). Evaluating the expression of the enzymes required for synthesis of the signaling peptides from the propeptides will help determine whether these newly resolved populations of VTA neurons can indeed generate the endogenous opioid peptides. All four opioid receptors were detected in this VTA study, with Oprm1 being most common and detected in many neurophenotypic clusters, followed by Oprd1 , with much lower detection frequencies of Oprk1 and Oprl1 . These findings are in contrast to strong detection of Oprk1 expression in dopamine neurons using RNAscope ( Liu et al., 2019 ) and KOR activity using physiological techniques ( Ford et al., 2007 ; Margolis et al., 2003 , reviewed in Margolis and Karkhanis, 2019 ), as well as the physiological responses to nociceptin reported in both dopamine and non-dopamine VTA neurons ( Driscoll et al., 2020 ; Zheng et al., 2002 ). Together, this kind of analysis of endogenous opioid peptide precursor message as well as that for their receptors can provide new perspectives on expression patterns, but need to be interpreted in the context provided by other techniques.

It is generally thought that propeptides are processed into signaling neuropeptides in rough endoplasmic reticulum then transported through the Golgi apparatus to be packaged into the dense core vesicles from which the peptides are released to the extracellular space. Large dense core vesicles are 80–120 nm in diameter, compared to small clear vesicles that contain fast neurotransmitters and are on average 40 nm in diameter. Therefore putative peptide-containing vesicles can be visually identified by size with sufficient magnification. In ultrastructure studies, dense core vesicles containing endogenous opioid peptides have mostly been observed in axon varicosities and axon terminals, with much less localization to dendrites and rare localization in somata. These vesicles are usually present in low abundance in any of these compartments. Terminals are typically filled with tens to hundreds of small clear vesicles but often co-contain no more than 2–3 dense core vesicles, for example reported in dynorphin containing terminals ( Pickel et al., 1993 ) and ENK containing terminals ( Sesack and Pickel, 1992 ) in the VTA. The apparent scarcity of dense core vesicles is curious given that it is generally thought that neuropeptides signal through volume transmission ( Fig. 3A ), diffusing far from release sites. This would probably require multiple release events in order to achieve pharmacologically relevant extracellular concentrations over a large volume of tissue. Many types of neurons express neuropeptides that are generally thought to have opposing physiological effects, such as lateral hypothalamic neurons that express both orexin (hypocretin) and PDYN ( Chou et al., 2001 ), or striatal MSNs that express substance P and PDYN ( Anderson and Reiner, 1990 ; Reiner, 1986 ). In a simple model where excitatory and inhibitory signals compete to control neural activity, such expression patterns are hard to interpret. One possibility is that the critical downstream intracellular signaling effects of the neuropeptides are as important, if not more important, than the modulation of contemporaneous action potential activity. Another is that the time course of the receptor responses differ. If something like these possibilities is the case, segregated vesicular loading would seem unnecessary ( Fig. 4A ). Sorting of peptides into different vesicles may also be unnecessary if neuromodulators signal through volume transmission ( Fig. 3A ; Agnati et al., 1995 ). Another possibility is that peptide transmission is more targeted, that the wiring diagram for neuropeptide control of neural circuits has higher spatial resolution than volume transmission ( Fig. 3B – E ). There is evidence that both of these arrangements are present in the brain. There are several reports of detection of colocalization of different peptides in the same dense core vesicles ( Fig. 4A and B ). For opioid peptides, one example is in lateral hypothalamus neurons where the generally excitatory neuropeptide orexin is colocalized with dynorphin in dense core vesicles found in axon varicosities, axon terminals, and dendrites ( Muschamp et al., 2014 ). On the other hand, in neurons of the arcuate nucleus of the hypothalamus, neuropeptides generated from different precursor molecules are not co-packaged into the same dense core vesicles ( Fig. 4A and B ). In neurons that produce kisspeptin, neurokinin B, and dynorphin, neither kisspeptin nor neurokinin B was found to colocalize with dynorphin A in dense core vesicles in the somata or axons of these neurons ( Murakawa et al., 2016 ). The peptides synthesized in POMC neurons, including different peptides that are cleaved from POMC itself, may be aggregated at specific release sites ( Fig. 4B ). For instance, in the dorsal raphe nucleus (DRN) most axon terminals that are labeled for β-endorphin are in the dorsomedial DRN. These axon terminals tend to contain either a mix of dense core vesicles with small clear vesicles or mostly dense core vesicles, often synapsing onto dendritic spines of GABAergic profiles ( Wang et al., 2001 ). However, in a separate study it was reported that axon terminals labeled for adrenocorticotropic hormone, also produced from POMC, contain an abundance of small clear vesicles with few dense core vesicles synapsing mostly onto dendritic shafts ( Léger et al., 1997 ). While this is not direct evidence against colocalization, the differing arrangements are consistent with specific axon terminals releasing different peptides ( Fig. 4 ). As proof-of-principle, separate packaging of the different peptides generated from the egg-laying hormone propeptide in Aplysia has been demonstrated to occur through sorting at the trans-Golgi ( Fisher et al., 1988 ). The extent of processing of propeptides prior to packaging into dense core vesicles and transporting of these vesicles to release sites may vary by neuron type. In some neurons there is evidence that propeptides are processed in axon termina

Terminals that contain endogenous opioid peptides have mostly been detected to synapse onto dendrites and other axon terminals in ultrastructural studies ( Fig. 3C and D ). Far less frequently axon terminals containing opioid peptides have been detected synapsing onto somata, such as dynorphin A terminals synapsing onto cell bodies in the substantia nigra pars compacta ( Riesenberg and Nitsch, 1990 ). An important question is: how close are opioid receptors to the release sites of endogenous opioid peptides? Similar to the ultrastructural localization of opioid peptide containing dense core vesicles, opioid receptors are also generally localized extrasynaptically, near the expected release sites for opioid peptide containing dense core vesicles (e.g., Aicher et al., 2003 ; Bockstaele et al., 1996 , 1997 ). Are opioid receptors more likely to be near peptide release sites compared to more distant neuronal profiles? There is some evidence that subpopulations of GPCRs are anchored to specific locations on the plasma membrane ( Thurner et al., 2014 ; Yanagawa et al., 2018 ) including MORs ( Metz et al., 2019 ; Suzuki et al., 2005 ). These single receptor tracking techniques suggest that individual receptors move within plasma membrane areas on the order of <10 μm 2 , consistent with some compartmentalization. Another clue for targeted localization of GPCRs to specific neural compartments is the distribution of G proteins. A recent study reported the distribution of Gαo proteins in Purkinje neurons in mouse cerebellum ( Roldán-Sastre et al., 2021 ). Using immunocytochemical labeling of the G protein and localization with electron microscopy, Roldán-Sastre and colleagues reported that for Purkinje neurons most labeling was within dendritic spines, virtually no labeling was localized to somata, with low labeling in main/proximal dendrites and moderate labeling in spiny branchlet dendrites. Further, frequency of Gαo particle detection peaked at 180 nm from the synaptic specialization, with much lower density reported at distances ≥300 nm. Gαo particles were also located within the postsynaptic density, closer to the peripheral edges than to the center (see also Fernández et al., 2009 ). It is unknown whether these might be associated with neuropeptide receptors like opioid receptors or GPCRs that respond to fast neurotransmitters such as GABA B receptors or metabotropic glutamate receptors. While we do not know from this report the distribution of specific GPCRs relative to the Gαo, it does suggest some overall organization to the distribution of GPCRs within neural compartments. As discussed above, another important component impacting the physiologically plausible distance(s) between endogenous opioid peptide release location and responding receptor site is enzymatic degradation of the peptides. It is clear that aminopeptidase B and endopeptidase-24.11 have strong regulatory control over peptide signaling because in experiments where these enzymes are blocked not only is the manipulation sufficient to change behavior in vivo but also to facilitate the detection of electrophysiological responses to endogenous opioid peptide release ex vivo (e.g., Bourgoin et al., 1986 ; Fournie-Zaluski et al., 1984 ; Waksman et al., 1985 ), described in the next Section. Thus it seems the distribution and organization of several elements of the endogenous opioid system at the subcellular level may be more consistent with models of relatively limited peptide diffusion away from release sites rather than broad diffusion events generating larger volumes of transmission ( Fig. 3A ). There may be other cellular structures that target diffusion to nearby opioid receptors as well. For example, in the LC axon terminals labeled for LENK contain both small clear and large dense core vesicles along the periphery of the synaptic terminal. MORs were detected on dendrites lateral to the postsynaptic density, and astrocytic processes surround these synapses in an arrangement that might limit diffusion of LENK to nearby MOR containing dendrites ( Bockstaele et al., 1996 ). Overall, though, patterns of proximity of opioid receptors to peptide release sites remain to be resolved.

8.1. Clinical development and experience Can this mechanistic neurobiological research inform our understanding of neuropsychiatric disorders or maladaptive behaviors? Because the opioid receptor antagonist NTX decreases alcohol consumption in humans and animals ( Gonzales and Weiss, 1998 ; Herz, 1997 ; Mitchell et al., 2009 ) it is strongly hypothesized that a key contributor to alcohol’s rewarding effects is activation of the endogenous opioid system. As an antagonist, NTX has an effect when endogenous opioid peptides are released and activating opioid receptors. Although NTX was originally approved in 1984 to treat heroin addiction, studies do not support its use for this indication ( Coffa and Snyder, 2019 ; Greenstein et al., 1981 ; Julius and Renault, 1976 (eds); Lee et al., 2018 ; Minozzi et al., 2011 ; NRC, 1978 ; Rawson and Tennant, 1984 ). Oral NTX was approved for treatment of AUD in 1994 ( Center for Substance Abuse Treatment, 2009 ). Early studies investigating the efficacy of NTX were performed in conjunction with psychotherapy ( Balldin et al., 2003 ; O’Brien et al., 1996 ; O’Malley et al., 1992 ; Oslin et al., 2008 ; Volpicelli et al., 1992 ) and found that NTX decreased alcohol craving, the frequency of return to alcohol consumption, and Addiction Severity Index (ASI) scores compared to placebo ( O’Malley et al., 1992 ; Volpicelli et al., 1992 ). ASI evaluates the severity of the subject’s alcohol use, family, legal, social, psychological, and medical problems and was assessed at the beginning and end of the 12 week study ( O’Malley et al., 1992 ). It is important to note that there was also an interaction between type of therapy and frequency of relapse: providing coping skills to patients in the treatment group proved best for relapse prevention ( O’Malley et al., 1992 ). This is consistent with studies in which cognitive behavioral therapy improved AUD treatment outcomes ( Ray et al., 2020 ). A six-month follow-up study showed that the NTX treated subjects were still less likely to drink heavily, suggesting ongoing efficacy ( O’Malley et al., 1996 ). A clue to how endogenous opioids contribute to the motivation to drink alcohol is in an early study where subjects reported that their alcohol induced “high” was diminished when on NTX ( Volpicelli et al., 1992 ). This effect should decrease the motivation to continue drinking because it makes consumption less rewarding. In fact, and consistent with this, studies indicate that rather than promoting abstinence, NTX reduces heavy drinking episodes ( O’Malley et al., 1992 ; Rohsenow, 2004 ). One issue with oral NTX is subject non-compliance ( Volpicelli et al., 1997 ), a common problem for substance use disorder treatments ( Herbeck et al., 2005 ). To mitigate this, a long-acting intramuscular injection that lasts 4 weeks, brand named Vivitrol ® , was developed ( Center for Substance Abuse Treatment, 2009 ; Jarvis et al., 2018 ; O’Brien et al., 1996 ). In addition to compliance, another advantage of the injectable is more consistent blood levels of NTX compared to daily oral dosing ( Johnson, 2007 ; Mannelli et al., 2014 ). Clinical trials for the long-acting injectable showed a 25% reduction in heavy drinking episodes compared to placebo ( Garbutt et al., 2005 ). For patients that were abstinent before treatment began, the time to first drink with extended-release treatment was 41 days versus 12 days for those who received placebo ( O’Malley et al., 2007 ). Importantly, only 14% of subjects discontinued treatment ( Garbutt et al., 2005 ). While duration of abstinence had long been used as the main clinical endpoint for AUD treatment studies, there is a mismatch between this endpoint and the model of NTX target action ( Garbutt et al., 2005 ; Leighty and Ansara, 2019 ; O’Malley et al., 2007 ; Pettinati et al., 2011 ). That is, if the hypothesized mechanism is correct, NTX should affect signaling during drinking that causes endogenous opioid peptide release, not decrease the desire to drink while abstinent. Consistent with this model, oral NTX decreased total drinking days per month, decreased heavy drinking days per month, and increased days until the first heavy drinking event ( O’Malley et al., 2007 ; Pettinati et al., 2011 ). Extended release NTX also decreased the number of heavy drinking days for patients with high severity AUD ( Espiridion, 2019 ; Pettinati et al., 2011 ). Fewer heavy drinking days can improve mental and physical health, mitigating many of the problems associated with AUD. Several side effects of oral and intramuscular NTX have been recorded, including drowsiness (30%), nausea (26%), vomiting (17%), decreased appetite (18%), abdominal pain (16%), insomnia (16%), dizziness (12%), and injection site reactions for intramuscular NTX ( Jonas et al., 2014 ; Kranzler and Soyka, 2018 ). Nausea was less prevalent in patients with lower severity drinking episodes and longer abstinence before treatment initiation, consistent with NT

There are many remaining questions about endogenous opioid peptide function. Very little is known regarding when and where endogenous opioid peptides are released. Additional questions that require study include: how is peptide release driven, how far do the peptides diffuse from a release site, what is the extracellular concentration of peptide produced by release, how long does released peptide impact preand postsynaptic function, which signaling pathways are activated by the different endogenous opioid peptides at each opioid receptor, and do these phenomena vary with brain region? Recent reviews describe advances and ongoing challenges in detecting endogenous opioid peptide fluctuations ( Conway et al., 2022 ; Girven et al., 2022 ). New transgenic rodents continue to be created that enable careful investigations of endogenous opioid peptide circuits. Improved resolution and sensitivity of sequencing based assays enable detection of lower abundance RNA message identifying previously unknown expression patterns for opioid peptides and receptors. For recent and future single cell transcriptome studies, when the datasets are made public researchers can probe them for opioid system expression patterns. Recent advances in sequential re-staining and co-registration of cleared tissue that enable labeling of a dozen or more proteins in the same sample promise the opportunity to integrate detection of elements of the endogenous opioid system into a broader context ( Hickey et al., 2022 ; Kim et al., 2022 ; Murray et al., 2015 ). Advances in electron microscopy and reconstruction techniques provide improved understanding of the 3 dimensional structures such as where the peptides are released and the available extracellular space for diffusion to their target receptors. New tools for imaging fluctuations in neuropeptide levels in vivo are being developed (e.g. Ding et al., 2019 ). At the systems level, many questions remain regarding the contributions of the endogenous opioid system to sex differences including in pain, AUD, and SUD, as well as in treatment efficacies for these and other disorders. Still, much progress has been made in our understanding of endogenous opioid peptide generation and how the different peptides bind and activate each of the opioid receptors. The opioid field has a wealth of tool compounds as well that have been used to investigate the behavioral impact of activating or blocking different elements of this system, and the subsequent observations have generated many hypotheses not only for how to improve on NTX efficacy for treating AUD, but also how to interact with the endogenous opioid system in novel ways to treat other indications. In addition to applying recent discoveries to design better analgesics that activate MOR and MDD treatments that block KOR, one exciting possibility is to boost the activity of endogenous opioid peptides with a positive allosteric modulator of MOR. Blockade of endogenous opioid peptide catabolism in vivo does not cause respiratory depression, raising the possibility that a positive allosteric modulator at MOR will generate analgesia without causing respiratory depression ( Boudinot et al., 2001 ; Fournie-Zaluski et al., 1984 ; Kayser et al., 1989 ). Preclinical studies indicate potential for this approach ( Kandasamy et al., 2021 ). Provocatively, positive allosteric activity at MORs might be one of the mechanisms by which ketamine produces analgesia ( Gupta et al., 2011 ; Petrocchi et al., 2019 ). The efficacy of a positive allosteric modulator requires endogenous opioid peptide activity at the orthosteric binding site of the receptor. Thus, continuing to investigate the conditions when different endogenous opioid peptides are released with new and exciting tools, along with new advances in understanding structure-activity relationships in the opioid systems, has the potential to not only generate exciting biological observations but also new therapeutic strategies for various clinical indications.